diff options
Diffstat (limited to 'servers/rendering/renderer_rd')
66 files changed, 5738 insertions, 2540 deletions
diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.h b/servers/rendering/renderer_rd/cluster_builder_rd.h index ebb81abdad..feafd4c2db 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.h +++ b/servers/rendering/renderer_rd/cluster_builder_rd.h @@ -235,7 +235,7 @@ public: Transform3D xform = view_xform * p_transform; float radius = xform.basis.get_uniform_scale(); - if (radius > 0.98 || radius < 1.02) { + if (radius < 0.98 || radius > 1.02) { xform.basis.orthonormalize(); } @@ -289,7 +289,7 @@ public: e.touches_near = min_d < z_near; } else { //contains camera inside light - Plane base_plane(xform.origin, -xform.basis.get_axis(Vector3::AXIS_Z)); + Plane base_plane(-xform.basis.get_axis(Vector3::AXIS_Z), xform.origin); float dist = base_plane.distance_to(Vector3()); if (dist >= 0 && dist < radius) { //inside, check angle diff --git a/servers/rendering/renderer_rd/effects_rd.cpp b/servers/rendering/renderer_rd/effects_rd.cpp index 5cf8895c8e..c69408a30b 100644 --- a/servers/rendering/renderer_rd/effects_rd.cpp +++ b/servers/rendering/renderer_rd/effects_rd.cpp @@ -37,6 +37,10 @@ #include "servers/rendering/renderer_rd/renderer_compositor_rd.h" #include "thirdparty/misc/cubemap_coeffs.h" +bool EffectsRD::get_prefer_raster_effects() { + return prefer_raster_effects; +} + static _FORCE_INLINE_ void store_camera(const CameraMatrix &p_mtx, float *p_array) { for (int i = 0; i < 4; i++) { for (int j = 0; j < 4; j++) { @@ -66,6 +70,28 @@ RID EffectsRD::_get_uniform_set_from_image(RID p_image) { return uniform_set; } +RID EffectsRD::_get_uniform_set_for_input(RID p_texture) { + if (input_to_uniform_set_cache.has(p_texture)) { + RID uniform_set = input_to_uniform_set_cache[p_texture]; + if (RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + return uniform_set; + } + } + + Vector<RD::Uniform> uniforms; + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_INPUT_ATTACHMENT; + u.binding = 0; + u.ids.push_back(p_texture); + uniforms.push_back(u); + // This is specific to our subpass shader + RID uniform_set = RD::get_singleton()->uniform_set_create(uniforms, tonemap.shader.version_get_shader(tonemap.shader_version, TONEMAP_MODE_SUBPASS), 0); + + input_to_uniform_set_cache[p_texture] = uniform_set; + + return uniform_set; +} + RID EffectsRD::_get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps) { if (texture_to_uniform_set_cache.has(p_texture)) { RID uniform_set = texture_to_uniform_set_cache[p_texture]; @@ -81,7 +107,7 @@ RID EffectsRD::_get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps) u.ids.push_back(p_use_mipmaps ? default_mipmap_sampler : default_sampler); u.ids.push_back(p_texture); uniforms.push_back(u); - //anything with the same configuration (one texture in binding 0 for set 0), is good + // anything with the same configuration (one texture in binding 0 for set 0), is good RID uniform_set = RD::get_singleton()->uniform_set_create(uniforms, tonemap.shader.version_get_shader(tonemap.shader_version, 0), 0); texture_to_uniform_set_cache[p_texture] = uniform_set; @@ -383,6 +409,8 @@ void EffectsRD::set_color(RID p_dest_texture, const Color &p_color, const Rect2i } void EffectsRD::gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back_texture, const Rect2i &p_region, bool p_8bit_dst) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use the compute version of the gaussian blur with the mobile renderer."); + memset(©.push_constant, 0, sizeof(CopyPushConstant)); uint32_t base_flags = 0; @@ -416,6 +444,8 @@ void EffectsRD::gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back } void EffectsRD::gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const Size2i &p_size, float p_strength, bool p_high_quality, bool p_first_pass, float p_luminance_cap, float p_exposure, float p_bloom, float p_hdr_bleed_treshold, float p_hdr_bleed_scale, RID p_auto_exposure, float p_auto_exposure_grey) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use the compute version of the gaussian glow with the mobile renderer."); + memset(©.push_constant, 0, sizeof(CopyPushConstant)); CopyMode copy_mode = p_first_pass && p_auto_exposure.is_valid() ? COPY_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE : COPY_MODE_GAUSSIAN_GLOW; @@ -449,6 +479,57 @@ void EffectsRD::gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const RD::get_singleton()->compute_list_end(); } +void EffectsRD::gaussian_glow_raster(RID p_source_rd_texture, RID p_framebuffer_half, RID p_rd_texture_half, RID p_dest_framebuffer, const Vector2 &p_pixel_size, float p_strength, bool p_high_quality, bool p_first_pass, float p_luminance_cap, float p_exposure, float p_bloom, float p_hdr_bleed_treshold, float p_hdr_bleed_scale, RID p_auto_exposure, float p_auto_exposure_grey) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use the raster version of the gaussian glow with the clustered renderer."); + + memset(&blur_raster.push_constant, 0, sizeof(BlurRasterPushConstant)); + + BlurRasterMode blur_mode = p_first_pass && p_auto_exposure.is_valid() ? BLUR_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE : BLUR_MODE_GAUSSIAN_GLOW; + uint32_t base_flags = 0; + + blur_raster.push_constant.pixel_size[0] = p_pixel_size.x; + blur_raster.push_constant.pixel_size[1] = p_pixel_size.y; + + blur_raster.push_constant.glow_strength = p_strength; + blur_raster.push_constant.glow_bloom = p_bloom; + blur_raster.push_constant.glow_hdr_threshold = p_hdr_bleed_treshold; + blur_raster.push_constant.glow_hdr_scale = p_hdr_bleed_scale; + blur_raster.push_constant.glow_exposure = p_exposure; + blur_raster.push_constant.glow_white = 0; //actually unused + blur_raster.push_constant.glow_luminance_cap = p_luminance_cap; + + blur_raster.push_constant.glow_auto_exposure_grey = p_auto_exposure_grey; //unused also + + //HORIZONTAL + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_framebuffer_half, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, blur_raster.pipelines[blur_mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer_half))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_rd_texture), 0); + if (p_auto_exposure.is_valid() && p_first_pass) { + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_auto_exposure), 1); + } + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + blur_raster.push_constant.flags = base_flags | BLUR_FLAG_HORIZONTAL | (p_first_pass ? BLUR_FLAG_GLOW_FIRST_PASS : 0); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &blur_raster.push_constant, sizeof(BlurRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + + blur_mode = BLUR_MODE_GAUSSIAN_GLOW; + + //VERTICAL + draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, blur_raster.pipelines[blur_mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_rd_texture_half), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + blur_raster.push_constant.flags = base_flags; + RD::get_singleton()->draw_list_set_push_constant(draw_list, &blur_raster.push_constant, sizeof(BlurRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, RenderingServer::EnvironmentSSRRoughnessQuality p_roughness_quality, RID p_blur_radius, RID p_blur_radius2, RID p_metallic, const Color &p_metallic_mask, RID p_depth, RID p_scale_depth, RID p_scale_normal, RID p_output, RID p_output_blur, const Size2i &p_screen_size, int p_max_steps, float p_fade_in, float p_fade_out, float p_tolerance, const CameraMatrix &p_camera) { RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); @@ -658,7 +739,27 @@ void EffectsRD::make_mipmap(RID p_source_rd_texture, RID p_dest_texture, const S RD::get_singleton()->compute_list_end(); } -void EffectsRD::copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dst_framebuffer, const Rect2 &p_rect, float p_z_near, float p_z_far, bool p_dp_flip) { +void EffectsRD::make_mipmap_raster(RID p_source_rd_texture, RID p_dest_framebuffer, const Size2i &p_size) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use the raster version of mipmap with the clustered renderer."); + + memset(&blur_raster.push_constant, 0, sizeof(BlurRasterPushConstant)); + + BlurRasterMode mode = BLUR_MIPMAP; + + blur_raster.push_constant.pixel_size[0] = 1.0 / float(p_size.x); + blur_raster.push_constant.pixel_size[1] = 1.0 / float(p_size.y); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, blur_raster.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_rd_texture), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &blur_raster.push_constant, sizeof(BlurRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + +void EffectsRD::copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dst_framebuffer, const Rect2 &p_rect, const Vector2 &p_dst_size, float p_z_near, float p_z_far, bool p_dp_flip) { CopyToDPPushConstant push_constant; push_constant.screen_rect[0] = p_rect.position.x; push_constant.screen_rect[1] = p_rect.position.y; @@ -666,7 +767,9 @@ void EffectsRD::copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dst_framebuffe push_constant.screen_rect[3] = p_rect.size.height; push_constant.z_far = p_z_far; push_constant.z_near = p_z_near; - push_constant.z_flip = p_dp_flip; + push_constant.texel_size[0] = 1.0f / p_dst_size.x; + push_constant.texel_size[1] = 1.0f / p_dst_size.y; + push_constant.texel_size[0] *= p_dp_flip ? -1.0f : 1.0f; // Encode dp flip as x size sign RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dst_framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, cube_to_dp.pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer))); @@ -709,6 +812,7 @@ void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const Tone tonemap.push_constant.exposure = p_settings.exposure; tonemap.push_constant.white = p_settings.white; tonemap.push_constant.auto_exposure_grey = p_settings.auto_exposure_grey; + tonemap.push_constant.luminance_multiplier = p_settings.luminance_multiplier; tonemap.push_constant.use_color_correction = p_settings.use_color_correction; @@ -719,11 +823,11 @@ void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const Tone if (p_settings.view_count > 1) { // Use MULTIVIEW versions - mode += 4; + mode += 6; } RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dst_framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); - RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer))); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer), false, RD::get_singleton()->draw_list_get_current_pass())); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_color), 0); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_settings.exposure_texture), 1); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_settings.glow_texture, true), 2); @@ -735,7 +839,49 @@ void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const Tone RD::get_singleton()->draw_list_end(); } +void EffectsRD::tonemapper(RD::DrawListID p_subpass_draw_list, RID p_source_color, RD::FramebufferFormatID p_dst_format_id, const TonemapSettings &p_settings) { + memset(&tonemap.push_constant, 0, sizeof(TonemapPushConstant)); + + tonemap.push_constant.use_bcs = p_settings.use_bcs; + tonemap.push_constant.bcs[0] = p_settings.brightness; + tonemap.push_constant.bcs[1] = p_settings.contrast; + tonemap.push_constant.bcs[2] = p_settings.saturation; + + ERR_FAIL_COND_MSG(p_settings.use_glow, "Glow is not supported when using subpasses."); + tonemap.push_constant.use_glow = p_settings.use_glow; + + int mode = p_settings.use_1d_color_correction ? TONEMAP_MODE_SUBPASS_1D_LUT : TONEMAP_MODE_SUBPASS; + if (p_settings.view_count > 1) { + // Use MULTIVIEW versions + mode += 6; + } + + tonemap.push_constant.tonemapper = p_settings.tonemap_mode; + tonemap.push_constant.use_auto_exposure = p_settings.use_auto_exposure; + tonemap.push_constant.exposure = p_settings.exposure; + tonemap.push_constant.white = p_settings.white; + tonemap.push_constant.auto_exposure_grey = p_settings.auto_exposure_grey; + + tonemap.push_constant.use_color_correction = p_settings.use_color_correction; + + tonemap.push_constant.use_debanding = p_settings.use_debanding; + tonemap.push_constant.luminance_multiplier = p_settings.luminance_multiplier; + + RD::get_singleton()->draw_list_bind_render_pipeline(p_subpass_draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, p_dst_format_id, false, RD::get_singleton()->draw_list_get_current_pass())); + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_for_input(p_source_color), 0); + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.exposure_texture), 1); // should be set to a default texture, it's ignored + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.glow_texture, true), 2); // should be set to a default texture, it's ignored + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.color_correction_texture), 3); + + RD::get_singleton()->draw_list_bind_index_array(p_subpass_draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(p_subpass_draw_list, &tonemap.push_constant, sizeof(TonemapPushConstant)); + RD::get_singleton()->draw_list_draw(p_subpass_draw_list, true); +} + void EffectsRD::luminance_reduction(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute version of luminance reduction with the mobile renderer."); + luminance_reduce.push_constant.source_size[0] = p_source_size.x; luminance_reduce.push_constant.source_size[1] = p_source_size.y; luminance_reduce.push_constant.max_luminance = p_max_luminance; @@ -774,7 +920,41 @@ void EffectsRD::luminance_reduction(RID p_source_texture, const Size2i p_source_ RD::get_singleton()->compute_list_end(); } -void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i &p_base_texture_size, RID p_secondary_texture, RID p_halfsize_texture1, RID p_halfsize_texture2, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_bokeh_size, RenderingServer::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, bool p_use_jitter, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal) { +void EffectsRD::luminance_reduction_raster(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, Vector<RID> p_fb, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster version of luminance reduction with the clustered renderer."); + ERR_FAIL_COND_MSG(p_reduce.size() != p_fb.size(), "Incorrect frame buffer account for luminance reduction."); + + luminance_reduce_raster.push_constant.max_luminance = p_max_luminance; + luminance_reduce_raster.push_constant.min_luminance = p_min_luminance; + luminance_reduce_raster.push_constant.exposure_adjust = p_adjust; + + for (int i = 0; i < p_reduce.size(); i++) { + luminance_reduce_raster.push_constant.source_size[0] = i == 0 ? p_source_size.x : luminance_reduce_raster.push_constant.dest_size[0]; + luminance_reduce_raster.push_constant.source_size[1] = i == 0 ? p_source_size.y : luminance_reduce_raster.push_constant.dest_size[1]; + luminance_reduce_raster.push_constant.dest_size[0] = MAX(luminance_reduce_raster.push_constant.source_size[0] / 8, 1); + luminance_reduce_raster.push_constant.dest_size[1] = MAX(luminance_reduce_raster.push_constant.source_size[1] / 8, 1); + + bool final = !p_set && (luminance_reduce_raster.push_constant.dest_size[0] == 1) && (luminance_reduce_raster.push_constant.dest_size[1] == 1); + LuminanceReduceRasterMode mode = final ? LUMINANCE_REDUCE_FRAGMENT_FINAL : (i == 0 ? LUMINANCE_REDUCE_FRAGMENT_FIRST : LUMINANCE_REDUCE_FRAGMENT); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_fb[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, luminance_reduce_raster.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_fb[i]))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(i == 0 ? p_source_texture : p_reduce[i - 1]), 0); + if (final) { + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_prev_luminance), 1); + } + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &luminance_reduce_raster.push_constant, sizeof(LuminanceReduceRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + } +} + +void EffectsRD::bokeh_dof(const BokehBuffers &p_buffers, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_bokeh_size, RenderingServer::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, bool p_use_jitter, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute version of BOKEH DOF with the mobile renderer."); + bokeh.push_constant.blur_far_active = p_dof_far; bokeh.push_constant.blur_far_begin = p_dof_far_begin; bokeh.push_constant.blur_far_end = p_dof_far_begin + p_dof_far_size; @@ -801,22 +981,22 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i // The alpha channel of the source color texture is filled with the expected circle size // If used for DOF far, the size is positive, if used for near, its negative. - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.pipelines[BOKEH_GEN_BLUR_SIZE]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.compute_pipelines[BOKEH_GEN_BLUR_SIZE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_depth_texture), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.base_texture), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.depth_texture), 1); - bokeh.push_constant.size[0] = p_base_texture_size.x; - bokeh.push_constant.size[1] = p_base_texture_size.y; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_buffers.base_texture_size.x, p_buffers.base_texture_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); if (p_bokeh_shape == RS::DOF_BOKEH_BOX || p_bokeh_shape == RS::DOF_BOKEH_HEXAGON) { //second pass - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.pipelines[p_bokeh_shape == RS::DOF_BOKEH_BOX ? BOKEH_GEN_BOKEH_BOX : BOKEH_GEN_BOKEH_HEXAGONAL]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.compute_pipelines[p_bokeh_shape == RS::DOF_BOKEH_BOX ? BOKEH_GEN_BOKEH_BOX : BOKEH_GEN_BOKEH_HEXAGONAL]); static const int quality_samples[4] = { 6, 12, 12, 24 }; @@ -825,18 +1005,18 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i if (p_quality == RS::DOF_BLUR_QUALITY_VERY_LOW || p_quality == RS::DOF_BLUR_QUALITY_LOW) { //box and hexagon are more or less the same, and they can work in either half (very low and low quality) or full (medium and high quality_ sizes) - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_halfsize_texture1), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_base_texture), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.half_texture[0]), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.base_texture), 1); - bokeh.push_constant.size[0] = p_base_texture_size.x >> 1; - bokeh.push_constant.size[1] = p_base_texture_size.y >> 1; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x >> 1; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y >> 1; bokeh.push_constant.half_size = true; bokeh.push_constant.blur_size *= 0.5; } else { //medium and high quality use full size - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_secondary_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_base_texture), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.secondary_texture), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.base_texture), 1); } RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); @@ -848,11 +1028,11 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i bokeh.push_constant.second_pass = true; if (p_quality == RS::DOF_BLUR_QUALITY_VERY_LOW || p_quality == RS::DOF_BLUR_QUALITY_LOW) { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_halfsize_texture2), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_halfsize_texture1), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.half_texture[1]), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.half_texture[0]), 1); } else { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_secondary_texture), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.base_texture), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.secondary_texture), 1); } RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); @@ -863,25 +1043,25 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i if (p_quality == RS::DOF_BLUR_QUALITY_VERY_LOW || p_quality == RS::DOF_BLUR_QUALITY_LOW) { //forth pass, upscale for low quality - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.pipelines[BOKEH_COMPOSITE]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.compute_pipelines[BOKEH_COMPOSITE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_halfsize_texture2), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.base_texture), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.half_texture[1]), 1); - bokeh.push_constant.size[0] = p_base_texture_size.x; - bokeh.push_constant.size[1] = p_base_texture_size.y; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y; bokeh.push_constant.half_size = false; bokeh.push_constant.second_pass = false; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_buffers.base_texture_size.x, p_buffers.base_texture_size.y, 1); } } else { //circle //second pass - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.pipelines[BOKEH_GEN_BOKEH_CIRCULAR]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.compute_pipelines[BOKEH_GEN_BOKEH_CIRCULAR]); static const float quality_scale[4] = { 8.0, 4.0, 1.0, 0.5 }; @@ -890,11 +1070,11 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i //circle always runs in half size, otherwise too expensive - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_halfsize_texture1), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_base_texture), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.half_texture[0]), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.base_texture), 1); - bokeh.push_constant.size[0] = p_base_texture_size.x >> 1; - bokeh.push_constant.size[1] = p_base_texture_size.y >> 1; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x >> 1; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y >> 1; bokeh.push_constant.half_size = true; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); @@ -906,24 +1086,198 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i // upscale - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.pipelines[BOKEH_COMPOSITE]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, bokeh.compute_pipelines[BOKEH_COMPOSITE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_halfsize_texture1), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_buffers.base_texture), 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_buffers.half_texture[0]), 1); - bokeh.push_constant.size[0] = p_base_texture_size.x; - bokeh.push_constant.size[1] = p_base_texture_size.y; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y; bokeh.push_constant.half_size = false; bokeh.push_constant.second_pass = false; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_buffers.base_texture_size.x, p_buffers.base_texture_size.y, 1); } RD::get_singleton()->compute_list_end(); } +void EffectsRD::bokeh_dof_raster(const BokehBuffers &p_buffers, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_dof_blur_amount, RenderingServer::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use blur DOF with the clustered renderer."); + + memset(&bokeh.push_constant, 0, sizeof(BokehPushConstant)); + + bokeh.push_constant.orthogonal = p_cam_orthogonal; + bokeh.push_constant.size[0] = p_buffers.base_texture_size.width; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.height; + bokeh.push_constant.z_far = p_cam_zfar; + bokeh.push_constant.z_near = p_cam_znear; + + bokeh.push_constant.second_pass = false; + bokeh.push_constant.half_size = false; + bokeh.push_constant.blur_size = p_dof_blur_amount; + + if (p_dof_far || p_dof_near) { + if (p_dof_far) { + bokeh.push_constant.blur_far_active = true; + bokeh.push_constant.blur_far_begin = p_dof_far_begin; + bokeh.push_constant.blur_far_end = p_dof_far_begin + p_dof_far_size; + } + + if (p_dof_near) { + bokeh.push_constant.blur_near_active = true; + bokeh.push_constant.blur_near_begin = p_dof_near_begin; + bokeh.push_constant.blur_near_end = p_dof_near_begin - p_dof_near_size; + } + + { + // generate our depth data + RID framebuffer = p_buffers.base_weight_fb; + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[BOKEH_GEN_BLUR_SIZE].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.depth_texture), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + } + + if (p_bokeh_shape == RS::DOF_BOKEH_BOX || p_bokeh_shape == RS::DOF_BOKEH_HEXAGON) { + // double pass approach + BokehMode mode = p_bokeh_shape == RS::DOF_BOKEH_BOX ? BOKEH_GEN_BOKEH_BOX : BOKEH_GEN_BOKEH_HEXAGONAL; + + if (p_quality == RS::DOF_BLUR_QUALITY_VERY_LOW || p_quality == RS::DOF_BLUR_QUALITY_LOW) { + //box and hexagon are more or less the same, and they can work in either half (very low and low quality) or full (medium and high quality_ sizes) + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x >> 1; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y >> 1; + bokeh.push_constant.half_size = true; + bokeh.push_constant.blur_size *= 0.5; + } + + static const int quality_samples[4] = { 6, 12, 12, 24 }; + bokeh.push_constant.blur_scale = 0.5; + bokeh.push_constant.steps = quality_samples[p_quality]; + + RID framebuffer = bokeh.push_constant.half_size ? p_buffers.half_fb[0] : p_buffers.secondary_fb; + + // Pass 1 + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.base_texture), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[0]), 1); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + + // Pass 2 + if (!bokeh.push_constant.half_size) { + // do not output weight, we're writing back into our base buffer + mode = p_bokeh_shape == RS::DOF_BOKEH_BOX ? BOKEH_GEN_BOKEH_BOX_NOWEIGHT : BOKEH_GEN_BOKEH_HEXAGONAL_NOWEIGHT; + } + bokeh.push_constant.second_pass = true; + + framebuffer = bokeh.push_constant.half_size ? p_buffers.half_fb[1] : p_buffers.base_fb; + RID texture = bokeh.push_constant.half_size ? p_buffers.half_texture[0] : p_buffers.secondary_texture; + RID weight = bokeh.push_constant.half_size ? p_buffers.weight_texture[2] : p_buffers.weight_texture[1]; + + draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(texture), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(weight), 1); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + + if (bokeh.push_constant.half_size) { + // Compose pass + mode = BOKEH_COMPOSITE; + framebuffer = p_buffers.base_fb; + + draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.half_texture[1]), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[3]), 1); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[0]), 2); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + } + + } else { + // circular is a single pass approach + BokehMode mode = BOKEH_GEN_BOKEH_CIRCULAR; + + { + // circle always runs in half size, otherwise too expensive (though the code below does support making this optional) + bokeh.push_constant.size[0] = p_buffers.base_texture_size.x >> 1; + bokeh.push_constant.size[1] = p_buffers.base_texture_size.y >> 1; + bokeh.push_constant.half_size = true; + // bokeh.push_constant.blur_size *= 0.5; + } + + static const float quality_scale[4] = { 8.0, 4.0, 1.0, 0.5 }; + bokeh.push_constant.blur_scale = quality_scale[p_quality]; + bokeh.push_constant.steps = 0.0; + + RID framebuffer = bokeh.push_constant.half_size ? p_buffers.half_fb[0] : p_buffers.secondary_fb; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.base_texture), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[0]), 1); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + + if (bokeh.push_constant.half_size) { + // Compose + mode = BOKEH_COMPOSITE; + framebuffer = p_buffers.base_fb; + + draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, bokeh.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.half_texture[0]), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[2]), 1); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.weight_texture[0]), 2); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &bokeh.push_constant, sizeof(BokehPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + } else { + // Just copy it back (we use our blur raster shader here).. + draw_list = RD::get_singleton()->draw_list_begin(p_buffers.base_fb, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, blur_raster.pipelines[BLUR_MODE_COPY].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_buffers.base_fb))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_buffers.secondary_texture), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + memset(&blur_raster.push_constant, 0, sizeof(BlurRasterPushConstant)); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &blur_raster.push_constant, sizeof(BlurRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); + } + } + } +} + void EffectsRD::gather_ssao(RD::ComputeListID p_compute_list, const Vector<RID> p_ao_slices, const SSAOSettings &p_settings, bool p_adaptive_base_pass, RID p_gather_uniform_set, RID p_importance_map_uniform_set) { RD::get_singleton()->compute_list_bind_uniform_set(p_compute_list, p_gather_uniform_set, 0); if ((p_settings.quality == RS::ENV_SSAO_QUALITY_ULTRA) && !p_adaptive_base_pass) { @@ -1046,6 +1400,7 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep ssao.gather_push_constant.half_screen_pixel_size_x025[0] = ssao.gather_push_constant.half_screen_pixel_size[0] * 0.25; ssao.gather_push_constant.half_screen_pixel_size_x025[1] = ssao.gather_push_constant.half_screen_pixel_size[1] * 0.25; + ssao.gather_push_constant.radius = p_settings.radius; float radius_near_limit = (p_settings.radius * 1.2f); if (p_settings.quality <= RS::ENV_SSAO_QUALITY_LOW) { radius_near_limit *= 1.50f; @@ -1053,12 +1408,8 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep if (p_settings.quality == RS::ENV_SSAO_QUALITY_VERY_LOW) { ssao.gather_push_constant.radius *= 0.8f; } - if (p_settings.half_size) { - ssao.gather_push_constant.radius *= 0.5f; - } } radius_near_limit /= tan_half_fov_y; - ssao.gather_push_constant.radius = p_settings.radius; ssao.gather_push_constant.intensity = p_settings.intensity; ssao.gather_push_constant.shadow_power = p_settings.power; ssao.gather_push_constant.shadow_clamp = 0.98; @@ -1188,8 +1539,9 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep if (p_settings.quality > RS::ENV_SSAO_QUALITY_VERY_LOW) { if (pass < blur_passes - 2) { blur_pipeline = SSAO_BLUR_PASS_WIDE; + } else { + blur_pipeline = SSAO_BLUR_PASS_SMART; } - blur_pipeline = SSAO_BLUR_PASS_SMART; } for (int i = 0; i < 4; i++) { @@ -1283,7 +1635,9 @@ void EffectsRD::roughness_limit(RID p_source_normal, RID p_roughness, const Size RD::get_singleton()->compute_list_end(); } -void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size) { +void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_texture, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap roughness with the mobile renderer."); + memset(&roughness.push_constant, 0, sizeof(CubemapRoughnessPushConstant)); roughness.push_constant.face_id = p_face_id > 9 ? 0 : p_face_id; @@ -1293,10 +1647,10 @@ void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffe roughness.push_constant.face_size = p_size; RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, roughness.pipeline); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, roughness.compute_pipeline); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_framebuffer), 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 1); RD::get_singleton()->compute_list_set_push_constant(compute_list, &roughness.push_constant, sizeof(CubemapRoughnessPushConstant)); @@ -1308,11 +1662,37 @@ void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffe RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_roughness_raster(RID p_source_rd_texture, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap roughness with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap roughness must process one side at a time."); + + memset(&roughness.push_constant, 0, sizeof(CubemapRoughnessPushConstant)); + + roughness.push_constant.face_id = p_face_id; + roughness.push_constant.roughness = p_roughness; + roughness.push_constant.sample_count = p_sample_count; + roughness.push_constant.use_direct_write = p_roughness == 0.0; + roughness.push_constant.face_size = p_size; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, roughness.raster_pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_rd_texture), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &roughness.push_constant, sizeof(CubemapRoughnessPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, const Size2i &p_size) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap downsample with the mobile renderer."); + cubemap_downsampler.push_constant.face_size = p_size.x; + cubemap_downsampler.push_constant.face_id = 0; // we render all 6 sides to each layer in one call RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, cubemap_downsampler.pipeline); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, cubemap_downsampler.compute_pipeline); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_cubemap), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_cubemap), 1); @@ -1326,7 +1706,27 @@ void EffectsRD::cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, con RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_downsample_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, const Size2i &p_size) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap downsample with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap downsample must process one side at a time."); + + cubemap_downsampler.push_constant.face_size = p_size.x; + cubemap_downsampler.push_constant.face_id = p_face_id; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, cubemap_downsampler.raster_pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_cubemap), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &cubemap_downsampler.push_constant, sizeof(CubemapDownsamplerPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, bool p_use_array) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap filter with the mobile renderer."); + Vector<RD::Uniform> uniforms; for (int i = 0; i < p_dest_cubemap.size(); i++) { RD::Uniform u; @@ -1338,12 +1738,12 @@ void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, if (RD::get_singleton()->uniform_set_is_valid(filter.image_uniform_set)) { RD::get_singleton()->free(filter.image_uniform_set); } - filter.image_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.shader.version_get_shader(filter.shader_version, 0), 2); + filter.image_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.compute_shader.version_get_shader(filter.shader_version, 0), 2); int pipeline = p_use_array ? FILTER_MODE_HIGH_QUALITY_ARRAY : FILTER_MODE_HIGH_QUALITY; pipeline = filter.use_high_quality ? pipeline : pipeline + 1; RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, filter.pipelines[pipeline]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, filter.compute_pipelines[pipeline]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_cubemap, true), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, filter.uniform_set, 1); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, filter.image_uniform_set, 2); @@ -1355,6 +1755,29 @@ void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_filter_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_mip_level) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap filter with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap filter must process one side at a time."); + + // TODO implement! + CubemapFilterRasterPushConstant push_constant; + push_constant.mip_level = p_mip_level; + push_constant.face_id = p_face_id; + + CubemapFilterMode mode = filter.use_high_quality ? FILTER_MODE_HIGH_QUALITY : FILTER_MODE_LOW_QUALITY; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, filter.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_cubemap), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, filter.uniform_set, 1); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(CubemapFilterRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_voxel_gi, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_voxel_gi, Vector2i p_screen_size, int p_samples, uint32_t p_barrier) { ResolvePushConstant push_constant; push_constant.screen_size[0] = p_screen_size.x; @@ -1464,8 +1887,35 @@ void EffectsRD::sort_buffer(RID p_uniform_set, int p_size) { RD::get_singleton()->compute_list_end(); } -EffectsRD::EffectsRD() { - { // Initialize copy +EffectsRD::EffectsRD(bool p_prefer_raster_effects) { + prefer_raster_effects = p_prefer_raster_effects; + + if (prefer_raster_effects) { + // init blur shader (on compute use copy shader) + + Vector<String> blur_modes; + blur_modes.push_back("\n#define MODE_MIPMAP\n"); // BLUR_MIPMAP + blur_modes.push_back("\n#define MODE_GAUSSIAN_BLUR\n"); // BLUR_MODE_GAUSSIAN_BLUR + blur_modes.push_back("\n#define MODE_GAUSSIAN_GLOW\n"); // BLUR_MODE_GAUSSIAN_GLOW + blur_modes.push_back("\n#define MODE_GAUSSIAN_GLOW\n#define GLOW_USE_AUTO_EXPOSURE\n"); // BLUR_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE + blur_modes.push_back("\n#define MODE_COPY\n"); // BLUR_MODE_COPY + + blur_raster.shader.initialize(blur_modes); + memset(&blur_raster.push_constant, 0, sizeof(BlurRasterPushConstant)); + blur_raster.shader_version = blur_raster.shader.version_create(); + + for (int i = 0; i < BLUR_MODE_MAX; i++) { + blur_raster.pipelines[i].setup(blur_raster.shader.version_get_shader(blur_raster.shader_version, i), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } + + } else { + // not used in clustered + for (int i = 0; i < BLUR_MODE_MAX; i++) { + blur_raster.pipelines[i].clear(); + } + } + + if (!prefer_raster_effects) { // Initialize copy Vector<String> copy_modes; copy_modes.push_back("\n#define MODE_GAUSSIAN_BLUR\n"); copy_modes.push_back("\n#define MODE_GAUSSIAN_BLUR\n#define DST_IMAGE_8BIT\n"); @@ -1483,10 +1933,21 @@ EffectsRD::EffectsRD() { copy.shader.initialize(copy_modes); memset(©.push_constant, 0, sizeof(CopyPushConstant)); + + if (prefer_raster_effects) { + // disable shaders we can't use + copy.shader.set_variant_enabled(COPY_MODE_GAUSSIAN_COPY, false); + copy.shader.set_variant_enabled(COPY_MODE_GAUSSIAN_COPY_8BIT, false); + copy.shader.set_variant_enabled(COPY_MODE_GAUSSIAN_GLOW, false); + copy.shader.set_variant_enabled(COPY_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE, false); + } + copy.shader_version = copy.shader.version_create(); for (int i = 0; i < COPY_MODE_MAX; i++) { - copy.pipelines[i] = RD::get_singleton()->compute_pipeline_create(copy.shader.version_get_shader(copy.shader_version, i)); + if (copy.shader.is_variant_enabled(i)) { + copy.pipelines[i] = RD::get_singleton()->compute_pipeline_create(copy.shader.version_get_shader(copy.shader_version, i)); + } } } { @@ -1510,11 +1971,22 @@ EffectsRD::EffectsRD() { // Initialize roughness Vector<String> cubemap_roughness_modes; cubemap_roughness_modes.push_back(""); - roughness.shader.initialize(cubemap_roughness_modes); - roughness.shader_version = roughness.shader.version_create(); + if (prefer_raster_effects) { + roughness.raster_shader.initialize(cubemap_roughness_modes); - roughness.pipeline = RD::get_singleton()->compute_pipeline_create(roughness.shader.version_get_shader(roughness.shader_version, 0)); + roughness.shader_version = roughness.raster_shader.version_create(); + + roughness.raster_pipeline.setup(roughness.raster_shader.version_get_shader(roughness.shader_version, 0), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + + } else { + roughness.compute_shader.initialize(cubemap_roughness_modes); + + roughness.shader_version = roughness.compute_shader.version_create(); + + roughness.compute_pipeline = RD::get_singleton()->compute_pipeline_create(roughness.compute_shader.version_get_shader(roughness.shader_version, 0)); + roughness.raster_pipeline.clear(); + } } { @@ -1524,12 +1996,16 @@ EffectsRD::EffectsRD() { tonemap_modes.push_back("\n#define USE_GLOW_FILTER_BICUBIC\n"); tonemap_modes.push_back("\n#define USE_1D_LUT\n"); tonemap_modes.push_back("\n#define USE_GLOW_FILTER_BICUBIC\n#define USE_1D_LUT\n"); + tonemap_modes.push_back("\n#define SUBPASS\n"); + tonemap_modes.push_back("\n#define SUBPASS\n#define USE_1D_LUT\n"); // multiview versions of our shaders tonemap_modes.push_back("\n#define MULTIVIEW\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_GLOW_FILTER_BICUBIC\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_1D_LUT\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_GLOW_FILTER_BICUBIC\n#define USE_1D_LUT\n"); + tonemap_modes.push_back("\n#define MULTIVIEW\n#define SUBPASS\n"); + tonemap_modes.push_back("\n#define MULTIVIEW\n#define SUBPASS\n#define USE_1D_LUT\n"); tonemap.shader.initialize(tonemap_modes); @@ -1538,6 +2014,8 @@ EffectsRD::EffectsRD() { tonemap.shader.set_variant_enabled(TONEMAP_MODE_BICUBIC_GLOW_FILTER_MULTIVIEW, false); tonemap.shader.set_variant_enabled(TONEMAP_MODE_1D_LUT_MULTIVIEW, false); tonemap.shader.set_variant_enabled(TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT_MULTIVIEW, false); + tonemap.shader.set_variant_enabled(TONEMAP_MODE_SUBPASS_MULTIVIEW, false); + tonemap.shader.set_variant_enabled(TONEMAP_MODE_SUBPASS_1D_LUT_MULTIVIEW, false); } tonemap.shader_version = tonemap.shader.version_create(); @@ -1551,7 +2029,20 @@ EffectsRD::EffectsRD() { } } - { + if (prefer_raster_effects) { + Vector<String> luminance_reduce_modes; + luminance_reduce_modes.push_back("\n#define FIRST_PASS\n"); // LUMINANCE_REDUCE_FRAGMENT_FIRST + luminance_reduce_modes.push_back("\n"); // LUMINANCE_REDUCE_FRAGMENT + luminance_reduce_modes.push_back("\n#define FINAL_PASS\n"); // LUMINANCE_REDUCE_FRAGMENT_FINAL + + luminance_reduce_raster.shader.initialize(luminance_reduce_modes); + memset(&luminance_reduce_raster.push_constant, 0, sizeof(LuminanceReduceRasterPushConstant)); + luminance_reduce_raster.shader_version = luminance_reduce_raster.shader.version_create(); + + for (int i = 0; i < LUMINANCE_REDUCE_FRAGMENT_MAX; i++) { + luminance_reduce_raster.pipelines[i].setup(luminance_reduce_raster.shader.version_get_shader(luminance_reduce_raster.shader_version, i), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } + } else { // Initialize luminance_reduce Vector<String> luminance_reduce_modes; luminance_reduce_modes.push_back("\n#define READ_TEXTURE\n"); @@ -1565,6 +2056,10 @@ EffectsRD::EffectsRD() { for (int i = 0; i < LUMINANCE_REDUCE_MAX; i++) { luminance_reduce.pipelines[i] = RD::get_singleton()->compute_pipeline_create(luminance_reduce.shader.version_get_shader(luminance_reduce.shader_version, i)); } + + for (int i = 0; i < LUMINANCE_REDUCE_FRAGMENT_MAX; i++) { + luminance_reduce_raster.pipelines[i].clear(); + } } { @@ -1583,25 +2078,43 @@ EffectsRD::EffectsRD() { cube_to_dp.pipeline.setup(shader, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), dss, RD::PipelineColorBlendState(), 0); } - { - // Initialize bokeh - Vector<String> bokeh_modes; - bokeh_modes.push_back("\n#define MODE_GEN_BLUR_SIZE\n"); - bokeh_modes.push_back("\n#define MODE_BOKEH_BOX\n"); - bokeh_modes.push_back("\n#define MODE_BOKEH_HEXAGONAL\n"); - bokeh_modes.push_back("\n#define MODE_BOKEH_CIRCULAR\n"); - bokeh_modes.push_back("\n#define MODE_COMPOSITE_BOKEH\n"); + // Initialize bokeh + Vector<String> bokeh_modes; + bokeh_modes.push_back("\n#define MODE_GEN_BLUR_SIZE\n"); + bokeh_modes.push_back("\n#define MODE_BOKEH_BOX\n#define OUTPUT_WEIGHT\n"); + bokeh_modes.push_back("\n#define MODE_BOKEH_BOX\n"); + bokeh_modes.push_back("\n#define MODE_BOKEH_HEXAGONAL\n#define OUTPUT_WEIGHT\n"); + bokeh_modes.push_back("\n#define MODE_BOKEH_HEXAGONAL\n"); + bokeh_modes.push_back("\n#define MODE_BOKEH_CIRCULAR\n#define OUTPUT_WEIGHT\n"); + bokeh_modes.push_back("\n#define MODE_COMPOSITE_BOKEH\n"); + if (prefer_raster_effects) { + bokeh.raster_shader.initialize(bokeh_modes); + + bokeh.shader_version = bokeh.raster_shader.version_create(); - bokeh.shader.initialize(bokeh_modes); + const int att_count[BOKEH_MAX] = { 1, 2, 1, 2, 1, 2, 1 }; + for (int i = 0; i < BOKEH_MAX; i++) { + RD::PipelineColorBlendState blend_state = (i == BOKEH_COMPOSITE) ? RD::PipelineColorBlendState::create_blend(att_count[i]) : RD::PipelineColorBlendState::create_disabled(att_count[i]); + bokeh.raster_pipelines[i].setup(bokeh.raster_shader.version_get_shader(bokeh.shader_version, i), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), blend_state, 0); + } + } else { + bokeh.compute_shader.initialize(bokeh_modes); + bokeh.compute_shader.set_variant_enabled(BOKEH_GEN_BOKEH_BOX_NOWEIGHT, false); + bokeh.compute_shader.set_variant_enabled(BOKEH_GEN_BOKEH_HEXAGONAL_NOWEIGHT, false); + bokeh.shader_version = bokeh.compute_shader.version_create(); - bokeh.shader_version = bokeh.shader.version_create(); + for (int i = 0; i < BOKEH_MAX; i++) { + if (bokeh.compute_shader.is_variant_enabled(i)) { + bokeh.compute_pipelines[i] = RD::get_singleton()->compute_pipeline_create(bokeh.compute_shader.version_get_shader(bokeh.shader_version, i)); + } + } for (int i = 0; i < BOKEH_MAX; i++) { - bokeh.pipelines[i] = RD::get_singleton()->compute_pipeline_create(bokeh.shader.version_get_shader(bokeh.shader_version, i)); + bokeh.raster_pipelines[i].clear(); } } - { + if (!prefer_raster_effects) { // Initialize ssao RD::SamplerState sampler; @@ -1657,10 +2170,8 @@ EffectsRD::EffectsRD() { for (int pass = 0; pass < 4; pass++) { for (int subPass = 0; subPass < sub_pass_count; subPass++) { int a = pass; - int b = subPass; - int spmap[5]{ 0, 1, 4, 3, 2 }; - b = spmap[subPass]; + int b = spmap[subPass]; float ca, sa; float angle0 = (float(a) + float(b) / float(sub_pass_count)) * Math_PI * 0.5f; @@ -1745,7 +2256,7 @@ EffectsRD::EffectsRD() { ERR_FAIL_COND(pipeline != SSAO_MAX); } - { + if (!prefer_raster_effects) { // Initialize roughness limiter Vector<String> shader_modes; shader_modes.push_back(""); @@ -1761,11 +2272,21 @@ EffectsRD::EffectsRD() { //Initialize cubemap downsampler Vector<String> cubemap_downsampler_modes; cubemap_downsampler_modes.push_back(""); - cubemap_downsampler.shader.initialize(cubemap_downsampler_modes); - cubemap_downsampler.shader_version = cubemap_downsampler.shader.version_create(); + if (prefer_raster_effects) { + cubemap_downsampler.raster_shader.initialize(cubemap_downsampler_modes); + + cubemap_downsampler.shader_version = cubemap_downsampler.raster_shader.version_create(); + + cubemap_downsampler.raster_pipeline.setup(cubemap_downsampler.raster_shader.version_get_shader(cubemap_downsampler.shader_version, 0), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } else { + cubemap_downsampler.compute_shader.initialize(cubemap_downsampler_modes); + + cubemap_downsampler.shader_version = cubemap_downsampler.compute_shader.version_create(); - cubemap_downsampler.pipeline = RD::get_singleton()->compute_pipeline_create(cubemap_downsampler.shader.version_get_shader(cubemap_downsampler.shader_version, 0)); + cubemap_downsampler.compute_pipeline = RD::get_singleton()->compute_pipeline_create(cubemap_downsampler.compute_shader.version_get_shader(cubemap_downsampler.shader_version, 0)); + cubemap_downsampler.raster_pipeline.clear(); + } } { @@ -1777,12 +2298,6 @@ EffectsRD::EffectsRD() { cubemap_filter_modes.push_back("\n#define USE_LOW_QUALITY\n"); cubemap_filter_modes.push_back("\n#define USE_HIGH_QUALITY\n#define USE_TEXTURE_ARRAY\n"); cubemap_filter_modes.push_back("\n#define USE_LOW_QUALITY\n#define USE_TEXTURE_ARRAY\n"); - filter.shader.initialize(cubemap_filter_modes); - filter.shader_version = filter.shader.version_create(); - - for (int i = 0; i < FILTER_MODE_MAX; i++) { - filter.pipelines[i] = RD::get_singleton()->compute_pipeline_create(filter.shader.version_get_shader(filter.shader_version, i)); - } if (filter.use_high_quality) { filter.coefficient_buffer = RD::get_singleton()->storage_buffer_create(sizeof(high_quality_coeffs)); @@ -1792,18 +2307,54 @@ EffectsRD::EffectsRD() { RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(low_quality_coeffs), &low_quality_coeffs[0]); } - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 0; - u.ids.push_back(filter.coefficient_buffer); - uniforms.push_back(u); + if (prefer_raster_effects) { + filter.raster_shader.initialize(cubemap_filter_modes); + + // array variants are not supported in raster + filter.raster_shader.set_variant_enabled(FILTER_MODE_HIGH_QUALITY_ARRAY, false); + filter.raster_shader.set_variant_enabled(FILTER_MODE_LOW_QUALITY_ARRAY, false); + + filter.shader_version = filter.raster_shader.version_create(); + + for (int i = 0; i < FILTER_MODE_MAX; i++) { + if (filter.raster_shader.is_variant_enabled(i)) { + filter.raster_pipelines[i].setup(filter.raster_shader.version_get_shader(filter.shader_version, i), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } else { + filter.raster_pipelines[i].clear(); + } + } + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 0; + u.ids.push_back(filter.coefficient_buffer); + uniforms.push_back(u); + } + filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.raster_shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); + } else { + filter.compute_shader.initialize(cubemap_filter_modes); + filter.shader_version = filter.compute_shader.version_create(); + + for (int i = 0; i < FILTER_MODE_MAX; i++) { + filter.compute_pipelines[i] = RD::get_singleton()->compute_pipeline_create(filter.compute_shader.version_get_shader(filter.shader_version, i)); + filter.raster_pipelines[i].clear(); + } + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 0; + u.ids.push_back(filter.coefficient_buffer); + uniforms.push_back(u); + } + filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.compute_shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); } - filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); } - { + if (!prefer_raster_effects) { Vector<String> specular_modes; specular_modes.push_back("\n#define MODE_MERGE\n"); specular_modes.push_back("\n#define MODE_MERGE\n#define MODE_SSR\n"); @@ -1839,72 +2390,74 @@ EffectsRD::EffectsRD() { } } - { - Vector<String> ssr_modes; - ssr_modes.push_back("\n"); - ssr_modes.push_back("\n#define MODE_ROUGH\n"); + if (!prefer_raster_effects) { + { + Vector<String> ssr_modes; + ssr_modes.push_back("\n"); + ssr_modes.push_back("\n#define MODE_ROUGH\n"); - ssr.shader.initialize(ssr_modes); + ssr.shader.initialize(ssr_modes); - ssr.shader_version = ssr.shader.version_create(); + ssr.shader_version = ssr.shader.version_create(); - for (int i = 0; i < SCREEN_SPACE_REFLECTION_MAX; i++) { - ssr.pipelines[i] = RD::get_singleton()->compute_pipeline_create(ssr.shader.version_get_shader(ssr.shader_version, i)); + for (int i = 0; i < SCREEN_SPACE_REFLECTION_MAX; i++) { + ssr.pipelines[i] = RD::get_singleton()->compute_pipeline_create(ssr.shader.version_get_shader(ssr.shader_version, i)); + } } - } - { - Vector<String> ssr_filter_modes; - ssr_filter_modes.push_back("\n"); - ssr_filter_modes.push_back("\n#define VERTICAL_PASS\n"); + { + Vector<String> ssr_filter_modes; + ssr_filter_modes.push_back("\n"); + ssr_filter_modes.push_back("\n#define VERTICAL_PASS\n"); - ssr_filter.shader.initialize(ssr_filter_modes); + ssr_filter.shader.initialize(ssr_filter_modes); - ssr_filter.shader_version = ssr_filter.shader.version_create(); + ssr_filter.shader_version = ssr_filter.shader.version_create(); - for (int i = 0; i < SCREEN_SPACE_REFLECTION_FILTER_MAX; i++) { - ssr_filter.pipelines[i] = RD::get_singleton()->compute_pipeline_create(ssr_filter.shader.version_get_shader(ssr_filter.shader_version, i)); + for (int i = 0; i < SCREEN_SPACE_REFLECTION_FILTER_MAX; i++) { + ssr_filter.pipelines[i] = RD::get_singleton()->compute_pipeline_create(ssr_filter.shader.version_get_shader(ssr_filter.shader_version, i)); + } } - } - { - Vector<String> ssr_scale_modes; - ssr_scale_modes.push_back("\n"); + { + Vector<String> ssr_scale_modes; + ssr_scale_modes.push_back("\n"); - ssr_scale.shader.initialize(ssr_scale_modes); + ssr_scale.shader.initialize(ssr_scale_modes); - ssr_scale.shader_version = ssr_scale.shader.version_create(); + ssr_scale.shader_version = ssr_scale.shader.version_create(); - ssr_scale.pipeline = RD::get_singleton()->compute_pipeline_create(ssr_scale.shader.version_get_shader(ssr_scale.shader_version, 0)); - } + ssr_scale.pipeline = RD::get_singleton()->compute_pipeline_create(ssr_scale.shader.version_get_shader(ssr_scale.shader_version, 0)); + } - { - Vector<String> sss_modes; - sss_modes.push_back("\n#define USE_11_SAMPLES\n"); - sss_modes.push_back("\n#define USE_17_SAMPLES\n"); - sss_modes.push_back("\n#define USE_25_SAMPLES\n"); + { + Vector<String> sss_modes; + sss_modes.push_back("\n#define USE_11_SAMPLES\n"); + sss_modes.push_back("\n#define USE_17_SAMPLES\n"); + sss_modes.push_back("\n#define USE_25_SAMPLES\n"); - sss.shader.initialize(sss_modes); + sss.shader.initialize(sss_modes); - sss.shader_version = sss.shader.version_create(); + sss.shader_version = sss.shader.version_create(); - for (int i = 0; i < sss_modes.size(); i++) { - sss.pipelines[i] = RD::get_singleton()->compute_pipeline_create(sss.shader.version_get_shader(sss.shader_version, i)); + for (int i = 0; i < sss_modes.size(); i++) { + sss.pipelines[i] = RD::get_singleton()->compute_pipeline_create(sss.shader.version_get_shader(sss.shader_version, i)); + } } - } - { - Vector<String> resolve_modes; - resolve_modes.push_back("\n#define MODE_RESOLVE_GI\n"); - resolve_modes.push_back("\n#define MODE_RESOLVE_GI\n#define VOXEL_GI_RESOLVE\n"); - resolve_modes.push_back("\n#define MODE_RESOLVE_DEPTH\n"); + { + Vector<String> resolve_modes; + resolve_modes.push_back("\n#define MODE_RESOLVE_GI\n"); + resolve_modes.push_back("\n#define MODE_RESOLVE_GI\n#define VOXEL_GI_RESOLVE\n"); + resolve_modes.push_back("\n#define MODE_RESOLVE_DEPTH\n"); - resolve.shader.initialize(resolve_modes); + resolve.shader.initialize(resolve_modes); - resolve.shader_version = resolve.shader.version_create(); + resolve.shader_version = resolve.shader.version_create(); - for (int i = 0; i < RESOLVE_MODE_MAX; i++) { - resolve.pipelines[i] = RD::get_singleton()->compute_pipeline_create(resolve.shader.version_get_shader(resolve.shader_version, i)); + for (int i = 0; i < RESOLVE_MODE_MAX; i++) { + resolve.pipelines[i] = RD::get_singleton()->compute_pipeline_create(resolve.shader.version_get_shader(resolve.shader_version, i)); + } } } @@ -1970,30 +2523,41 @@ EffectsRD::~EffectsRD() { RD::get_singleton()->free(index_buffer); //array gets freed as dependency RD::get_singleton()->free(filter.coefficient_buffer); - RD::get_singleton()->free(ssao.mirror_sampler); - RD::get_singleton()->free(ssao.gather_constants_buffer); - RD::get_singleton()->free(ssao.importance_map_load_counter); - - bokeh.shader.version_free(bokeh.shader_version); - copy.shader.version_free(copy.shader_version); + if (prefer_raster_effects) { + blur_raster.shader.version_free(blur_raster.shader_version); + bokeh.raster_shader.version_free(blur_raster.shader_version); + luminance_reduce_raster.shader.version_free(luminance_reduce_raster.shader_version); + roughness.raster_shader.version_free(roughness.shader_version); + cubemap_downsampler.raster_shader.version_free(cubemap_downsampler.shader_version); + filter.raster_shader.version_free(filter.shader_version); + } else { + bokeh.compute_shader.version_free(bokeh.shader_version); + luminance_reduce.shader.version_free(luminance_reduce.shader_version); + roughness.compute_shader.version_free(roughness.shader_version); + cubemap_downsampler.compute_shader.version_free(cubemap_downsampler.shader_version); + filter.compute_shader.version_free(filter.shader_version); + } + if (!prefer_raster_effects) { + copy.shader.version_free(copy.shader_version); + resolve.shader.version_free(resolve.shader_version); + specular_merge.shader.version_free(specular_merge.shader_version); + ssao.blur_shader.version_free(ssao.blur_shader_version); + ssao.gather_shader.version_free(ssao.gather_shader_version); + ssao.downsample_shader.version_free(ssao.downsample_shader_version); + ssao.interleave_shader.version_free(ssao.interleave_shader_version); + ssao.importance_map_shader.version_free(ssao.importance_map_shader_version); + roughness_limiter.shader.version_free(roughness_limiter.shader_version); + ssr.shader.version_free(ssr.shader_version); + ssr_filter.shader.version_free(ssr_filter.shader_version); + ssr_scale.shader.version_free(ssr_scale.shader_version); + sss.shader.version_free(sss.shader_version); + + RD::get_singleton()->free(ssao.mirror_sampler); + RD::get_singleton()->free(ssao.gather_constants_buffer); + RD::get_singleton()->free(ssao.importance_map_load_counter); + } copy_to_fb.shader.version_free(copy_to_fb.shader_version); cube_to_dp.shader.version_free(cube_to_dp.shader_version); - cubemap_downsampler.shader.version_free(cubemap_downsampler.shader_version); - filter.shader.version_free(filter.shader_version); - luminance_reduce.shader.version_free(luminance_reduce.shader_version); - resolve.shader.version_free(resolve.shader_version); - roughness.shader.version_free(roughness.shader_version); - roughness_limiter.shader.version_free(roughness_limiter.shader_version); sort.shader.version_free(sort.shader_version); - specular_merge.shader.version_free(specular_merge.shader_version); - ssao.blur_shader.version_free(ssao.blur_shader_version); - ssao.gather_shader.version_free(ssao.gather_shader_version); - ssao.downsample_shader.version_free(ssao.downsample_shader_version); - ssao.interleave_shader.version_free(ssao.interleave_shader_version); - ssao.importance_map_shader.version_free(ssao.importance_map_shader_version); - ssr.shader.version_free(ssr.shader_version); - ssr_filter.shader.version_free(ssr_filter.shader_version); - ssr_scale.shader.version_free(ssr_scale.shader_version); - sss.shader.version_free(sss.shader_version); tonemap.shader.version_free(tonemap.shader_version); } diff --git a/servers/rendering/renderer_rd/effects_rd.h b/servers/rendering/renderer_rd/effects_rd.h index 33d32f0c57..0db0919dbc 100644 --- a/servers/rendering/renderer_rd/effects_rd.h +++ b/servers/rendering/renderer_rd/effects_rd.h @@ -33,14 +33,20 @@ #include "core/math/camera_matrix.h" #include "servers/rendering/renderer_rd/pipeline_cache_rd.h" +#include "servers/rendering/renderer_rd/shaders/blur_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/bokeh_dof.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/bokeh_dof_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/copy.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/copy_to_fb.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cube_to_dp.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_filter.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/luminance_reduce.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/luminance_reduce_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/resolve.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/roughness_limiter.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/screen_space_reflection.glsl.gen.h" @@ -60,6 +66,50 @@ #include "servers/rendering_server.h" class EffectsRD { +private: + bool prefer_raster_effects; + + enum BlurRasterMode { + BLUR_MIPMAP, + + BLUR_MODE_GAUSSIAN_BLUR, + BLUR_MODE_GAUSSIAN_GLOW, + BLUR_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE, + BLUR_MODE_COPY, + + BLUR_MODE_MAX + }; + + enum { + BLUR_FLAG_HORIZONTAL = (1 << 0), + BLUR_FLAG_USE_ORTHOGONAL_PROJECTION = (1 << 1), + BLUR_FLAG_GLOW_FIRST_PASS = (1 << 2), + }; + + struct BlurRasterPushConstant { + float pixel_size[2]; + uint32_t flags; + uint32_t pad; + + //glow + float glow_strength; + float glow_bloom; + float glow_hdr_threshold; + float glow_hdr_scale; + + float glow_exposure; + float glow_white; + float glow_luminance_cap; + float glow_auto_exposure_grey; + }; + + struct BlurRaster { + BlurRasterPushConstant push_constant; + BlurRasterShaderRD shader; + RID shader_version; + PipelineCacheRD pipelines[BLUR_MODE_MAX]; + } blur_raster; + enum CopyMode { COPY_MODE_GAUSSIAN_COPY, COPY_MODE_GAUSSIAN_COPY_8BIT, @@ -161,9 +211,11 @@ class EffectsRD { struct CubemapRoughness { CubemapRoughnessPushConstant push_constant; - CubemapRoughnessShaderRD shader; + CubemapRoughnessShaderRD compute_shader; + CubemapRoughnessRasterShaderRD raster_shader; RID shader_version; - RID pipeline; + RID compute_pipeline; + PipelineCacheRD raster_pipeline; } roughness; enum TonemapMode { @@ -171,39 +223,43 @@ class EffectsRD { TONEMAP_MODE_BICUBIC_GLOW_FILTER, TONEMAP_MODE_1D_LUT, TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT, + TONEMAP_MODE_SUBPASS, + TONEMAP_MODE_SUBPASS_1D_LUT, TONEMAP_MODE_NORMAL_MULTIVIEW, TONEMAP_MODE_BICUBIC_GLOW_FILTER_MULTIVIEW, TONEMAP_MODE_1D_LUT_MULTIVIEW, TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT_MULTIVIEW, + TONEMAP_MODE_SUBPASS_MULTIVIEW, + TONEMAP_MODE_SUBPASS_1D_LUT_MULTIVIEW, TONEMAP_MODE_MAX }; struct TonemapPushConstant { - float bcs[3]; - uint32_t use_bcs; + float bcs[3]; // 12 - 12 + uint32_t use_bcs; // 4 - 16 - uint32_t use_glow; - uint32_t use_auto_exposure; - uint32_t use_color_correction; - uint32_t tonemapper; + uint32_t use_glow; // 4 - 20 + uint32_t use_auto_exposure; // 4 - 24 + uint32_t use_color_correction; // 4 - 28 + uint32_t tonemapper; // 4 - 32 - uint32_t glow_texture_size[2]; - float glow_intensity; - uint32_t pad3; + uint32_t glow_texture_size[2]; // 8 - 40 + float glow_intensity; // 4 - 44 + uint32_t pad3; // 4 - 48 - uint32_t glow_mode; - float glow_levels[7]; + uint32_t glow_mode; // 4 - 52 + float glow_levels[7]; // 28 - 80 - float exposure; - float white; - float auto_exposure_grey; - uint32_t pad2; + float exposure; // 4 - 84 + float white; // 4 - 88 + float auto_exposure_grey; // 4 - 92 + float luminance_multiplier; // 4 - 96 - float pixel_size[2]; - uint32_t use_fxaa; - uint32_t use_debanding; + float pixel_size[2]; // 8 - 104 + uint32_t use_fxaa; // 4 - 108 + uint32_t use_debanding; // 4 - 112 }; /* tonemap actually writes to a framebuffer, which is @@ -239,11 +295,33 @@ class EffectsRD { RID pipelines[LUMINANCE_REDUCE_MAX]; } luminance_reduce; + enum LuminanceReduceRasterMode { + LUMINANCE_REDUCE_FRAGMENT_FIRST, + LUMINANCE_REDUCE_FRAGMENT, + LUMINANCE_REDUCE_FRAGMENT_FINAL, + LUMINANCE_REDUCE_FRAGMENT_MAX + }; + + struct LuminanceReduceRasterPushConstant { + int32_t source_size[2]; + int32_t dest_size[2]; + float exposure_adjust; + float min_luminance; + float max_luminance; + uint32_t pad1; + }; + + struct LuminanceReduceFragment { + LuminanceReduceRasterPushConstant push_constant; + LuminanceReduceRasterShaderRD shader; + RID shader_version; + PipelineCacheRD pipelines[LUMINANCE_REDUCE_FRAGMENT_MAX]; + } luminance_reduce_raster; + struct CopyToDPPushConstant { float z_far; float z_near; - uint32_t z_flip; - uint32_t pad; + float texel_size[2]; float screen_rect[4]; }; @@ -281,7 +359,9 @@ class EffectsRD { enum BokehMode { BOKEH_GEN_BLUR_SIZE, BOKEH_GEN_BOKEH_BOX, + BOKEH_GEN_BOKEH_BOX_NOWEIGHT, BOKEH_GEN_BOKEH_HEXAGONAL, + BOKEH_GEN_BOKEH_HEXAGONAL_NOWEIGHT, BOKEH_GEN_BOKEH_CIRCULAR, BOKEH_COMPOSITE, BOKEH_MAX @@ -289,9 +369,11 @@ class EffectsRD { struct Bokeh { BokehPushConstant push_constant; - BokehDofShaderRD shader; + BokehDofShaderRD compute_shader; + BokehDofRasterShaderRD raster_shader; RID shader_version; - RID pipelines[BOKEH_MAX]; + RID compute_pipelines[BOKEH_MAX]; + PipelineCacheRD raster_pipelines[BOKEH_MAX]; } bokeh; enum SSAOMode { @@ -426,15 +508,17 @@ class EffectsRD { struct CubemapDownsamplerPushConstant { uint32_t face_size; - float pad[3]; + uint32_t face_id; + float pad[2]; }; struct CubemapDownsampler { CubemapDownsamplerPushConstant push_constant; - CubemapDownsamplerShaderRD shader; + CubemapDownsamplerShaderRD compute_shader; + CubemapDownsamplerRasterShaderRD raster_shader; RID shader_version; - RID pipeline; - + RID compute_pipeline; + PipelineCacheRD raster_pipeline; } cubemap_downsampler; enum CubemapFilterMode { @@ -445,10 +529,19 @@ class EffectsRD { FILTER_MODE_MAX, }; + struct CubemapFilterRasterPushConstant { + uint32_t mip_level; + uint32_t face_id; + float pad[2]; + }; + struct CubemapFilter { - CubemapFilterShaderRD shader; + CubemapFilterShaderRD compute_shader; + CubemapFilterRasterShaderRD raster_shader; RID shader_version; - RID pipelines[FILTER_MODE_MAX]; + RID compute_pipelines[FILTER_MODE_MAX]; + PipelineCacheRD raster_pipelines[FILTER_MODE_MAX]; + RID uniform_set; RID image_uniform_set; RID coefficient_buffer; @@ -617,6 +710,7 @@ class EffectsRD { RID index_array; Map<RID, RID> texture_to_uniform_set_cache; + Map<RID, RID> input_to_uniform_set_cache; Map<RID, RID> image_to_uniform_set_cache; @@ -650,6 +744,7 @@ class EffectsRD { Map<TextureSamplerPair, RID> texture_sampler_to_compute_uniform_set_cache; RID _get_uniform_set_from_image(RID p_texture); + RID _get_uniform_set_for_input(RID p_texture); RID _get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps = false); RID _get_compute_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps = false); RID _get_compute_uniform_set_from_texture_and_sampler(RID p_texture, RID p_sampler); @@ -657,6 +752,8 @@ class EffectsRD { RID _get_compute_uniform_set_from_image_pair(RID p_texture, RID p_texture2); public: + bool get_prefer_raster_effects(); + void copy_to_fb_rect(RID p_source_rd_texture, RID p_dest_framebuffer, const Rect2i &p_rect, bool p_flip_y = false, bool p_force_luminance = false, bool p_alpha_to_zero = false, bool p_srgb = false, RID p_secondary = RID()); void copy_to_rect(RID p_source_rd_texture, RID p_dest_texture, const Rect2i &p_rect, bool p_flip_y = false, bool p_force_luminance = false, bool p_all_source = false, bool p_8_bit_dst = false, bool p_alpha_to_one = false); void copy_cubemap_to_panorama(RID p_source_cube, RID p_dest_panorama, const Size2i &p_panorama_size, float p_lod, bool p_is_array); @@ -666,12 +763,36 @@ public: void gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back_texture, const Rect2i &p_region, bool p_8bit_dst = false); void set_color(RID p_dest_texture, const Color &p_color, const Rect2i &p_region, bool p_8bit_dst = false); void gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const Size2i &p_size, float p_strength = 1.0, bool p_high_quality = false, bool p_first_pass = false, float p_luminance_cap = 16.0, float p_exposure = 1.0, float p_bloom = 0.0, float p_hdr_bleed_treshold = 1.0, float p_hdr_bleed_scale = 1.0, RID p_auto_exposure = RID(), float p_auto_exposure_grey = 1.0); + void gaussian_glow_raster(RID p_source_rd_texture, RID p_framebuffer_half, RID p_rd_texture_half, RID p_dest_framebuffer, const Vector2 &p_pixel_size, float p_strength = 1.0, bool p_high_quality = false, bool p_first_pass = false, float p_luminance_cap = 16.0, float p_exposure = 1.0, float p_bloom = 0.0, float p_hdr_bleed_treshold = 1.0, float p_hdr_bleed_scale = 1.0, RID p_auto_exposure = RID(), float p_auto_exposure_grey = 1.0); - void cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size); + void cubemap_roughness(RID p_source_rd_texture, RID p_dest_texture, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size); + void cubemap_roughness_raster(RID p_source_rd_texture, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size); void make_mipmap(RID p_source_rd_texture, RID p_dest_texture, const Size2i &p_size); - void copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dest_texture, const Rect2 &p_rect, float p_z_near, float p_z_far, bool p_dp_flip); + void make_mipmap_raster(RID p_source_rd_texture, RID p_dest_framebuffer, const Size2i &p_size); + void copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dst_framebuffer, const Rect2 &p_rect, const Vector2 &p_dst_size, float p_z_near, float p_z_far, bool p_dp_flip); void luminance_reduction(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set = false); - void bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i &p_base_texture_size, RID p_secondary_texture, RID p_bokeh_texture1, RID p_bokeh_texture2, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_bokeh_size, RS::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, bool p_use_jitter, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal); + void luminance_reduction_raster(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, Vector<RID> p_fb, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set = false); + + struct BokehBuffers { + // bokeh buffers + + // textures + Size2i base_texture_size; + RID base_texture; + RID depth_texture; + RID secondary_texture; + RID half_texture[2]; + + // raster only + RID base_fb; + RID secondary_fb; // with weights + RID half_fb[2]; // with weights + RID base_weight_fb; + RID weight_texture[4]; + }; + + void bokeh_dof(const BokehBuffers &p_buffers, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_bokeh_size, RS::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, bool p_use_jitter, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal); + void bokeh_dof_raster(const BokehBuffers &p_buffers, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_dof_blur_amount, RenderingServer::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal); struct TonemapSettings { bool use_glow = false; @@ -697,6 +818,7 @@ public: bool use_auto_exposure = false; float auto_exposure_grey = 0.5; RID exposure_texture; + float luminance_multiplier = 1.0; bool use_bcs = false; float brightness = 1.0; @@ -734,13 +856,16 @@ public: }; void tonemapper(RID p_source_color, RID p_dst_framebuffer, const TonemapSettings &p_settings); + void tonemapper(RD::DrawListID p_subpass_draw_list, RID p_source_color, RD::FramebufferFormatID p_dst_format_id, const TonemapSettings &p_settings); void gather_ssao(RD::ComputeListID p_compute_list, const Vector<RID> p_ao_slices, const SSAOSettings &p_settings, bool p_adaptive_base_pass, RID p_gather_uniform_set, RID p_importance_map_uniform_set); void generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_depth_mipmaps_texture, const Vector<RID> &depth_mipmaps, RID p_ao, const Vector<RID> p_ao_slices, RID p_ao_pong, const Vector<RID> p_ao_pong_slices, RID p_upscale_buffer, RID p_importance_map, RID p_importance_map_pong, const CameraMatrix &p_projection, const SSAOSettings &p_settings, bool p_invalidate_uniform_sets, RID &r_downsample_uniform_set, RID &r_gather_uniform_set, RID &r_importance_map_uniform_set); void roughness_limit(RID p_source_normal, RID p_roughness, const Size2i &p_size, float p_curve); void cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, const Size2i &p_size); + void cubemap_downsample_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, const Size2i &p_size); void cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, bool p_use_array); + void cubemap_filter_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_mip_level); void screen_space_reflection(RID p_diffuse, RID p_normal_roughness, RS::EnvironmentSSRRoughnessQuality p_roughness_quality, RID p_blur_radius, RID p_blur_radius2, RID p_metallic, const Color &p_metallic_mask, RID p_depth, RID p_scale_depth, RID p_scale_normal, RID p_output, RID p_output_blur, const Size2i &p_screen_size, int p_max_steps, float p_fade_in, float p_fade_out, float p_tolerance, const CameraMatrix &p_camera); void merge_specular(RID p_dest_framebuffer, RID p_specular, RID p_base, RID p_reflection); @@ -751,7 +876,7 @@ public: void sort_buffer(RID p_uniform_set, int p_size); - EffectsRD(); + EffectsRD(bool p_prefer_raster_effects); ~EffectsRD(); }; diff --git a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp index ac20515c28..0deb822e86 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -183,7 +183,7 @@ void RenderForwardClustered::RenderBufferDataForwardClustered::clear() { } } -void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { +void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { clear(); ERR_FAIL_COND_MSG(p_view_count != 1, "Multiple views is currently not supported in this renderer, please use the mobile renderer for VR support"); @@ -223,7 +223,6 @@ void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_c RD::TEXTURE_SAMPLES_2, RD::TEXTURE_SAMPLES_4, RD::TEXTURE_SAMPLES_8, - RD::TEXTURE_SAMPLES_16 }; texture_samples = ts[p_msaa]; @@ -329,6 +328,8 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p push_constant.uv_offset = 0; } + bool should_request_redraw = false; + for (uint32_t i = p_from_element; i < p_to_element; i++) { const GeometryInstanceSurfaceDataCache *surf = p_params->elements[i]; const RenderElementInfo &element_info = p_params->element_info[i]; @@ -366,6 +367,11 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p continue; } + //request a redraw if one of the shaders uses TIME + if (shader->uses_time) { + should_request_redraw = true; + } + //find cull variant SceneShaderForwardClustered::ShaderData::CullVariant cull_variant; @@ -483,8 +489,8 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p } if (material_uniform_set != prev_material_uniform_set) { - //update uniform set - if (material_uniform_set.is_valid()) { + // Update uniform set. + if (material_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(material_uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_uniform_set, MATERIAL_UNIFORM_SET); } @@ -501,6 +507,11 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p RD::get_singleton()->draw_list_draw(draw_list, index_array_rd.is_valid(), instance_count); i += element_info.repeat - 1; //skip equal elements } + + // Make the actual redraw request + if (should_request_redraw) { + RenderingServerDefault::redraw_request(); + } } void RenderForwardClustered::_render_list(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { @@ -754,19 +765,10 @@ void RenderForwardClustered::_setup_environment(const RenderDataRD *p_render_dat scene_state.ubo.ssao_ao_affect = environment_get_ssao_ao_affect(p_render_data->environment); scene_state.ubo.ssao_light_affect = environment_get_ssao_light_affect(p_render_data->environment); - Color ao_color = environment_get_ao_color(p_render_data->environment).to_linear(); - scene_state.ubo.ao_color[0] = ao_color.r; - scene_state.ubo.ao_color[1] = ao_color.g; - scene_state.ubo.ao_color[2] = ao_color.b; - scene_state.ubo.ao_color[3] = ao_color.a; - scene_state.ubo.fog_enabled = environment_is_fog_enabled(p_render_data->environment); scene_state.ubo.fog_density = environment_get_fog_density(p_render_data->environment); scene_state.ubo.fog_height = environment_get_fog_height(p_render_data->environment); scene_state.ubo.fog_height_density = environment_get_fog_height_density(p_render_data->environment); - if (scene_state.ubo.fog_height_density >= 0.0001) { - scene_state.ubo.fog_height_density = 1.0 / scene_state.ubo.fog_height_density; - } scene_state.ubo.fog_aerial_perspective = environment_get_fog_aerial_perspective(p_render_data->environment); Color fog_color = environment_get_fog_light_color(p_render_data->environment).to_linear(); @@ -914,7 +916,7 @@ void RenderForwardClustered::_fill_render_list(RenderListType p_render_list, con } uint32_t lightmap_captures_used = 0; - Plane near_plane(p_render_data->cam_transform.origin, -p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z)); + Plane near_plane = Plane(-p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z), p_render_data->cam_transform.origin); near_plane.d += p_render_data->cam_projection.get_z_near(); float z_max = p_render_data->cam_projection.get_z_far() - p_render_data->cam_projection.get_z_near(); @@ -1163,7 +1165,7 @@ void RenderForwardClustered::_render_scene(RenderDataRD *p_render_data, const Co render_buffer = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_data->render_buffers); } RendererSceneEnvironmentRD *env = get_environment(p_render_data->environment); - static const int texture_multisamples[RS::VIEWPORT_MSAA_MAX] = { 1, 2, 4, 8, 16 }; + static const int texture_multisamples[RS::VIEWPORT_MSAA_MAX] = { 1, 2, 4, 8 }; //first of all, make a new render pass //fill up ubo @@ -1533,6 +1535,16 @@ void RenderForwardClustered::_render_scene(RenderDataRD *p_render_data, const Co } } + if (scene_state.used_screen_texture) { + // Copy screen texture to backbuffer so we can read from it + _render_buffers_copy_screen_texture(p_render_data); + } + + if (scene_state.used_depth_texture) { + // Copy depth texture to backbuffer so we can read from it + _render_buffers_copy_depth_texture(p_render_data); + } + RENDER_TIMESTAMP("Render Transparent Pass"); RD::get_singleton()->draw_command_begin_label("Render Transparent Pass"); @@ -1555,6 +1567,14 @@ void RenderForwardClustered::_render_scene(RenderDataRD *p_render_data, const Co } RD::get_singleton()->draw_command_end_label(); + + if (p_render_data->render_buffers.is_valid()) { + _debug_draw_cluster(p_render_data->render_buffers); + + RENDER_TIMESTAMP("Tonemap"); + + _render_buffers_post_process_and_tonemap(p_render_data); + } } void RenderForwardClustered::_render_shadow_begin() { @@ -2211,7 +2231,8 @@ RID RenderForwardClustered::_setup_render_pass_uniform_set(RenderListType p_rend RD::Uniform u; u.binding = 9; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - RID texture = (false && rb && rb->depth.is_valid()) ? rb->depth : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); + RID dbt = rb ? render_buffers_get_back_depth_texture(p_render_data->render_buffers) : RID(); + RID texture = (dbt.is_valid()) ? dbt : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); u.ids.push_back(texture); uniforms.push_back(u); } @@ -2533,7 +2554,7 @@ void RenderForwardClustered::_geometry_instance_add_surface_with_material(Geomet SceneShaderForwardClustered::MaterialData *material_shadow = nullptr; void *surface_shadow = nullptr; - if (!p_material->shader_data->uses_particle_trails && !p_material->shader_data->writes_modelview_or_projection && !p_material->shader_data->uses_vertex && !p_material->shader_data->uses_discard && !p_material->shader_data->uses_depth_pre_pass) { + if (!p_material->shader_data->uses_particle_trails && !p_material->shader_data->writes_modelview_or_projection && !p_material->shader_data->uses_vertex && !p_material->shader_data->uses_position && !p_material->shader_data->uses_discard && !p_material->shader_data->uses_depth_pre_pass) { flags |= GeometryInstanceSurfaceDataCache::FLAG_USES_SHARED_SHADOW_MATERIAL; material_shadow = (SceneShaderForwardClustered::MaterialData *)storage->material_get_data(scene_shader.default_material, RendererStorageRD::SHADER_TYPE_3D); @@ -2677,7 +2698,7 @@ void RenderForwardClustered::_geometry_instance_update(GeometryInstance *p_geome } break; #if 0 case RS::INSTANCE_IMMEDIATE: { - RasterizerStorageGLES3::Immediate *immediate = storage->immediate_owner.getornull(inst->base); + RasterizerStorageGLES3::Immediate *immediate = storage->immediate_owner.get_or_null(inst->base); ERR_CONTINUE(!immediate); _add_geometry(immediate, inst, nullptr, -1, p_depth_pass, p_shadow_pass); @@ -3043,7 +3064,7 @@ RenderForwardClustered::RenderForwardClustered(RendererStorageRD *p_storage) : defines += "\n#define USE_RADIANCE_CUBEMAP_ARRAY \n"; } defines += "\n#define SDFGI_OCT_SIZE " + itos(gi.sdfgi_get_lightprobe_octahedron_size()) + "\n"; - defines += "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(get_max_directional_lights()) + "\n"; + defines += "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(MAX_DIRECTIONAL_LIGHTS) + "\n"; { //lightmaps diff --git a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h index 6682c5e9b0..ff3d2fc6cb 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h @@ -112,7 +112,7 @@ class RenderForwardClustered : public RendererSceneRenderRD { void ensure_specular(); void ensure_voxelgi(); void clear(); - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); ~RenderBufferDataForwardClustered(); }; @@ -257,8 +257,6 @@ class RenderForwardClustered : public RendererSceneRenderRD { float roughness_limiter_limit; uint32_t roughness_limiter_pad[2]; - float ao_color[4]; - float sdf_to_bounds[16]; int32_t sdf_offset[3]; diff --git a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp index 333e87bdbd..a1c3481ed6 100644 --- a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp @@ -66,6 +66,7 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { unshaded = false; uses_vertex = false; + uses_position = false; uses_sss = false; uses_transmittance = false; uses_screen_texture = false; @@ -126,13 +127,13 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { actions.write_flag_pointers["MODELVIEW_MATRIX"] = &writes_modelview_or_projection; actions.write_flag_pointers["PROJECTION_MATRIX"] = &writes_modelview_or_projection; actions.write_flag_pointers["VERTEX"] = &uses_vertex; + actions.write_flag_pointers["POSITION"] = &uses_position; actions.uniforms = &uniforms; SceneShaderForwardClustered *shader_singleton = (SceneShaderForwardClustered *)SceneShaderForwardClustered::singleton; Error err = shader_singleton->compiler.compile(RS::SHADER_SPATIAL, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); + ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); if (version.is_null()) { version = shader_singleton->shader.version_create(); @@ -343,36 +344,36 @@ void SceneShaderForwardClustered::ShaderData::set_default_texture_param(const St void SceneShaderForwardClustered::ShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { Map<int, StringName> order; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { continue; } - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); + if (E.value.texture_order >= 0) { + order[E.value.texture_order + 100000] = E.key; } else { - order[E->get().order] = E->key(); + order[E.value.order] = E.key; } } - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); + for (const KeyValue<int, StringName> &E : order) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E.value]); + pi.name = E.value; p_param_list->push_back(pi); } } void SceneShaderForwardClustered::ShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p.info = ShaderLanguage::uniform_to_property_info(E.value); + p.info.name = E.key; //supply name + p.index = E.value.instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E.value.default_value, E.value.type, E.value.array_size, E.value.hint); p_param_list->push_back(p); } } @@ -397,7 +398,7 @@ Variant SceneShaderForwardClustered::ShaderData::get_default_parameter(const Str if (uniforms.has(p_parameter)) { ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.array_size, uniform.hint); } return Variant(); } @@ -602,10 +603,10 @@ void SceneShaderForwardClustered::init(RendererStorageRD *p_storage, const Strin actions.usage_defines["UV2"] = "#define UV2_USED\n"; actions.usage_defines["BONE_INDICES"] = "#define BONES_USED\n"; actions.usage_defines["BONE_WEIGHTS"] = "#define WEIGHTS_USED\n"; - actions.usage_defines["CUSTOM0"] = "#define CUSTOM0\n"; - actions.usage_defines["CUSTOM1"] = "#define CUSTOM1\n"; - actions.usage_defines["CUSTOM2"] = "#define CUSTOM2\n"; - actions.usage_defines["CUSTOM3"] = "#define CUSTOM3\n"; + actions.usage_defines["CUSTOM0"] = "#define CUSTOM0_USED\n"; + actions.usage_defines["CUSTOM1"] = "#define CUSTOM1_USED\n"; + actions.usage_defines["CUSTOM2"] = "#define CUSTOM2_USED\n"; + actions.usage_defines["CUSTOM3"] = "#define CUSTOM3_USED\n"; actions.usage_defines["NORMAL_MAP"] = "#define NORMAL_MAP_USED\n"; actions.usage_defines["NORMAL_MAP_DEPTH"] = "@NORMAL_MAP"; actions.usage_defines["COLOR"] = "#define COLOR_USED\n"; @@ -684,6 +685,8 @@ void SceneShaderForwardClustered::init(RendererStorageRD *p_storage, const Strin default_shader = storage->shader_allocate(); storage->shader_initialize(default_shader); storage->shader_set_code(default_shader, R"( +// Default 3D material shader (clustered). + shader_type spatial; void vertex() { @@ -713,6 +716,8 @@ void fragment() { storage->shader_initialize(overdraw_material_shader); // Use relatively low opacity so that more "layers" of overlapping objects can be distinguished. storage->shader_set_code(overdraw_material_shader, R"( +// 3D editor Overdraw debug draw mode shader (clustered). + shader_type spatial; render_mode blend_add, unshaded; diff --git a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h index 8d75f30a20..09ef425e2e 100644 --- a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h +++ b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h @@ -137,6 +137,7 @@ public: bool unshaded; bool uses_vertex; + bool uses_position; bool uses_sss; bool uses_transmittance; bool uses_screen_texture; diff --git a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp index 438bbff53c..1e5854a174 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp @@ -78,34 +78,70 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::clear() { color = RID(); depth = RID(); - color_fb = RID(); + for (int i = 0; i < FB_CONFIG_MAX; i++) { + color_fbs[i] = RID(); + } } -void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { +void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { clear(); msaa = p_msaa; + Size2i target_size = RD::get_singleton()->texture_size(p_target_buffer); + width = p_width; height = p_height; + bool is_scaled = (target_size.width != p_width) || (target_size.height != p_height); view_count = p_view_count; color = p_color_buffer; depth = p_depth_buffer; - RD::DataFormat color_format = RenderForwardMobile::singleton->_render_buffers_get_color_format(); + // We are creating 4 configurations here for our framebuffers. if (p_msaa == RS::VIEWPORT_MSAA_DISABLED) { - if (color_format == RD::DATA_FORMAT_A2B10G10R10_UNORM_PACK32) { - // @TODO add a second color buffer for alpha as this format is RGB only - } - Vector<RID> fb; - fb.push_back(p_color_buffer); - fb.push_back(depth); + fb.push_back(p_color_buffer); // 0 - color buffer + fb.push_back(depth); // 1 - depth buffer + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + color_fbs[FB_CONFIG_ONE_PASS] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + + // - add sky pass + passes.push_back(pass); + color_fbs[FB_CONFIG_TWO_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + + // - add alpha pass + passes.push_back(pass); + color_fbs[FB_CONFIG_THREE_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); - color_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, view_count); + if (!is_scaled) { + // - add blit to 2D pass + fb.push_back(p_target_buffer); // 2 - target buffer + + RD::FramebufferPass blit_pass; + blit_pass.color_attachments.push_back(2); + blit_pass.input_attachments.push_back(0); + passes.push_back(blit_pass); + + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + } else { + // can't do our blit pass if resolutions don't match + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RID(); + } } else { + RD::DataFormat color_format = RenderForwardMobile::singleton->_render_buffers_get_color_format(); + RD::TextureFormat tf; if (view_count > 1) { tf.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; @@ -123,7 +159,6 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_b RD::TEXTURE_SAMPLES_2, RD::TEXTURE_SAMPLES_4, RD::TEXTURE_SAMPLES_8, - RD::TEXTURE_SAMPLES_16 }; texture_samples = ts[p_msaa]; @@ -138,14 +173,87 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_b { Vector<RID> fb; - fb.push_back(color_msaa); - fb.push_back(depth_msaa); + fb.push_back(color_msaa); // 0 - msaa color buffer + fb.push_back(depth_msaa); // 1 - msaa depth buffer + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + + // - add sky pass + fb.push_back(color); // 2 - color buffer + passes.push_back(pass); // without resolve for our 3 + 4 subpass config + { + // but with resolve for our 2 subpass config + Vector<RD::FramebufferPass> two_passes; + two_passes.push_back(pass); // opaque subpass without resolve + pass.resolve_attachments.push_back(2); + two_passes.push_back(pass); // sky subpass with resolve + + color_fbs[FB_CONFIG_TWO_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, two_passes, RenderingDevice::INVALID_ID, view_count); + } + + // - add alpha pass (with resolve, we just added that above) + passes.push_back(pass); + color_fbs[FB_CONFIG_THREE_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + + { + // we also need our one pass with resolve + Vector<RD::FramebufferPass> one_pass_with_resolve; + one_pass_with_resolve.push_back(pass); // note our pass configuration already has resolve.. + color_fbs[FB_CONFIG_ONE_PASS] = RD::get_singleton()->framebuffer_create_multipass(fb, one_pass_with_resolve, RenderingDevice::INVALID_ID, view_count); + } - color_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, view_count); + if (!is_scaled) { + // - add blit to 2D pass + fb.push_back(p_target_buffer); // 3 - target buffer + RD::FramebufferPass blit_pass; + blit_pass.color_attachments.push_back(3); + blit_pass.input_attachments.push_back(2); + passes.push_back(blit_pass); + + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + } else { + // can't do our blit pass if resolutions don't match + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RID(); + } } } } +RID RenderForwardMobile::reflection_probe_create_framebuffer(RID p_color, RID p_depth) { + // Our attachments + Vector<RID> fb; + fb.push_back(p_color); // 0 + fb.push_back(p_depth); // 1 + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + + // - sky pass + passes.push_back(pass); + + // - alpha pass + passes.push_back(pass); + + return RD::get_singleton()->framebuffer_create_multipass(fb, passes); +} + RenderForwardMobile::RenderBufferDataForwardMobile::~RenderBufferDataForwardMobile() { clear(); } @@ -163,17 +271,20 @@ bool RenderForwardMobile::free(RID p_rid) { /* Render functions */ +float RenderForwardMobile::_render_buffers_get_luminance_multiplier() { + // On mobile renderer we need to multiply source colors by 2 due to using a UNORM buffer + // and multiplying by the output color during 3D rendering by 0.5 + return 2.0; +} + RD::DataFormat RenderForwardMobile::_render_buffers_get_color_format() { - // Using 32bit buffers enables AFBC on mobile devices which should have a definate performance improvement (MALI G710 and newer support this on 64bit RTs) - // NO ALPHA and unsigned float. - // @TODO No alpha is an issue, recommendation here is to add a second RT for alpha + // Using 32bit buffers enables AFBC on mobile devices which should have a definite performance improvement (MALI G710 and newer support this on 64bit RTs) return RD::DATA_FORMAT_A2B10G10R10_UNORM_PACK32; } bool RenderForwardMobile::_render_buffers_can_be_storage() { - // Using 32bit buffers enables AFBC on mobile devices which should have a definate performance improvement (MALI G710 and newer support this on 64bit RTs) - // NO ALPHA and unsigned float. - // @TODO No alpha is an issue, recommendation here is to add a second RT for alpha + // Using 32bit buffers enables AFBC on mobile devices which should have a definite performance improvement (MALI G710 and newer support this on 64bit RTs) + // Doesn't support storage return false; } @@ -309,7 +420,8 @@ RID RenderForwardMobile::_setup_render_pass_uniform_set(RenderListType p_render_ RD::Uniform u; u.binding = 9; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - RID texture = (false && rb && rb->depth.is_valid()) ? rb->depth : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); + RID dbt = rb ? render_buffers_get_back_depth_texture(p_render_data->render_buffers) : RID(); + RID texture = (dbt.is_valid()) ? dbt : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); u.ids.push_back(texture); uniforms.push_back(u); } @@ -372,14 +484,28 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color scene_state.ubo.viewport_size[1] = vp_he.y; scene_state.ubo.directional_light_count = 0; + // We can only use our full subpass approach if we're: + // - not reading from SCREEN_TEXTURE/DEPTH_TEXTURE + // - not using ssr/sss (currently not supported) + // - not using glow or other post effects (can't do 4th subpass) + // - rendering to a half sized render buffer (can't do 4th subpass) + // We'll need to restrict how far we're going with subpasses based on this. + Size2i screen_size; - RID opaque_framebuffer; - RID alpha_framebuffer; + RID framebuffer; bool reverse_cull = false; + bool using_subpass_transparent = true; + bool using_subpass_post_process = true; - // I don't think we support either of these in our mobile renderer so probably should phase them out - bool using_ssr = false; - bool using_sss = false; + bool using_ssr = false; // I don't think we support this in our mobile renderer so probably should phase it out + bool using_sss = false; // I don't think we support this in our mobile renderer so probably should phase it out + + // fill our render lists early so we can find out if we use various features + _fill_render_list(RENDER_LIST_OPAQUE, p_render_data, PASS_MODE_COLOR); + render_list[RENDER_LIST_OPAQUE].sort_by_key(); + render_list[RENDER_LIST_ALPHA].sort_by_reverse_depth_and_priority(); + _fill_element_info(RENDER_LIST_OPAQUE); + _fill_element_info(RENDER_LIST_ALPHA); if (p_render_data->render_info) { p_render_data->render_info->info[RS::VIEWPORT_RENDER_INFO_TYPE_VISIBLE][RS::VIEWPORT_RENDER_INFO_DRAW_CALLS_IN_FRAME] = p_render_data->instances->size(); @@ -391,15 +517,36 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color screen_size.x = render_buffer->width; screen_size.y = render_buffer->height; - opaque_framebuffer = render_buffer->color_fb; - alpha_framebuffer = opaque_framebuffer; + if (render_buffer->color_fbs[FB_CONFIG_FOUR_SUBPASSES].is_null()) { + // can't do blit subpass + using_subpass_post_process = false; + } else if (env && (env->glow_enabled || env->auto_exposure || camera_effects_uses_dof(p_render_data->camera_effects))) { + // can't do blit subpass + using_subpass_post_process = false; + } + + if (using_ssr || using_sss || scene_state.used_screen_texture || scene_state.used_depth_texture) { + // can't use our last two subpasses + using_subpass_transparent = false; + using_subpass_post_process = false; + } + + if (using_subpass_post_process) { + // all as subpasses + framebuffer = render_buffer->color_fbs[FB_CONFIG_FOUR_SUBPASSES]; + } else if (using_subpass_transparent) { + // our tonemap pass is separate + framebuffer = render_buffer->color_fbs[FB_CONFIG_THREE_SUBPASSES]; + } else { + // only opaque and sky as subpasses + framebuffer = render_buffer->color_fbs[FB_CONFIG_TWO_SUBPASSES]; + } } else if (p_render_data->reflection_probe.is_valid()) { uint32_t resolution = reflection_probe_instance_get_resolution(p_render_data->reflection_probe); screen_size.x = resolution; screen_size.y = resolution; - opaque_framebuffer = reflection_probe_instance_get_framebuffer(p_render_data->reflection_probe, p_render_data->reflection_probe_pass); - alpha_framebuffer = opaque_framebuffer; + framebuffer = reflection_probe_instance_get_framebuffer(p_render_data->reflection_probe, p_render_data->reflection_probe_pass); if (storage->reflection_probe_is_interior(reflection_probe_instance_get_probe(p_render_data->reflection_probe))) { p_render_data->environment = RID(); //no environment on interiors @@ -407,6 +554,8 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color } reverse_cull = true; + using_subpass_transparent = true; // we ignore our screen/depth texture here + using_subpass_post_process = false; // not applicable at all for reflection probes. } else { ERR_FAIL(); //bug? } @@ -418,17 +567,7 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color _update_render_base_uniform_set(); //may have changed due to the above (light buffer enlarged, as an example) - _fill_render_list(RENDER_LIST_OPAQUE, p_render_data, PASS_MODE_COLOR); - render_list[RENDER_LIST_OPAQUE].sort_by_key(); - render_list[RENDER_LIST_ALPHA].sort_by_reverse_depth_and_priority(); - - // we no longer use this... - _fill_instance_data(RENDER_LIST_OPAQUE); - _fill_instance_data(RENDER_LIST_ALPHA); - - RD::get_singleton()->draw_command_end_label(); - - // note, no depth prepass here! + RD::get_singleton()->draw_command_end_label(); // Render Setup // setup environment RID radiance_texture; @@ -497,103 +636,234 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color RID sky_rid = env->sky; if (sky_rid.is_valid()) { - sky.update(env, projection, p_render_data->cam_transform, time); + sky.update(env, projection, p_render_data->cam_transform, time, _render_buffers_get_luminance_multiplier()); radiance_texture = sky.sky_get_radiance_texture_rd(sky_rid); } else { // do not try to draw sky if invalid draw_sky = false; } - RD::get_singleton()->draw_command_end_label(); + RD::get_singleton()->draw_command_end_label(); // Setup Sky } } else { clear_color = p_default_bg_color; } - // opaque pass + // update sky buffers (if required) + if (draw_sky || draw_sky_fog_only) { + // !BAS! @TODO See if we can limit doing some things double and maybe even move this into _pre_opaque_render + // and change Forward Clustered in the same way as we have here (but without using subpasses) + RENDER_TIMESTAMP("Setup Sky resolution buffers"); - // !BAS! Look into this, seems most of the code in here related to clustered only, may want to move this code into ForwardClustered/RenderForwardMobile before calling it from here - // does trigger shadow map rendering so kinda important - _pre_opaque_render(p_render_data, false, false, RID(), RID()); + RD::get_singleton()->draw_command_begin_label("Setup Sky resolution buffers"); - RD::get_singleton()->draw_command_begin_label("Render Opaque Pass"); + if (p_render_data->reflection_probe.is_valid()) { + CameraMatrix correction; + correction.set_depth_correction(true); + CameraMatrix projection = correction * p_render_data->cam_projection; + sky.update_res_buffers(env, 1, &projection, p_render_data->cam_transform, time); + } else { + sky.update_res_buffers(env, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time); + } - scene_state.ubo.directional_light_count = p_render_data->directional_light_count; + RD::get_singleton()->draw_command_end_label(); // Setup Sky resolution buffers + } - _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, p_render_data->render_buffers.is_valid()); + _pre_opaque_render(p_render_data, false, false, RID(), RID()); - RENDER_TIMESTAMP("Render Opaque Pass"); + uint32_t spec_constant_base_flags = 0; - RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, p_render_data, radiance_texture, true); + { + //figure out spec constants - bool can_continue_color = !scene_state.used_screen_texture && !using_ssr && !using_sss; - bool can_continue_depth = !scene_state.used_depth_texture && !using_ssr && !using_sss; + if (p_render_data->directional_light_count > 0) { + if (p_render_data->directional_light_soft_shadows) { + spec_constant_base_flags |= 1 << SPEC_CONSTANT_USING_DIRECTIONAL_SOFT_SHADOWS; + } + } else { + spec_constant_base_flags |= 1 << SPEC_CONSTANT_DISABLE_DIRECTIONAL_LIGHTS; + } + if (!is_environment(p_render_data->environment) || environment_is_fog_enabled(p_render_data->environment)) { + spec_constant_base_flags |= 1 << SPEC_CONSTANT_DISABLE_FOG; + } + } { - bool will_continue_color = (can_continue_color || draw_sky || draw_sky_fog_only); - bool will_continue_depth = (can_continue_depth || draw_sky || draw_sky_fog_only); + if (render_buffer) { + RD::get_singleton()->draw_command_begin_label("Render 3D Pass"); + } else { + RD::get_singleton()->draw_command_begin_label("Render Reflection Probe Pass"); + } - // regular forward for now - Vector<Color> c; - c.push_back(clear_color.to_linear()); + // opaque pass - RenderListParameters render_list_params(render_list[RENDER_LIST_OPAQUE].elements.ptr(), render_list[RENDER_LIST_OPAQUE].element_info.ptr(), render_list[RENDER_LIST_OPAQUE].elements.size(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); - _render_list_with_threads(&render_list_params, opaque_framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); - } + RD::get_singleton()->draw_command_begin_label("Render Opaque Subpass"); - RD::get_singleton()->draw_command_end_label(); + scene_state.ubo.directional_light_count = p_render_data->directional_light_count; - if (draw_sky || draw_sky_fog_only) { - RENDER_TIMESTAMP("Render Sky"); + _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, p_render_data->render_buffers.is_valid()); - RD::get_singleton()->draw_command_begin_label("Draw Sky"); + if (using_subpass_transparent && using_subpass_post_process) { + RENDER_TIMESTAMP("Render Opaque + Transparent + Tonemap"); + } else if (using_subpass_transparent) { + RENDER_TIMESTAMP("Render Opaque + Transparent"); + } else { + RENDER_TIMESTAMP("Render Opaque"); + } - if (p_render_data->reflection_probe.is_valid()) { - CameraMatrix correction; - correction.set_depth_correction(true); - CameraMatrix projection = correction * p_render_data->cam_projection; - sky.draw(env, can_continue_color, can_continue_depth, opaque_framebuffer, 1, &projection, p_render_data->cam_transform, time); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, p_render_data, radiance_texture, true); + + bool can_continue_color = !using_subpass_transparent && !scene_state.used_screen_texture && !using_ssr && !using_sss; + bool can_continue_depth = !using_subpass_transparent && !scene_state.used_depth_texture && !using_ssr && !using_sss; + + { + // regular forward for now + Vector<Color> c; + c.push_back(clear_color.to_linear()); // our render buffer + if (render_buffer) { + if (render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { + c.push_back(clear_color.to_linear()); // our resolve buffer + } + if (using_subpass_post_process) { + c.push_back(Color()); // our 2D buffer we're copying into + } + } + + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + RenderListParameters render_list_params(render_list[RENDER_LIST_OPAQUE].elements.ptr(), render_list[RENDER_LIST_OPAQUE].element_info.ptr(), render_list[RENDER_LIST_OPAQUE].elements.size(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, spec_constant_base_flags, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_begin_split(framebuffer, thread_draw_lists.size(), thread_draw_lists.ptr(), keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, can_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, can_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, can_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, can_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + } + } + + RD::get_singleton()->draw_command_end_label(); //Render Opaque Subpass + + if (draw_sky || draw_sky_fog_only) { + RD::get_singleton()->draw_command_begin_label("Draw Sky Subpass"); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + + if (p_render_data->reflection_probe.is_valid()) { + CameraMatrix correction; + correction.set_depth_correction(true); + CameraMatrix projection = correction * p_render_data->cam_projection; + sky.draw(draw_list, env, framebuffer, 1, &projection, p_render_data->cam_transform, time, _render_buffers_get_luminance_multiplier()); + } else { + sky.draw(draw_list, env, framebuffer, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time, _render_buffers_get_luminance_multiplier()); + } + + RD::get_singleton()->draw_command_end_label(); // Draw Sky Subpass + + // note, if MSAA is used in 2-subpass approach we should get an automatic resolve here } else { - sky.draw(env, can_continue_color, can_continue_depth, opaque_framebuffer, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time); + // switch to subpass but we do nothing here so basically we skip (though this should trigger resolve with 2-subpass MSAA). + RD::get_singleton()->draw_list_switch_to_next_pass(); } - RD::get_singleton()->draw_command_end_label(); - } - if (render_buffer && !can_continue_color && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color); - /* - if (using_separate_specular) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->specular_msaa, render_buffer->specular); + if (!using_subpass_transparent) { + // We're done with our subpasses so end our container pass + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + + RD::get_singleton()->draw_command_end_label(); // Render 3D Pass / Render Reflection Probe Pass } - */ - } - if (render_buffer && !can_continue_depth && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->depth_msaa, render_buffer->depth); - } + if (scene_state.used_screen_texture) { + // Copy screen texture to backbuffer so we can read from it + _render_buffers_copy_screen_texture(p_render_data); + } - // transparent pass - RENDER_TIMESTAMP("Render Transparent Pass"); + if (scene_state.used_depth_texture) { + // Copy depth texture to backbuffer so we can read from it + _render_buffers_copy_depth_texture(p_render_data); + } - RD::get_singleton()->draw_command_begin_label("Render Transparent Pass"); + // transparent pass - rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_ALPHA, p_render_data, radiance_texture, true); + RD::get_singleton()->draw_command_begin_label("Render Transparent Subpass"); - _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, false); + rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_ALPHA, p_render_data, radiance_texture, true); - { - RenderListParameters render_list_params(render_list[RENDER_LIST_ALPHA].elements.ptr(), render_list[RENDER_LIST_ALPHA].element_info.ptr(), render_list[RENDER_LIST_ALPHA].elements.size(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); - _render_list_with_threads(&render_list_params, alpha_framebuffer, can_continue_color ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, can_continue_depth ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); + if (using_subpass_transparent) { + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + RenderListParameters render_list_params(render_list[RENDER_LIST_ALPHA].elements.ptr(), render_list[RENDER_LIST_ALPHA].element_info.ptr(), render_list[RENDER_LIST_ALPHA].elements.size(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, spec_constant_base_flags, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_switch_to_next_pass_split(thread_draw_lists.size(), thread_draw_lists.ptr()); + render_list_params.subpass = RD::get_singleton()->draw_list_get_current_pass(); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + render_list_params.subpass = RD::get_singleton()->draw_list_get_current_pass(); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + } + + RD::get_singleton()->draw_command_end_label(); // Render Transparent Subpass + + // note if we are using MSAA we should get an automatic resolve through our subpass configuration. + + // blit to tonemap + if (render_buffer && using_subpass_post_process) { + _post_process_subpass(render_buffer->color, framebuffer, p_render_data); + } + + RD::get_singleton()->draw_command_end_label(); // Render 3D Pass / Render Reflection Probe Pass + + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } else { + RENDER_TIMESTAMP("Render Transparent"); + + framebuffer = render_buffer->color_fbs[FB_CONFIG_ONE_PASS]; + + // this may be needed if we re-introduced steps that change info, not sure which do so in the previous implementation + // _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, false); + + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + RenderListParameters render_list_params(render_list[RENDER_LIST_ALPHA].elements.ptr(), render_list[RENDER_LIST_ALPHA].element_info.ptr(), render_list[RENDER_LIST_ALPHA].elements.size(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, spec_constant_base_flags, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_begin_split(framebuffer, thread_draw_lists.size(), thread_draw_lists.ptr(), can_continue_color ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, can_continue_depth ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, can_continue_color ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, can_continue_depth ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } + + RD::get_singleton()->draw_command_end_label(); // Render Transparent Subpass + } } - RD::get_singleton()->draw_command_end_label(); + if (render_buffer && !using_subpass_post_process) { + RD::get_singleton()->draw_command_begin_label("Post process pass"); - RD::get_singleton()->draw_command_begin_label("Resolve"); + // If we need extra effects we do this in its own pass + RENDER_TIMESTAMP("Tonemap"); - if (render_buffer && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color); + _render_buffers_post_process_and_tonemap(p_render_data); + + RD::get_singleton()->draw_command_end_label(); // Post process pass } - RD::get_singleton()->draw_command_end_label(); + if (render_buffer) { + _disable_clear_request(p_render_data); + } } /* these are being called from RendererSceneRenderRD::_pre_opaque_render */ @@ -642,7 +912,7 @@ void RenderForwardMobile::_render_shadow_append(RID p_framebuffer, const PagedAr _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode, true); uint32_t render_list_size = render_list[RENDER_LIST_SECONDARY].elements.size() - render_list_from; render_list[RENDER_LIST_SECONDARY].sort_by_key_range(render_list_from, render_list_size); - _fill_instance_data(RENDER_LIST_SECONDARY, render_list_from, render_list_size, false); + _fill_element_info(RENDER_LIST_SECONDARY, render_list_from, render_list_size); { //regular forward for now @@ -687,7 +957,7 @@ void RenderForwardMobile::_render_shadow_end(uint32_t p_barrier) { for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; - RenderListParameters render_list_parameters(render_list[RENDER_LIST_SECONDARY].elements.ptr() + shadow_pass.element_from, render_list[RENDER_LIST_SECONDARY].element_info.ptr() + shadow_pass.element_from, shadow_pass.element_count, shadow_pass.flip_cull, shadow_pass.pass_mode, shadow_pass.rp_uniform_set, false, Vector2(), shadow_pass.camera_plane, shadow_pass.lod_distance_multiplier, shadow_pass.screen_lod_threshold, 1, shadow_pass.element_from, RD::BARRIER_MASK_NO_BARRIER); + RenderListParameters render_list_parameters(render_list[RENDER_LIST_SECONDARY].elements.ptr() + shadow_pass.element_from, render_list[RENDER_LIST_SECONDARY].element_info.ptr() + shadow_pass.element_from, shadow_pass.element_count, shadow_pass.flip_cull, shadow_pass.pass_mode, shadow_pass.rp_uniform_set, 0, false, Vector2(), shadow_pass.camera_plane, shadow_pass.lod_distance_multiplier, shadow_pass.screen_lod_threshold, 1, shadow_pass.element_from, RD::BARRIER_MASK_NO_BARRIER); _render_list_with_threads(&render_list_parameters, shadow_pass.framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, shadow_pass.initial_depth_action, shadow_pass.final_depth_action, Vector<Color>(), 1.0, 0, shadow_pass.rect); } @@ -720,14 +990,14 @@ void RenderForwardMobile::_render_material(const Transform3D &p_cam_transform, c PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); RENDER_TIMESTAMP("Render Material"); { - RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, rp_uniform_set); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, rp_uniform_set, 0); //regular forward for now Vector<Color> clear; clear.push_back(Color(0, 0, 0, 0)); @@ -761,14 +1031,14 @@ void RenderForwardMobile::_render_uv2(const PagedArray<GeometryInstance *> &p_in PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); RENDER_TIMESTAMP("Render Material"); { - RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, rp_uniform_set, true); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, rp_uniform_set, true, 0); //regular forward for now Vector<Color> clear; clear.push_back(Color(0, 0, 0, 0)); @@ -834,7 +1104,7 @@ void RenderForwardMobile::_render_particle_collider_heightfield(RID p_fb, const _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); @@ -842,7 +1112,7 @@ void RenderForwardMobile::_render_particle_collider_heightfield(RID p_fb, const { //regular forward for now - RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), false, pass_mode, rp_uniform_set); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), false, pass_mode, rp_uniform_set, 0); _render_list_with_threads(&render_list_params, p_fb, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ); } RD::get_singleton()->draw_command_end_label(); @@ -1052,7 +1322,7 @@ void RenderForwardMobile::_fill_render_list(RenderListType p_render_list, const } uint32_t lightmap_captures_used = 0; - Plane near_plane(p_render_data->cam_transform.origin, -p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z)); + Plane near_plane(-p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z), p_render_data->cam_transform.origin); near_plane.d += p_render_data->cam_projection.get_z_near(); float z_max = p_render_data->cam_projection.get_z_far() - p_render_data->cam_projection.get_z_near(); @@ -1363,19 +1633,10 @@ void RenderForwardMobile::_setup_environment(const RenderDataRD *p_render_data, scene_state.ubo.ssao_ao_affect = environment_get_ssao_ao_affect(p_render_data->environment); scene_state.ubo.ssao_light_affect = environment_get_ssao_light_affect(p_render_data->environment); - Color ao_color = environment_get_ao_color(p_render_data->environment).to_linear(); - scene_state.ubo.ao_color[0] = ao_color.r; - scene_state.ubo.ao_color[1] = ao_color.g; - scene_state.ubo.ao_color[2] = ao_color.b; - scene_state.ubo.ao_color[3] = ao_color.a; - scene_state.ubo.fog_enabled = environment_is_fog_enabled(p_render_data->environment); scene_state.ubo.fog_density = environment_get_fog_density(p_render_data->environment); scene_state.ubo.fog_height = environment_get_fog_height(p_render_data->environment); scene_state.ubo.fog_height_density = environment_get_fog_height_density(p_render_data->environment); - if (scene_state.ubo.fog_height_density >= 0.0001) { - scene_state.ubo.fog_height_density = 1.0 / scene_state.ubo.fog_height_density; - } scene_state.ubo.fog_aerial_perspective = environment_get_fog_aerial_perspective(p_render_data->environment); Color fog_color = environment_get_fog_light_color(p_render_data->environment).to_linear(); @@ -1420,9 +1681,7 @@ void RenderForwardMobile::_setup_environment(const RenderDataRD *p_render_data, RD::get_singleton()->buffer_update(scene_state.uniform_buffers[p_index], 0, sizeof(SceneState::UBO), &scene_state.ubo, RD::BARRIER_MASK_RASTER); } -void RenderForwardMobile::_fill_instance_data(RenderListType p_render_list, uint32_t p_offset, int32_t p_max_elements, bool p_update_buffer) { - // !BAS! Rename this to make clear this is not the same as with the forward renderer and remove p_update_buffer? - +void RenderForwardMobile::_fill_element_info(RenderListType p_render_list, uint32_t p_offset, int32_t p_max_elements) { RenderList *rl = &render_list[p_render_list]; uint32_t element_total = p_max_elements >= 0 ? uint32_t(p_max_elements) : rl->elements.size(); @@ -1487,7 +1746,7 @@ void RenderForwardMobile::_render_list_with_threads(RenderListParameters *p_para } } -void RenderForwardMobile::_fill_push_constant_instance_indices(GeometryInstanceForwardMobile::PushConstant *p_push_constant, const GeometryInstanceForwardMobile *p_instance) { +void RenderForwardMobile::_fill_push_constant_instance_indices(GeometryInstanceForwardMobile::PushConstant *p_push_constant, uint32_t &spec_constants, const GeometryInstanceForwardMobile *p_instance) { // first zero out our indices p_push_constant->omni_lights[0] = 0xFFFF; @@ -1502,6 +1761,19 @@ void RenderForwardMobile::_fill_push_constant_instance_indices(GeometryInstanceF p_push_constant->reflection_probes[0] = 0xFFFF; p_push_constant->reflection_probes[1] = 0xFFFF; + if (p_instance->omni_light_count == 0) { + spec_constants |= 1 << SPEC_CONSTANT_DISABLE_OMNI_LIGHTS; + } + if (p_instance->spot_light_count == 0) { + spec_constants |= 1 << SPEC_CONSTANT_DISABLE_SPOT_LIGHTS; + } + if (p_instance->reflection_probe_count == 0) { + spec_constants |= 1 << SPEC_CONSTANT_DISABLE_REFLECTION_PROBES; + } + if (p_instance->decals_count == 0) { + spec_constants |= 1 << SPEC_CONSTANT_DISABLE_DECALS; + } + for (uint32_t i = 0; i < MAX_RDL_CULL; i++) { uint32_t ofs = i < 4 ? 0 : 1; uint32_t shift = (i & 0x3) << 3; @@ -1549,6 +1821,8 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr const RenderElementInfo &element_info = p_params->element_info[i]; const GeometryInstanceForwardMobile *inst = surf->owner; + uint32_t base_spec_constants = p_params->spec_constant_base_flags; + // GeometryInstanceForwardMobile::PushConstant push_constant = inst->push_constant; GeometryInstanceForwardMobile::PushConstant push_constant; @@ -1584,7 +1858,13 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr mesh_surface = surf->surface_shadow; } else { - _fill_push_constant_instance_indices(&push_constant, inst); + if (inst->use_projector) { + base_spec_constants |= 1 << SPEC_CONSTANT_USING_PROJECTOR; + } + if (inst->use_soft_shadow) { + base_spec_constants |= 1 << SPEC_CONSTANT_USING_SOFT_SHADOWS; + } + _fill_push_constant_instance_indices(&push_constant, base_spec_constants, inst); #ifdef DEBUG_ENABLED if (unlikely(get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_LIGHTING)) { @@ -1676,7 +1956,7 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr prev_index_array_rd = index_array_rd; } - RID pipeline_rd = pipeline->get_render_pipeline(vertex_format, framebuffer_format, p_params->force_wireframe); + RID pipeline_rd = pipeline->get_render_pipeline(vertex_format, framebuffer_format, p_params->force_wireframe, p_params->subpass, base_spec_constants); if (pipeline_rd != prev_pipeline_rd) { // checking with prev shader does not make so much sense, as @@ -1691,8 +1971,8 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr } if (material_uniform_set != prev_material_uniform_set) { - //update uniform set - if (material_uniform_set.is_valid()) { + // Update uniform set. + if (material_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(material_uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_uniform_set, MATERIAL_UNIFORM_SET); } @@ -1935,6 +2215,11 @@ void RenderForwardMobile::geometry_instance_pair_voxel_gi_instances(GeometryInst } void RenderForwardMobile::geometry_instance_set_softshadow_projector_pairing(GeometryInstance *p_geometry_instance, bool p_softshadow, bool p_projector) { + GeometryInstanceForwardMobile *ginstance = static_cast<GeometryInstanceForwardMobile *>(p_geometry_instance); + ERR_FAIL_COND(!ginstance); + + ginstance->use_projector = p_projector; + ginstance->use_soft_shadow = p_softshadow; } void RenderForwardMobile::_geometry_instance_mark_dirty(GeometryInstance *p_geometry_instance) { @@ -2146,7 +2431,7 @@ void RenderForwardMobile::_geometry_instance_update(GeometryInstance *p_geometry } break; #if 0 case RS::INSTANCE_IMMEDIATE: { - RasterizerStorageGLES3::Immediate *immediate = storage->immediate_owner.getornull(inst->base); + RasterizerStorageGLES3::Immediate *immediate = storage->immediate_owner.get_or_null(inst->base); ERR_CONTINUE(!immediate); _add_geometry(immediate, inst, nullptr, -1, p_depth_pass, p_shadow_pass); @@ -2313,12 +2598,12 @@ void RenderForwardMobile::_update_shader_quality_settings() { spec_constants.push_back(sc); sc.type = RD::PIPELINE_SPECIALIZATION_CONSTANT_TYPE_BOOL; - sc.constant_id = SPEC_CONSTANT_DECAL_FILTER; + sc.constant_id = SPEC_CONSTANT_DECAL_USE_MIPMAPS; sc.bool_value = decals_get_filter() == RS::DECAL_FILTER_NEAREST_MIPMAPS || decals_get_filter() == RS::DECAL_FILTER_LINEAR_MIPMAPS || decals_get_filter() == RS::DECAL_FILTER_LINEAR_MIPMAPS_ANISOTROPIC; spec_constants.push_back(sc); - sc.constant_id = SPEC_CONSTANT_PROJECTOR_FILTER; + sc.constant_id = SPEC_CONSTANT_PROJECTOR_USE_MIPMAPS; sc.bool_value = light_projectors_get_filter() == RS::LIGHT_PROJECTOR_FILTER_NEAREST_MIPMAPS || light_projectors_get_filter() == RS::LIGHT_PROJECTOR_FILTER_LINEAR_MIPMAPS || light_projectors_get_filter() == RS::LIGHT_PROJECTOR_FILTER_LINEAR_MIPMAPS_ANISOTROPIC; spec_constants.push_back(sc); @@ -2332,6 +2617,8 @@ RenderForwardMobile::RenderForwardMobile(RendererStorageRD *p_storage) : RendererSceneRenderRD(p_storage) { singleton = this; + sky.set_texture_format(_render_buffers_get_color_format()); + String defines; defines += "\n#define MAX_ROUGHNESS_LOD " + itos(get_roughness_layers() - 1) + ".0\n"; @@ -2339,7 +2626,7 @@ RenderForwardMobile::RenderForwardMobile(RendererStorageRD *p_storage) : defines += "\n#define USE_RADIANCE_CUBEMAP_ARRAY \n"; } // defines += "\n#define SDFGI_OCT_SIZE " + itos(gi.sdfgi_get_lightprobe_octahedron_size()) + "\n"; - defines += "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(get_max_directional_lights()) + "\n"; + defines += "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(MAX_DIRECTIONAL_LIGHTS) + "\n"; { //lightmaps diff --git a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h index 973925d562..36a92e1464 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h @@ -65,12 +65,27 @@ protected: }; enum { - SPEC_CONSTANT_SOFT_SHADOW_SAMPLES = 6, - SPEC_CONSTANT_PENUMBRA_SHADOW_SAMPLES = 7, - SPEC_CONSTANT_DIRECTIONAL_SOFT_SHADOW_SAMPLES = 8, - SPEC_CONSTANT_DIRECTIONAL_PENUMBRA_SHADOW_SAMPLES = 9, - SPEC_CONSTANT_DECAL_FILTER = 10, - SPEC_CONSTANT_PROJECTOR_FILTER = 11, + + SPEC_CONSTANT_USING_PROJECTOR = 0, + SPEC_CONSTANT_USING_SOFT_SHADOWS = 1, + SPEC_CONSTANT_USING_DIRECTIONAL_SOFT_SHADOWS = 2, + + SPEC_CONSTANT_SOFT_SHADOW_SAMPLES = 3, + SPEC_CONSTANT_PENUMBRA_SHADOW_SAMPLES = 4, + SPEC_CONSTANT_DIRECTIONAL_SOFT_SHADOW_SAMPLES = 5, + SPEC_CONSTANT_DIRECTIONAL_PENUMBRA_SHADOW_SAMPLES = 6, + + SPEC_CONSTANT_DECAL_USE_MIPMAPS = 7, + SPEC_CONSTANT_PROJECTOR_USE_MIPMAPS = 8, + + SPEC_CONSTANT_DISABLE_OMNI_LIGHTS = 9, + SPEC_CONSTANT_DISABLE_SPOT_LIGHTS = 10, + SPEC_CONSTANT_DISABLE_REFLECTION_PROBES = 11, + SPEC_CONSTANT_DISABLE_DIRECTIONAL_LIGHTS = 12, + + SPEC_CONSTANT_DISABLE_DECALS = 13, + SPEC_CONSTANT_DISABLE_FOG = 14, + }; enum { @@ -92,6 +107,18 @@ protected: /* Render Buffer */ + // We can have: + // - 4 subpasses combining the full render cycle + // - 3 subpasses + 1 normal pass for tonemapping/glow/dof/etc (using fb for 2D buffer) + // - 2 subpasses + 1 normal pass for transparent + 1 normal pass for tonemapping/glow/dof/etc (using fb for 2D buffer) + enum RenderBufferMobileFramebufferConfigType { + FB_CONFIG_ONE_PASS, // Single pass frame buffer for alpha pass + FB_CONFIG_TWO_SUBPASSES, // Opaque + Sky sub pass + FB_CONFIG_THREE_SUBPASSES, // Opaque + Sky + Alpha sub pass + FB_CONFIG_FOUR_SUBPASSES, // Opaque + Sky + Alpha sub pass + Tonemap pass + FB_CONFIG_MAX + }; + struct RenderBufferDataForwardMobile : public RenderBufferData { RID color; RID depth; @@ -104,12 +131,12 @@ protected: RID depth_msaa; // RID normal_roughness_buffer_msaa; - RID color_fb; + RID color_fbs[FB_CONFIG_MAX]; int width, height; uint32_t view_count; void clear(); - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); ~RenderBufferDataForwardMobile(); }; @@ -147,13 +174,15 @@ protected: bool force_wireframe = false; Vector2 uv_offset; Plane lod_plane; + uint32_t spec_constant_base_flags = 0; float lod_distance_multiplier = 0.0; float screen_lod_threshold = 0.0; RD::FramebufferFormatID framebuffer_format = 0; uint32_t element_offset = 0; uint32_t barrier = RD::BARRIER_MASK_ALL; + uint32_t subpass = 0; - RenderListParameters(GeometryInstanceSurfaceDataCache **p_elements, RenderElementInfo *p_element_info, int p_element_count, bool p_reverse_cull, PassMode p_pass_mode, RID p_render_pass_uniform_set, bool p_force_wireframe = false, const Vector2 &p_uv_offset = Vector2(), const Plane &p_lod_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, uint32_t p_view_count = 1, uint32_t p_element_offset = 0, uint32_t p_barrier = RD::BARRIER_MASK_ALL) { + RenderListParameters(GeometryInstanceSurfaceDataCache **p_elements, RenderElementInfo *p_element_info, int p_element_count, bool p_reverse_cull, PassMode p_pass_mode, RID p_render_pass_uniform_set, uint32_t p_spec_constant_base_flags = 0, bool p_force_wireframe = false, const Vector2 &p_uv_offset = Vector2(), const Plane &p_lod_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, uint32_t p_view_count = 1, uint32_t p_element_offset = 0, uint32_t p_barrier = RD::BARRIER_MASK_ALL) { elements = p_elements; element_info = p_element_info; element_count = p_element_count; @@ -169,9 +198,11 @@ protected: screen_lod_threshold = p_screen_lod_threshold; element_offset = p_element_offset; barrier = p_barrier; + spec_constant_base_flags = p_spec_constant_base_flags; } }; + virtual float _render_buffers_get_luminance_multiplier() override; virtual RD::DataFormat _render_buffers_get_color_format() override; virtual bool _render_buffers_can_be_storage() override; @@ -195,7 +226,7 @@ protected: virtual RID _render_buffers_get_normal_texture(RID p_render_buffers) override; void _fill_render_list(RenderListType p_render_list, const RenderDataRD *p_render_data, PassMode p_pass_mode, bool p_append = false); - void _fill_instance_data(RenderListType p_render_list, uint32_t p_offset = 0, int32_t p_max_elements = -1, bool p_update_buffer = true); + void _fill_element_info(RenderListType p_render_list, uint32_t p_offset = 0, int32_t p_max_elements = -1); // void _update_instance_data_buffer(RenderListType p_render_list); static RenderForwardMobile *singleton; @@ -263,8 +294,6 @@ protected: float roughness_limiter_limit; uint32_t roughness_limiter_pad[2]; - float ao_color[4]; - // Fog uint32_t fog_enabled; float fog_density; @@ -513,6 +542,8 @@ protected: RID transforms_uniform_set; float depth = 0; bool mirror = false; + bool use_projector = false; + bool use_soft_shadow = false; Transform3D transform; bool store_transform_cache = true; // if true we copy our transform into our PushConstant, if false we use our transforms UBO and clear our PushConstants transform bool non_uniform_scale = false; @@ -571,11 +602,13 @@ protected: dirty_list_element(this) {} }; - _FORCE_INLINE_ void _fill_push_constant_instance_indices(GeometryInstanceForwardMobile::PushConstant *p_push_constant, const GeometryInstanceForwardMobile *p_instance); + _FORCE_INLINE_ void _fill_push_constant_instance_indices(GeometryInstanceForwardMobile::PushConstant *p_push_constant, uint32_t &spec_constants, const GeometryInstanceForwardMobile *p_instance); void _update_shader_quality_settings() override; public: + virtual RID reflection_probe_create_framebuffer(RID p_color, RID p_depth) override; + static void _geometry_instance_dependency_changed(RendererStorage::DependencyChangedNotification p_notification, RendererStorage::DependencyTracker *p_tracker); static void _geometry_instance_dependency_deleted(const RID &p_dependency, RendererStorage::DependencyTracker *p_tracker); diff --git a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp index bcdcb05653..16d650a540 100644 --- a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp @@ -135,8 +135,7 @@ void SceneShaderForwardMobile::ShaderData::set_code(const String &p_code) { SceneShaderForwardMobile *shader_singleton = (SceneShaderForwardMobile *)SceneShaderForwardMobile::singleton; Error err = shader_singleton->compiler.compile(RS::SHADER_SPATIAL, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); + ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); if (version.is_null()) { version = shader_singleton->shader.version_create(); @@ -337,36 +336,36 @@ void SceneShaderForwardMobile::ShaderData::set_default_texture_param(const Strin void SceneShaderForwardMobile::ShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { Map<int, StringName> order; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { continue; } - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); + if (E.value.texture_order >= 0) { + order[E.value.texture_order + 100000] = E.key; } else { - order[E->get().order] = E->key(); + order[E.value.order] = E.key; } } - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); + for (const KeyValue<int, StringName> &E : order) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E.value]); + pi.name = E.value; p_param_list->push_back(pi); } } void SceneShaderForwardMobile::ShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p.info = ShaderLanguage::uniform_to_property_info(E.value); + p.info.name = E.key; //supply name + p.index = E.value.instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E.value.default_value, E.value.type, E.value.array_size, E.value.hint); p_param_list->push_back(p); } } @@ -391,7 +390,7 @@ Variant SceneShaderForwardMobile::ShaderData::get_default_parameter(const String if (uniforms.has(p_parameter)) { ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.array_size, uniform.hint); } return Variant(); } @@ -594,10 +593,10 @@ void SceneShaderForwardMobile::init(RendererStorageRD *p_storage, const String p actions.usage_defines["UV2"] = "#define UV2_USED\n"; actions.usage_defines["BONE_INDICES"] = "#define BONES_USED\n"; actions.usage_defines["BONE_WEIGHTS"] = "#define WEIGHTS_USED\n"; - actions.usage_defines["CUSTOM0"] = "#define CUSTOM0\n"; - actions.usage_defines["CUSTOM1"] = "#define CUSTOM1\n"; - actions.usage_defines["CUSTOM2"] = "#define CUSTOM2\n"; - actions.usage_defines["CUSTOM3"] = "#define CUSTOM3\n"; + actions.usage_defines["CUSTOM0"] = "#define CUSTOM0_USED\n"; + actions.usage_defines["CUSTOM1"] = "#define CUSTOM1_USED\n"; + actions.usage_defines["CUSTOM2"] = "#define CUSTOM2_USED\n"; + actions.usage_defines["CUSTOM3"] = "#define CUSTOM3_USED\n"; actions.usage_defines["NORMAL_MAP"] = "#define NORMAL_MAP_USED\n"; actions.usage_defines["NORMAL_MAP_DEPTH"] = "@NORMAL_MAP"; actions.usage_defines["COLOR"] = "#define COLOR_USED\n"; @@ -666,6 +665,8 @@ void SceneShaderForwardMobile::init(RendererStorageRD *p_storage, const String p actions.global_buffer_array_variable = "global_variables.data"; actions.instance_uniform_index_variable = "draw_call.instance_uniforms_ofs"; + actions.apply_luminance_multiplier = true; // apply luminance multiplier to screen texture + compiler.initialize(actions); } @@ -674,6 +675,8 @@ void SceneShaderForwardMobile::init(RendererStorageRD *p_storage, const String p default_shader = storage->shader_allocate(); storage->shader_initialize(default_shader); storage->shader_set_code(default_shader, R"( +// Default 3D material shader (mobile). + shader_type spatial; void vertex() { @@ -702,6 +705,8 @@ void fragment() { storage->shader_initialize(overdraw_material_shader); // Use relatively low opacity so that more "layers" of overlapping objects can be distinguished. storage->shader_set_code(overdraw_material_shader, R"( +// 3D editor Overdraw debug draw mode shader (mobile). + shader_type spatial; render_mode blend_add, unshaded; diff --git a/servers/rendering/renderer_rd/pipeline_cache_rd.cpp b/servers/rendering/renderer_rd/pipeline_cache_rd.cpp index aefe926cb0..0d60052666 100644 --- a/servers/rendering/renderer_rd/pipeline_cache_rd.cpp +++ b/servers/rendering/renderer_rd/pipeline_cache_rd.cpp @@ -35,8 +35,10 @@ RID PipelineCacheRD::_generate_version(RD::VertexFormatID p_vertex_format_id, RD RD::PipelineMultisampleState multisample_state_version = multisample_state; multisample_state_version.sample_count = RD::get_singleton()->framebuffer_format_get_texture_samples(p_framebuffer_format_id, p_render_pass); + bool wireframe = p_wireframe || rasterization_state.wireframe; + RD::PipelineRasterizationState raster_state_version = rasterization_state; - raster_state_version.wireframe = p_wireframe; + raster_state_version.wireframe = wireframe; Vector<RD::PipelineSpecializationConstant> specialization_constants = base_specialization_constants; @@ -59,7 +61,7 @@ RID PipelineCacheRD::_generate_version(RD::VertexFormatID p_vertex_format_id, RD versions = (Version *)memrealloc(versions, sizeof(Version) * (version_count + 1)); versions[version_count].framebuffer_id = p_framebuffer_format_id; versions[version_count].vertex_id = p_vertex_format_id; - versions[version_count].wireframe = p_wireframe; + versions[version_count].wireframe = wireframe; versions[version_count].pipeline = pipeline; versions[version_count].render_pass = p_render_pass; versions[version_count].bool_specializations = p_bool_specializations; diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index 6a66e9fa01..c69c9eeadf 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -480,6 +480,10 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend case Item::Command::TYPE_RECT: { const Item::CommandRect *rect = static_cast<const Item::CommandRect *>(c); + if (rect->flags & CANVAS_RECT_TILE) { + current_repeat = RenderingServer::CanvasItemTextureRepeat::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED; + } + //bind pipeline { RID pipeline = pipeline_variants->variants[light_mode][PIPELINE_VARIANT_QUAD].get_render_pipeline(RD::INVALID_ID, p_framebuffer_format); @@ -537,6 +541,14 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend src_rect = Rect2(0, 0, 1, 1); } + if (rect->flags & CANVAS_RECT_MSDF) { + push_constant.flags |= FLAGS_USE_MSDF; + push_constant.msdf[0] = rect->px_range; // Pixel range. + push_constant.msdf[1] = rect->outline; // Outline size. + push_constant.msdf[2] = 0.f; // Reserved. + push_constant.msdf[3] = 0.f; // Reserved. + } + push_constant.modulation[0] = rect->modulate.r * base_color.r; push_constant.modulation[1] = rect->modulate.g * base_color.g; push_constant.modulation[2] = rect->modulate.b * base_color.b; @@ -620,7 +632,7 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend RD::get_singleton()->draw_list_bind_index_array(p_draw_list, shader.quad_index_array); RD::get_singleton()->draw_list_draw(p_draw_list, true); - //restore if overrided + // Restore if overridden. push_constant.color_texture_pixel_size[0] = texpixel_size.x; push_constant.color_texture_pixel_size[1] = texpixel_size.y; @@ -1074,7 +1086,7 @@ void RendererCanvasRenderRD::_render_items(RID p_to_render_target, int p_item_co } } - RID material = ci->material; + RID material = ci->material_owner == nullptr ? ci->material : ci->material_owner->material; if (material.is_null() && ci->canvas_group != nullptr) { material = default_canvas_group_material; @@ -1089,7 +1101,8 @@ void RendererCanvasRenderRD::_render_items(RID p_to_render_target, int p_item_co if (material_data) { if (material_data->shader_data->version.is_valid() && material_data->shader_data->valid) { pipeline_variants = &material_data->shader_data->pipeline_variants; - if (material_data->uniform_set.is_valid()) { + // Update uniform set. + if (material_data->uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(material_data->uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_data->uniform_set, MATERIAL_UNIFORM_SET); } } else { @@ -1131,7 +1144,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p continue; } - CanvasLight *clight = canvas_light_owner.getornull(l->light_internal); + CanvasLight *clight = canvas_light_owner.get_or_null(l->light_internal); if (!clight) { //unused or invalid texture l->render_index_cache = -1; l = l->next_ptr; @@ -1194,7 +1207,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p continue; } - CanvasLight *clight = canvas_light_owner.getornull(l->light_internal); + CanvasLight *clight = canvas_light_owner.get_or_null(l->light_internal); if (!clight) { //unused or invalid texture l->render_index_cache = -1; l = l->next_ptr; @@ -1341,8 +1354,10 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p } } - if (ci->material.is_valid()) { - MaterialData *md = (MaterialData *)storage->material_get_data(ci->material, RendererStorageRD::SHADER_TYPE_2D); + RID material = ci->material_owner == nullptr ? ci->material : ci->material_owner->material; + + if (material.is_valid()) { + MaterialData *md = (MaterialData *)storage->material_get_data(material, RendererStorageRD::SHADER_TYPE_2D); if (md && md->shader_data->valid) { if (md->shader_data->uses_screen_texture && canvas_group_owner == nullptr) { if (!material_screen_texture_found) { @@ -1362,7 +1377,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p if (!RD::get_singleton()->uniform_set_is_valid(md->uniform_set)) { // uniform set may be gone because a dependency was erased. In this case, it will happen // if a texture is deleted, so just re-create it. - storage->material_force_update_textures(ci->material, RendererStorageRD::SHADER_TYPE_2D); + storage->material_force_update_textures(material, RendererStorageRD::SHADER_TYPE_2D); } } } @@ -1379,6 +1394,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p update_skeletons = true; } } + c = c->next; } } @@ -1465,7 +1481,7 @@ RID RendererCanvasRenderRD::light_create() { } void RendererCanvasRenderRD::light_set_texture(RID p_rid, RID p_texture) { - CanvasLight *cl = canvas_light_owner.getornull(p_rid); + CanvasLight *cl = canvas_light_owner.get_or_null(p_rid); ERR_FAIL_COND(!cl); if (cl->texture == p_texture) { return; @@ -1481,7 +1497,7 @@ void RendererCanvasRenderRD::light_set_texture(RID p_rid, RID p_texture) { } void RendererCanvasRenderRD::light_set_use_shadow(RID p_rid, bool p_enable) { - CanvasLight *cl = canvas_light_owner.getornull(p_rid); + CanvasLight *cl = canvas_light_owner.get_or_null(p_rid); ERR_FAIL_COND(!cl); cl->shadow.enabled = p_enable; @@ -1521,7 +1537,7 @@ void RendererCanvasRenderRD::_update_shadow_atlas() { } } void RendererCanvasRenderRD::light_update_shadow(RID p_rid, int p_shadow_index, const Transform2D &p_light_xform, int p_light_mask, float p_near, float p_far, LightOccluderInstance *p_occluders) { - CanvasLight *cl = canvas_light_owner.getornull(p_rid); + CanvasLight *cl = canvas_light_owner.get_or_null(p_rid); ERR_FAIL_COND(!cl->shadow.enabled); _update_shadow_atlas(); @@ -1575,7 +1591,7 @@ void RendererCanvasRenderRD::light_update_shadow(RID p_rid, int p_shadow_index, LightOccluderInstance *instance = p_occluders; while (instance) { - OccluderPolygon *co = occluder_polygon_owner.getornull(instance->occluder); + OccluderPolygon *co = occluder_polygon_owner.get_or_null(instance->occluder); if (!co || co->index_array.is_null() || !(p_light_mask & instance->light_mask)) { instance = instance->next; @@ -1599,14 +1615,14 @@ void RendererCanvasRenderRD::light_update_shadow(RID p_rid, int p_shadow_index, } void RendererCanvasRenderRD::light_update_directional_shadow(RID p_rid, int p_shadow_index, const Transform2D &p_light_xform, int p_light_mask, float p_cull_distance, const Rect2 &p_clip_rect, LightOccluderInstance *p_occluders) { - CanvasLight *cl = canvas_light_owner.getornull(p_rid); + CanvasLight *cl = canvas_light_owner.get_or_null(p_rid); ERR_FAIL_COND(!cl->shadow.enabled); _update_shadow_atlas(); Vector2 light_dir = p_light_xform.elements[1].normalized(); - Vector2 center = p_clip_rect.position + p_clip_rect.size * 0.5; + Vector2 center = p_clip_rect.get_center(); float to_edge_distance = ABS(light_dir.dot(p_clip_rect.get_support(light_dir)) - light_dir.dot(center)); @@ -1650,7 +1666,7 @@ void RendererCanvasRenderRD::light_update_directional_shadow(RID p_rid, int p_sh LightOccluderInstance *instance = p_occluders; while (instance) { - OccluderPolygon *co = occluder_polygon_owner.getornull(instance->occluder); + OccluderPolygon *co = occluder_polygon_owner.get_or_null(instance->occluder); if (!co || co->index_array.is_null() || !(p_light_mask & instance->light_mask)) { instance = instance->next; @@ -1716,7 +1732,7 @@ void RendererCanvasRenderRD::render_sdf(RID p_render_target, LightOccluderInstan LightOccluderInstance *instance = p_occluders; while (instance) { - OccluderPolygon *co = occluder_polygon_owner.getornull(instance->occluder); + OccluderPolygon *co = occluder_polygon_owner.get_or_null(instance->occluder); if (!co || co->sdf_index_array.is_null()) { instance = instance->next; @@ -1750,7 +1766,7 @@ RID RendererCanvasRenderRD::occluder_polygon_create() { } void RendererCanvasRenderRD::occluder_polygon_set_shape(RID p_occluder, const Vector<Vector2> &p_points, bool p_closed) { - OccluderPolygon *oc = occluder_polygon_owner.getornull(p_occluder); + OccluderPolygon *oc = occluder_polygon_owner.get_or_null(p_occluder); ERR_FAIL_COND(!oc); Vector<Vector2> lines; @@ -1919,7 +1935,7 @@ void RendererCanvasRenderRD::occluder_polygon_set_shape(RID p_occluder, const Ve } void RendererCanvasRenderRD::occluder_polygon_set_cull_mode(RID p_occluder, RS::CanvasOccluderPolygonCullMode p_mode) { - OccluderPolygon *oc = occluder_polygon_owner.getornull(p_occluder); + OccluderPolygon *oc = occluder_polygon_owner.get_or_null(p_occluder); ERR_FAIL_COND(!oc); oc->cull_mode = p_mode; } @@ -1965,8 +1981,7 @@ void RendererCanvasRenderRD::ShaderData::set_code(const String &p_code) { RendererCanvasRenderRD *canvas_singleton = (RendererCanvasRenderRD *)RendererCanvasRender::singleton; Error err = canvas_singleton->shader.compiler.compile(RS::SHADER_CANVAS_ITEM, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); + ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); if (version.is_null()) { version = canvas_singleton->shader.canvas_shader.version_create(); @@ -2118,35 +2133,35 @@ void RendererCanvasRenderRD::ShaderData::set_default_texture_param(const StringN void RendererCanvasRenderRD::ShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { Map<int, StringName> order; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { continue; } - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); + if (E.value.texture_order >= 0) { + order[E.value.texture_order + 100000] = E.key; } else { - order[E->get().order] = E->key(); + order[E.value.order] = E.key; } } - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); + for (const KeyValue<int, StringName> &E : order) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E.value]); + pi.name = E.value; p_param_list->push_back(pi); } } void RendererCanvasRenderRD::ShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p.info = ShaderLanguage::uniform_to_property_info(E.value); + p.info.name = E.key; //supply name + p.index = E.value.instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E.value.default_value, E.value.type, E.value.array_size, E.value.hint); p_param_list->push_back(p); } } @@ -2171,7 +2186,7 @@ Variant RendererCanvasRenderRD::ShaderData::get_default_parameter(const StringNa if (uniforms.has(p_parameter)) { ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.array_size, uniform.hint); } return Variant(); } @@ -2571,6 +2586,8 @@ RendererCanvasRenderRD::RendererCanvasRenderRD(RendererStorageRD *p_storage) { storage->shader_initialize(default_canvas_group_shader); storage->shader_set_code(default_canvas_group_shader, R"( +// Default CanvasGroup shader. + shader_type canvas_item; void fragment() { @@ -2594,7 +2611,7 @@ void fragment() { bool RendererCanvasRenderRD::free(RID p_rid) { if (canvas_light_owner.owns(p_rid)) { - CanvasLight *cl = canvas_light_owner.getornull(p_rid); + CanvasLight *cl = canvas_light_owner.get_or_null(p_rid); ERR_FAIL_COND_V(!cl, false); light_set_use_shadow(p_rid, false); canvas_light_owner.free(p_rid); diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h index 7c4f62832c..ec7d7e2854 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h @@ -84,8 +84,9 @@ class RendererCanvasRenderRD : public RendererCanvasRender { FLAGS_LIGHT_COUNT_SHIFT = 20, FLAGS_DEFAULT_NORMAL_MAP_USED = (1 << 26), - FLAGS_DEFAULT_SPECULAR_MAP_USED = (1 << 27) + FLAGS_DEFAULT_SPECULAR_MAP_USED = (1 << 27), + FLAGS_USE_MSDF = (1 << 28), }; enum { @@ -388,7 +389,10 @@ class RendererCanvasRenderRD : public RendererCanvasRender { //rect struct { float modulation[4]; - float ninepatch_margins[4]; + union { + float msdf[4]; + float ninepatch_margins[4]; + }; float dst_rect[4]; float src_rect[4]; float pad[2]; diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp index e6ae66d56f..559e6d5ad7 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp @@ -46,6 +46,8 @@ void RendererCompositorRD::blit_render_targets_to_screen(DisplayServer::WindowID RID rd_texture = storage->texture_get_rd_texture(texture); ERR_CONTINUE(rd_texture.is_null()); + // TODO if keep_3d_linear was set when rendering to this render target we need to add a linear->sRGB conversion in. + if (!render_target_descriptors.has(rd_texture) || !RD::get_singleton()->uniform_set_is_valid(render_target_descriptors[rd_texture])) { Vector<RD::Uniform> uniforms; RD::Uniform u; @@ -65,10 +67,14 @@ void RendererCompositorRD::blit_render_targets_to_screen(DisplayServer::WindowID RD::get_singleton()->draw_list_bind_index_array(draw_list, blit.array); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, render_target_descriptors[rd_texture], 0); - blit.push_constant.rect[0] = p_render_targets[i].rect.position.x / screen_size.width; - blit.push_constant.rect[1] = p_render_targets[i].rect.position.y / screen_size.height; - blit.push_constant.rect[2] = p_render_targets[i].rect.size.width / screen_size.width; - blit.push_constant.rect[3] = p_render_targets[i].rect.size.height / screen_size.height; + blit.push_constant.src_rect[0] = p_render_targets[i].src_rect.position.x; + blit.push_constant.src_rect[1] = p_render_targets[i].src_rect.position.y; + blit.push_constant.src_rect[2] = p_render_targets[i].src_rect.size.width; + blit.push_constant.src_rect[3] = p_render_targets[i].src_rect.size.height; + blit.push_constant.dst_rect[0] = p_render_targets[i].dst_rect.position.x / screen_size.width; + blit.push_constant.dst_rect[1] = p_render_targets[i].dst_rect.position.y / screen_size.height; + blit.push_constant.dst_rect[2] = p_render_targets[i].dst_rect.size.width / screen_size.width; + blit.push_constant.dst_rect[3] = p_render_targets[i].dst_rect.size.height / screen_size.height; blit.push_constant.layer = p_render_targets[i].multi_view.layer; blit.push_constant.eye_center[0] = p_render_targets[i].lens_distortion.eye_center.x; blit.push_constant.eye_center[1] = p_render_targets[i].lens_distortion.eye_center.y; @@ -191,7 +197,7 @@ void RendererCompositorRD::set_boot_image(const Ref<Image> &p_image, const Color } } else { screenrect = imgrect; - screenrect.position += ((Size2(window_size.width, window_size.height) - screenrect.size) / 2.0).floor(); + screenrect.position += ((window_size - screenrect.size) / 2.0).floor(); } screenrect.position /= window_size; @@ -203,10 +209,14 @@ void RendererCompositorRD::set_boot_image(const Ref<Image> &p_image, const Color RD::get_singleton()->draw_list_bind_index_array(draw_list, blit.array); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, uset, 0); - blit.push_constant.rect[0] = screenrect.position.x; - blit.push_constant.rect[1] = screenrect.position.y; - blit.push_constant.rect[2] = screenrect.size.width; - blit.push_constant.rect[3] = screenrect.size.height; + blit.push_constant.src_rect[0] = 0.0; + blit.push_constant.src_rect[1] = 0.0; + blit.push_constant.src_rect[2] = 1.0; + blit.push_constant.src_rect[3] = 1.0; + blit.push_constant.dst_rect[0] = screenrect.position.x; + blit.push_constant.dst_rect[1] = screenrect.position.y; + blit.push_constant.dst_rect[2] = screenrect.size.width; + blit.push_constant.dst_rect[3] = screenrect.size.height; blit.push_constant.layer = 0; blit.push_constant.eye_center[0] = 0; blit.push_constant.eye_center[1] = 0; @@ -222,7 +232,7 @@ void RendererCompositorRD::set_boot_image(const Ref<Image> &p_image, const Color RD::get_singleton()->swap_buffers(); - RD::get_singleton()->free(texture); + storage->free(texture); } RendererCompositorRD *RendererCompositorRD::singleton = nullptr; @@ -280,6 +290,11 @@ RendererCompositorRD::RendererCompositorRD() { // default to our high end renderer scene = memnew(RendererSceneRenderImplementation::RenderForwardClustered(storage)); } + + scene->init(); + + // now we're ready to create our effects, + storage->init_effects(!scene->_render_buffers_can_be_storage()); } RendererCompositorRD::~RendererCompositorRD() { diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.h b/servers/rendering/renderer_rd/renderer_compositor_rd.h index 15b3b77ed9..0230c46800 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.h +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.h @@ -55,7 +55,8 @@ protected: }; struct BlitPushConstant { - float rect[4]; + float src_rect[4]; + float dst_rect[4]; float eye_center[2]; float k1; @@ -80,7 +81,7 @@ protected: Map<RID, RID> render_target_descriptors; double time; - float delta; + double delta; static uint64_t frame; @@ -100,7 +101,7 @@ public: void finalize(); _ALWAYS_INLINE_ uint64_t get_frame_number() const { return frame; } - _ALWAYS_INLINE_ float get_frame_delta_time() const { return delta; } + _ALWAYS_INLINE_ double get_frame_delta_time() const { return delta; } _ALWAYS_INLINE_ double get_total_time() const { return time; } static Error is_viable() { diff --git a/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp index d631cb4bac..bb7fbbcdc2 100644 --- a/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp @@ -32,13 +32,12 @@ uint64_t RendererSceneEnvironmentRD::auto_exposure_counter = 2; -void RendererSceneEnvironmentRD::set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color) { +void RendererSceneEnvironmentRD::set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source) { ambient_light = p_color; ambient_source = p_ambient; ambient_light_energy = p_energy; ambient_sky_contribution = p_sky_contribution; reflection_source = p_reflection_source; - ao_color = p_ao_color; } void RendererSceneEnvironmentRD::set_tonemap(RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) { diff --git a/servers/rendering/renderer_rd/renderer_scene_environment_rd.h b/servers/rendering/renderer_rd/renderer_scene_environment_rd.h index 992c4bf471..bc47abbff5 100644 --- a/servers/rendering/renderer_rd/renderer_scene_environment_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_environment_rd.h @@ -52,7 +52,6 @@ public: float ambient_light_energy = 1.0; float ambient_sky_contribution = 1.0; RS::EnvironmentReflectionSource reflection_source = RS::ENV_REFLECTION_SOURCE_BG; - Color ao_color; /// Tonemap @@ -142,7 +141,7 @@ public: bool use_1d_color_correction = false; RID color_correction = RID(); - void set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color); + void set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source); void set_tonemap(RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale); void set_glow(bool p_enable, Vector<float> p_levels, float p_intensity, float p_strength, float p_mix, float p_bloom_threshold, RS::EnvironmentGlowBlendMode p_blend_mode, float p_hdr_bleed_threshold, float p_hdr_bleed_scale, float p_hdr_luminance_cap); void set_sdfgi(bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias); diff --git a/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp index 98d08f68e8..ea73dbf26e 100644 --- a/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp @@ -1450,7 +1450,7 @@ void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform3D &p_transform, Re break; } - RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_scene_render->render_state.sdfgi_update_data->directional_lights->get(j)); + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.get_or_null(p_scene_render->render_state.sdfgi_update_data->directional_lights->get(j)); ERR_CONTINUE(!li); if (storage->light_directional_is_sky_only(li->light)) { @@ -1469,7 +1469,7 @@ void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform3D &p_transform, Re lights[idx].color[1] = color.g; lights[idx].color[2] = color.b; lights[idx].type = RS::LIGHT_DIRECTIONAL; - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY) * storage->light_get_param(li->light, RS::LIGHT_PARAM_INDIRECT_ENERGY); lights[idx].has_shadow = storage->light_has_shadow(li->light); idx++; @@ -1484,7 +1484,7 @@ void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform3D &p_transform, Re break; } - RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_scene_render->render_state.sdfgi_update_data->positional_light_instances[j]); + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.get_or_null(p_scene_render->render_state.sdfgi_update_data->positional_light_instances[j]); ERR_CONTINUE(!li); uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); @@ -1514,7 +1514,7 @@ void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform3D &p_transform, Re lights[idx].color[1] = color.g; lights[idx].color[2] = color.b; lights[idx].type = storage->light_get_type(li->light); - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY) * storage->light_get_param(li->light, RS::LIGHT_PARAM_INDIRECT_ENERGY); lights[idx].has_shadow = storage->light_has_shadow(li->light); lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); @@ -1534,7 +1534,7 @@ void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform3D &p_transform, Re void RendererSceneGIRD::SDFGI::render_region(RID p_render_buffers, int p_region, const PagedArray<RendererSceneRender::GeometryInstance *> &p_instances, RendererSceneRenderRD *p_scene_render) { //print_line("rendering region " + itos(p_region)); - RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); // we wouldn't be here if this failed but... AABB bounds; Vector3i from; @@ -1892,7 +1892,7 @@ void RendererSceneGIRD::SDFGI::render_region(RID p_render_buffers, int p_region, } void RendererSceneGIRD::SDFGI::render_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_light_cull_result, RendererSceneRenderRD *p_scene_render) { - RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); // we wouldn't be here if this failed but... RD::get_singleton()->draw_command_begin_label("SDFGI Render Static Lighs"); @@ -1921,7 +1921,7 @@ void RendererSceneGIRD::SDFGI::render_static_lights(RID p_render_buffers, uint32 break; } - RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_positional_light_cull_result[i][j]); + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.get_or_null(p_positional_light_cull_result[i][j]); ERR_CONTINUE(!li); uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); @@ -1953,7 +1953,7 @@ void RendererSceneGIRD::SDFGI::render_static_lights(RID p_render_buffers, uint32 lights[idx].color[0] = color.r; lights[idx].color[1] = color.g; lights[idx].color[2] = color.b; - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY) * storage->light_get_param(li->light, RS::LIGHT_PARAM_INDIRECT_ENERGY); lights[idx].has_shadow = storage->light_has_shadow(li->light); lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); @@ -2575,7 +2575,7 @@ void RendererSceneGIRD::VoxelGIInstance::update(bool p_update_light_instances, c Vector3 render_dir = render_z[j]; Vector3 up_dir = render_up[j]; - Vector3 center = aabb.position + aabb.size * 0.5; + Vector3 center = aabb.get_center(); Transform3D xform; xform.set_look_at(center - aabb.size * 0.5 * render_dir, center, up_dir); @@ -2812,8 +2812,6 @@ void RendererSceneGIRD::init(RendererStorageRD *p_storage, RendererSceneSkyRD *p { //kinda complicated to compute the amount of slots, we try to use as many as we can - voxel_gi_max_lights = 32; - voxel_gi_lights = memnew_arr(VoxelGILight, voxel_gi_max_lights); voxel_gi_lights_uniform = RD::get_singleton()->uniform_buffer_create(voxel_gi_max_lights * sizeof(VoxelGILight)); voxel_gi_quality = RS::VoxelGIQuality(CLAMP(int(GLOBAL_GET("rendering/global_illumination/voxel_gi/quality")), 0, 1)); @@ -3009,7 +3007,9 @@ void RendererSceneGIRD::free() { sdfgi_shader.integrate.version_free(sdfgi_shader.integrate_shader); sdfgi_shader.preprocess.version_free(sdfgi_shader.preprocess_shader); - memdelete_arr(voxel_gi_lights); + if (voxel_gi_lights) { + memdelete_arr(voxel_gi_lights); + } } RendererSceneGIRD::SDFGI *RendererSceneGIRD::create_sdfgi(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size) { @@ -3024,7 +3024,7 @@ void RendererSceneGIRD::setup_voxel_gi_instances(RID p_render_buffers, const Tra r_voxel_gi_instances_used = 0; // feels a little dirty to use our container this way but.... - RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(rb == nullptr); RID voxel_gi_buffer = p_scene_render->render_buffers_get_voxel_gi_buffer(p_render_buffers); @@ -3119,9 +3119,8 @@ void RendererSceneGIRD::setup_voxel_gi_instances(RID p_render_buffers, const Tra void RendererSceneGIRD::process_gi(RID p_render_buffers, RID p_normal_roughness_buffer, RID p_voxel_gi_buffer, RID p_environment, const CameraMatrix &p_projection, const Transform3D &p_transform, const PagedArray<RID> &p_voxel_gi_instances, RendererSceneRenderRD *p_scene_render) { RD::get_singleton()->draw_command_begin_label("GI Render"); - RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(rb == nullptr); - RendererSceneEnvironmentRD *env = p_scene_render->environment_owner.getornull(p_environment); if (rb->ambient_buffer.is_null() || rb->gi.using_half_size_gi != half_resolution) { if (rb->ambient_buffer.is_valid()) { @@ -3160,16 +3159,6 @@ void RendererSceneGIRD::process_gi(RID p_render_buffers, RID p_normal_roughness_ bool use_sdfgi = rb->sdfgi != nullptr; bool use_voxel_gi_instances = push_constant.max_voxel_gi_instances > 0; - if (env) { - push_constant.ao_color[0] = env->ao_color.r; - push_constant.ao_color[1] = env->ao_color.g; - push_constant.ao_color[2] = env->ao_color.b; - } else { - push_constant.ao_color[0] = 0; - push_constant.ao_color[1] = 0; - push_constant.ao_color[2] = 0; - } - push_constant.cam_rotation[0] = p_transform.basis[0][0]; push_constant.cam_rotation[1] = p_transform.basis[1][0]; push_constant.cam_rotation[2] = p_transform.basis[2][0]; @@ -3393,7 +3382,7 @@ void RendererSceneGIRD::voxel_gi_update(RID p_probe, bool p_update_light_instanc } void RendererSceneGIRD::debug_voxel_gi(RID p_voxel_gi, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha) { - VoxelGIInstance *voxel_gi = voxel_gi_instance_owner.getornull(p_voxel_gi); + VoxelGIInstance *voxel_gi = voxel_gi_instance_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->debug(p_draw_list, p_framebuffer, p_camera_with_transform, p_lighting, p_emission, p_alpha); diff --git a/servers/rendering/renderer_rd/renderer_scene_gi_rd.h b/servers/rendering/renderer_rd/renderer_scene_gi_rd.h index 128bf09063..9f6fc8a9b7 100644 --- a/servers/rendering/renderer_rd/renderer_scene_gi_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_gi_rd.h @@ -110,8 +110,8 @@ private: float pad[3]; }; - VoxelGILight *voxel_gi_lights; - uint32_t voxel_gi_max_lights; + VoxelGILight *voxel_gi_lights = nullptr; + uint32_t voxel_gi_max_lights = 32; RID voxel_gi_lights_uniform; enum { @@ -383,7 +383,7 @@ public: mutable RID_Owner<VoxelGIInstance> voxel_gi_instance_owner; _FORCE_INLINE_ VoxelGIInstance *get_probe_instance(RID p_probe) const { - return voxel_gi_instance_owner.getornull(p_probe); + return voxel_gi_instance_owner.get_or_null(p_probe); }; _FORCE_INLINE_ RID voxel_gi_instance_get_texture(RID p_probe) { @@ -623,12 +623,11 @@ public: float z_far; float proj_info[4]; - float ao_color[3]; - uint32_t max_voxel_gi_instances; + uint32_t max_voxel_gi_instances; uint32_t high_quality_vct; uint32_t orthogonal; - uint32_t pad[2]; + uint32_t pad; float cam_rotation[12]; }; diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp index 4870f5f397..d128578d0b 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp @@ -48,8 +48,8 @@ void get_vogel_disk(float *r_kernel, int p_sample_count) { } void RendererSceneRenderRD::sdfgi_update(RID p_render_buffers, RID p_environment, const Vector3 &p_world_position) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_environment); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); bool needs_sdfgi = env && env->sdfgi_enabled; if (!needs_sdfgi) { @@ -83,7 +83,7 @@ void RendererSceneRenderRD::sdfgi_update(RID p_render_buffers, RID p_environment } int RendererSceneRenderRD::sdfgi_get_pending_region_count(RID p_render_buffers) const { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(rb == nullptr, 0); @@ -113,7 +113,7 @@ AABB RendererSceneRenderRD::sdfgi_get_pending_region_bounds(RID p_render_buffers AABB bounds; Vector3i from; Vector3i size; - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(rb == nullptr, AABB()); ERR_FAIL_COND_V(rb->sdfgi == nullptr, AABB()); @@ -126,7 +126,7 @@ uint32_t RendererSceneRenderRD::sdfgi_get_pending_region_cascade(RID p_render_bu AABB bounds; Vector3i from; Vector3i size; - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(rb == nullptr, -1); ERR_FAIL_COND_V(rb->sdfgi == nullptr, -1); @@ -164,139 +164,133 @@ void RendererSceneRenderRD::environment_initialize(RID p_rid) { } void RendererSceneRenderRD::environment_set_background(RID p_env, RS::EnvironmentBG p_bg) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->background = p_bg; } void RendererSceneRenderRD::environment_set_sky(RID p_env, RID p_sky) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->sky = p_sky; } void RendererSceneRenderRD::environment_set_sky_custom_fov(RID p_env, float p_scale) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->sky_custom_fov = p_scale; } void RendererSceneRenderRD::environment_set_sky_orientation(RID p_env, const Basis &p_orientation) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->sky_orientation = p_orientation; } void RendererSceneRenderRD::environment_set_bg_color(RID p_env, const Color &p_color) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->bg_color = p_color; } void RendererSceneRenderRD::environment_set_bg_energy(RID p_env, float p_energy) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->bg_energy = p_energy; } void RendererSceneRenderRD::environment_set_canvas_max_layer(RID p_env, int p_max_layer) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->canvas_max_layer = p_max_layer; } -void RendererSceneRenderRD::environment_set_ambient_light(RID p_env, const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); +void RendererSceneRenderRD::environment_set_ambient_light(RID p_env, const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source) { + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); - env->set_ambient_light(p_color, p_ambient, p_energy, p_sky_contribution, p_reflection_source, p_ao_color); + env->set_ambient_light(p_color, p_ambient, p_energy, p_sky_contribution, p_reflection_source); } RS::EnvironmentBG RendererSceneRenderRD::environment_get_background(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, RS::ENV_BG_MAX); return env->background; } RID RendererSceneRenderRD::environment_get_sky(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, RID()); return env->sky; } float RendererSceneRenderRD::environment_get_sky_custom_fov(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->sky_custom_fov; } Basis RendererSceneRenderRD::environment_get_sky_orientation(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, Basis()); return env->sky_orientation; } Color RendererSceneRenderRD::environment_get_bg_color(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, Color()); return env->bg_color; } float RendererSceneRenderRD::environment_get_bg_energy(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->bg_energy; } int RendererSceneRenderRD::environment_get_canvas_max_layer(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->canvas_max_layer; } Color RendererSceneRenderRD::environment_get_ambient_light_color(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, Color()); return env->ambient_light; } RS::EnvironmentAmbientSource RendererSceneRenderRD::environment_get_ambient_source(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, RS::ENV_AMBIENT_SOURCE_BG); return env->ambient_source; } float RendererSceneRenderRD::environment_get_ambient_light_energy(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->ambient_light_energy; } float RendererSceneRenderRD::environment_get_ambient_sky_contribution(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->ambient_sky_contribution; } RS::EnvironmentReflectionSource RendererSceneRenderRD::environment_get_reflection_source(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, RS::ENV_REFLECTION_SOURCE_DISABLED); return env->reflection_source; } -Color RendererSceneRenderRD::environment_get_ao_color(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); - ERR_FAIL_COND_V(!env, Color()); - return env->ao_color; -} - void RendererSceneRenderRD::environment_set_tonemap(RID p_env, RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->set_tonemap(p_tone_mapper, p_exposure, p_white, p_auto_exposure, p_min_luminance, p_max_luminance, p_auto_exp_speed, p_auto_exp_scale); } void RendererSceneRenderRD::environment_set_glow(RID p_env, bool p_enable, Vector<float> p_levels, float p_intensity, float p_strength, float p_mix, float p_bloom_threshold, RS::EnvironmentGlowBlendMode p_blend_mode, float p_hdr_bleed_threshold, float p_hdr_bleed_scale, float p_hdr_luminance_cap) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->set_glow(p_enable, p_levels, p_intensity, p_strength, p_mix, p_bloom_threshold, p_blend_mode, p_hdr_bleed_threshold, p_hdr_bleed_scale, p_hdr_luminance_cap); } @@ -310,7 +304,7 @@ void RendererSceneRenderRD::environment_glow_set_use_high_quality(bool p_enable) } void RendererSceneRenderRD::environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); if (!is_dynamic_gi_supported()) { @@ -321,58 +315,58 @@ void RendererSceneRenderRD::environment_set_sdfgi(RID p_env, bool p_enable, RS:: } void RendererSceneRenderRD::environment_set_fog(RID p_env, bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_fog_aerial_perspective) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->set_fog(p_enable, p_light_color, p_light_energy, p_sun_scatter, p_density, p_height, p_height_density, p_fog_aerial_perspective); } bool RendererSceneRenderRD::environment_is_fog_enabled(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, false); return env->fog_enabled; } Color RendererSceneRenderRD::environment_get_fog_light_color(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, Color()); return env->fog_light_color; } float RendererSceneRenderRD::environment_get_fog_light_energy(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_light_energy; } float RendererSceneRenderRD::environment_get_fog_sun_scatter(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_sun_scatter; } float RendererSceneRenderRD::environment_get_fog_density(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_density; } float RendererSceneRenderRD::environment_get_fog_height(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_height; } float RendererSceneRenderRD::environment_get_fog_height_density(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_height_density; } float RendererSceneRenderRD::environment_get_fog_aerial_perspective(RID p_env) const { - const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_aerial_perspective; } void RendererSceneRenderRD::environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); if (!is_volumetric_supported()) { @@ -403,7 +397,7 @@ void RendererSceneRenderRD::environment_set_sdfgi_frames_to_update_light(RS::Env } void RendererSceneRenderRD::environment_set_ssr(RID p_env, bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->set_ssr(p_enable, p_max_steps, p_fade_int, p_fade_out, p_depth_tolerance); @@ -418,7 +412,7 @@ RS::EnvironmentSSRRoughnessQuality RendererSceneRenderRD::environment_get_ssr_ro } void RendererSceneRenderRD::environment_set_ssao(RID p_env, bool p_enable, float p_radius, float p_intensity, float p_power, float p_detail, float p_horizon, float p_sharpness, float p_light_affect, float p_ao_channel_affect) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->set_ssao(p_enable, p_radius, p_intensity, p_power, p_detail, p_horizon, p_sharpness, p_light_affect, p_ao_channel_affect); @@ -434,30 +428,30 @@ void RendererSceneRenderRD::environment_set_ssao_quality(RS::EnvironmentSSAOQual } bool RendererSceneRenderRD::environment_is_ssao_enabled(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, false); return env->ssao_enabled; } float RendererSceneRenderRD::environment_get_ssao_ao_affect(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0.0); return env->ssao_ao_channel_affect; } float RendererSceneRenderRD::environment_get_ssao_light_affect(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, 0.0); return env->ssao_direct_light_affect; } bool RendererSceneRenderRD::environment_is_ssr_enabled(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, false); return env->ssr_enabled; } bool RendererSceneRenderRD::environment_is_sdfgi_enabled(RID p_env) const { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, false); return env->sdfgi_enabled; } @@ -467,7 +461,7 @@ bool RendererSceneRenderRD::is_environment(RID p_env) const { } Ref<Image> RendererSceneRenderRD::environment_bake_panorama(RID p_env, bool p_bake_irradiance, const Size2i &p_size) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND_V(!env, Ref<Image>()); if (env->background == RS::ENV_BG_CAMERA_FEED || env->background == RS::ENV_BG_CANVAS || env->background == RS::ENV_BG_KEEP) { @@ -522,7 +516,7 @@ RID RendererSceneRenderRD::reflection_atlas_create() { } void RendererSceneRenderRD::reflection_atlas_set_size(RID p_ref_atlas, int p_reflection_size, int p_reflection_count) { - ReflectionAtlas *ra = reflection_atlas_owner.getornull(p_ref_atlas); + ReflectionAtlas *ra = reflection_atlas_owner.get_or_null(p_ref_atlas); ERR_FAIL_COND(!ra); if (ra->size == p_reflection_size && ra->count == p_reflection_count) { @@ -557,7 +551,7 @@ void RendererSceneRenderRD::reflection_atlas_set_size(RID p_ref_atlas, int p_ref } int RendererSceneRenderRD::reflection_atlas_get_size(RID p_ref_atlas) const { - ReflectionAtlas *ra = reflection_atlas_owner.getornull(p_ref_atlas); + ReflectionAtlas *ra = reflection_atlas_owner.get_or_null(p_ref_atlas); ERR_FAIL_COND_V(!ra, 0); return ra->size; @@ -573,7 +567,7 @@ RID RendererSceneRenderRD::reflection_probe_instance_create(RID p_probe) { } void RendererSceneRenderRD::reflection_probe_instance_set_transform(RID p_instance, const Transform3D &p_transform) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND(!rpi); rpi->transform = p_transform; @@ -581,13 +575,13 @@ void RendererSceneRenderRD::reflection_probe_instance_set_transform(RID p_instan } void RendererSceneRenderRD::reflection_probe_release_atlas_index(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND(!rpi); if (rpi->atlas.is_null()) { return; //nothing to release } - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(rpi->atlas); ERR_FAIL_COND(!atlas); ERR_FAIL_INDEX(rpi->atlas_index, atlas->reflections.size()); atlas->reflections.write[rpi->atlas_index].owner = RID(); @@ -596,7 +590,7 @@ void RendererSceneRenderRD::reflection_probe_release_atlas_index(RID p_instance) } bool RendererSceneRenderRD::reflection_probe_instance_needs_redraw(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, false); if (rpi->rendering) { @@ -615,20 +609,22 @@ bool RendererSceneRenderRD::reflection_probe_instance_needs_redraw(RID p_instanc } bool RendererSceneRenderRD::reflection_probe_instance_has_reflection(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, false); return rpi->atlas.is_valid(); } bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instance, RID p_reflection_atlas) { - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(p_reflection_atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(p_reflection_atlas); ERR_FAIL_COND_V(!atlas, false); - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, false); + RD::get_singleton()->draw_command_begin_label("Reflection probe render"); + if (storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS && atlas->reflection.is_valid() && atlas->size != 256) { WARN_PRINT("ReflectionProbes set to UPDATE_ALWAYS must have an atlas size of 256. Please update the atlas size in the ProjectSettings."); reflection_atlas_set_size(p_reflection_atlas, 256, atlas->count); @@ -675,12 +671,9 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc } atlas->reflections.resize(atlas->count); for (int i = 0; i < atlas->count; i++) { - atlas->reflections.write[i].data.update_reflection_data(atlas->size, mipmaps, false, atlas->reflection, i * 6, storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS, sky.roughness_layers); + atlas->reflections.write[i].data.update_reflection_data(storage, atlas->size, mipmaps, false, atlas->reflection, i * 6, storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS, sky.roughness_layers, _render_buffers_get_color_format()); for (int j = 0; j < 6; j++) { - Vector<RID> fb; - fb.push_back(atlas->reflections.write[i].data.layers[0].mipmaps[0].views[j]); - fb.push_back(atlas->depth_buffer); - atlas->reflections.write[i].fbs[j] = RD::get_singleton()->framebuffer_create(fb); + atlas->reflections.write[i].fbs[j] = reflection_probe_create_framebuffer(atlas->reflections.write[i].data.layers[0].mipmaps[0].views[j], atlas->depth_buffer); } } @@ -702,7 +695,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc uint64_t pass_min = 0; for (int i = 0; i < atlas->reflections.size(); i++) { - ReflectionProbeInstance *rpi2 = reflection_probe_instance_owner.getornull(atlas->reflections[i].owner); + ReflectionProbeInstance *rpi2 = reflection_probe_instance_owner.get_or_null(atlas->reflections[i].owner); if (rpi2->last_pass < pass_min) { pass_min = rpi2->last_pass; rpi->atlas_index = i; @@ -721,16 +714,25 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc rpi->processing_layer = 1; rpi->processing_side = 0; + RD::get_singleton()->draw_command_end_label(); + return true; } +RID RendererSceneRenderRD::reflection_probe_create_framebuffer(RID p_color, RID p_depth) { + Vector<RID> fb; + fb.push_back(p_color); + fb.push_back(p_depth); + return RD::get_singleton()->framebuffer_create(fb); +} + bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, false); ERR_FAIL_COND_V(!rpi->rendering, false); ERR_FAIL_COND_V(rpi->atlas.is_null(), false); - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(rpi->atlas); if (!atlas || rpi->atlas_index == -1) { //does not belong to an atlas anymore, cancel (was removed from atlas or atlas changed while rendering) rpi->rendering = false; @@ -771,30 +773,30 @@ bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_ins } uint32_t RendererSceneRenderRD::reflection_probe_instance_get_resolution(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, 0); - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(rpi->atlas); ERR_FAIL_COND_V(!atlas, 0); return atlas->size; } RID RendererSceneRenderRD::reflection_probe_instance_get_framebuffer(RID p_instance, int p_index) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, RID()); ERR_FAIL_INDEX_V(p_index, 6, RID()); - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(rpi->atlas); ERR_FAIL_COND_V(!atlas, RID()); return atlas->reflections[rpi->atlas_index].fbs[p_index]; } RID RendererSceneRenderRD::reflection_probe_instance_get_depth_framebuffer(RID p_instance, int p_index) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, RID()); ERR_FAIL_INDEX_V(p_index, 6, RID()); - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(rpi->atlas); ERR_FAIL_COND_V(!atlas, RID()); return atlas->depth_fb; } @@ -821,7 +823,7 @@ void RendererSceneRenderRD::_update_shadow_atlas(ShadowAtlas *shadow_atlas) { } void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size, bool p_16_bits) { - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND(!shadow_atlas); ERR_FAIL_COND(p_size < 0); p_size = next_power_of_2(p_size); @@ -842,8 +844,8 @@ void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size, bool } //erase shadow atlas reference from lights - for (Map<RID, uint32_t>::Element *E = shadow_atlas->shadow_owners.front(); E; E = E->next()) { - LightInstance *li = light_instance_owner.getornull(E->key()); + for (const KeyValue<RID, uint32_t> &E : shadow_atlas->shadow_owners) { + LightInstance *li = light_instance_owner.get_or_null(E.key); ERR_CONTINUE(!li); li->shadow_atlases.erase(p_atlas); } @@ -852,11 +854,11 @@ void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size, bool shadow_atlas->shadow_owners.clear(); shadow_atlas->size = p_size; - shadow_atlas->use_16_bits = p_size; + shadow_atlas->use_16_bits = p_16_bits; } void RendererSceneRenderRD::shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision) { - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND(!shadow_atlas); ERR_FAIL_INDEX(p_quadrant, 4); ERR_FAIL_INDEX(p_subdivision, 16384); @@ -878,7 +880,7 @@ void RendererSceneRenderRD::shadow_atlas_set_quadrant_subdivision(RID p_atlas, i for (int i = 0; i < shadow_atlas->quadrants[p_quadrant].shadows.size(); i++) { if (shadow_atlas->quadrants[p_quadrant].shadows[i].owner.is_valid()) { shadow_atlas->shadow_owners.erase(shadow_atlas->quadrants[p_quadrant].shadows[i].owner); - LightInstance *li = light_instance_owner.getornull(shadow_atlas->quadrants[p_quadrant].shadows[i].owner); + LightInstance *li = light_instance_owner.get_or_null(shadow_atlas->quadrants[p_quadrant].shadows[i].owner); ERR_CONTINUE(!li); li->shadow_atlases.erase(p_atlas); } @@ -927,7 +929,7 @@ bool RendererSceneRenderRD::_shadow_atlas_find_shadow(ShadowAtlas *shadow_atlas, //look for an empty space int sc = shadow_atlas->quadrants[qidx].shadows.size(); - ShadowAtlas::Quadrant::Shadow *sarr = shadow_atlas->quadrants[qidx].shadows.ptrw(); + const ShadowAtlas::Quadrant::Shadow *sarr = shadow_atlas->quadrants[qidx].shadows.ptr(); int found_free_idx = -1; //found a free one int found_used_idx = -1; //found existing one, must steal it @@ -939,7 +941,7 @@ bool RendererSceneRenderRD::_shadow_atlas_find_shadow(ShadowAtlas *shadow_atlas, break; } - LightInstance *sli = light_instance_owner.getornull(sarr[j].owner); + LightInstance *sli = light_instance_owner.get_or_null(sarr[j].owner); ERR_CONTINUE(!sli); if (sli->last_scene_pass != scene_pass) { @@ -972,11 +974,83 @@ bool RendererSceneRenderRD::_shadow_atlas_find_shadow(ShadowAtlas *shadow_atlas, return false; } +bool RendererSceneRenderRD::_shadow_atlas_find_omni_shadows(ShadowAtlas *shadow_atlas, int *p_in_quadrants, int p_quadrant_count, int p_current_subdiv, uint64_t p_tick, int &r_quadrant, int &r_shadow) { + for (int i = p_quadrant_count - 1; i >= 0; i--) { + int qidx = p_in_quadrants[i]; + + if (shadow_atlas->quadrants[qidx].subdivision == (uint32_t)p_current_subdiv) { + return false; + } + + //look for an empty space + int sc = shadow_atlas->quadrants[qidx].shadows.size(); + const ShadowAtlas::Quadrant::Shadow *sarr = shadow_atlas->quadrants[qidx].shadows.ptr(); + + int found_idx = -1; + uint64_t min_pass = 0; // sum of currently selected spots, try to get the least recently used pair + + for (int j = 0; j < sc - 1; j++) { + uint64_t pass = 0; + + if (sarr[j].owner.is_valid()) { + LightInstance *sli = light_instance_owner.get_or_null(sarr[j].owner); + ERR_CONTINUE(!sli); + + if (sli->last_scene_pass == scene_pass) { + continue; + } + + //was just allocated, don't kill it so soon, wait a bit.. + if (p_tick - sarr[j].alloc_tick < shadow_atlas_realloc_tolerance_msec) { + continue; + } + pass += sli->last_scene_pass; + } + + if (sarr[j + 1].owner.is_valid()) { + LightInstance *sli = light_instance_owner.get_or_null(sarr[j + 1].owner); + ERR_CONTINUE(!sli); + + if (sli->last_scene_pass == scene_pass) { + continue; + } + + //was just allocated, don't kill it so soon, wait a bit.. + if (p_tick - sarr[j + 1].alloc_tick < shadow_atlas_realloc_tolerance_msec) { + continue; + } + pass += sli->last_scene_pass; + } + + if (found_idx == -1 || pass < min_pass) { + found_idx = j; + min_pass = pass; + + // we found two empty spots, no need to check the rest + if (pass == 0) { + break; + } + } + } + + if (found_idx == -1) { + continue; //nothing found + } + + r_quadrant = qidx; + r_shadow = found_idx; + + return true; + } + + return false; +} + bool RendererSceneRenderRD::shadow_atlas_update_light(RID p_atlas, RID p_light_intance, float p_coverage, uint64_t p_light_version) { - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND_V(!shadow_atlas, false); - LightInstance *li = light_instance_owner.getornull(p_light_intance); + LightInstance *li = light_instance_owner.get_or_null(p_light_intance); ERR_FAIL_COND_V(!li, false); if (shadow_atlas->size == 0 || shadow_atlas->smallest_subdiv == 0) { @@ -1017,94 +1091,104 @@ bool RendererSceneRenderRD::shadow_atlas_update_light(RID p_atlas, RID p_light_i uint64_t tick = OS::get_singleton()->get_ticks_msec(); - //see if it already exists + uint32_t old_key = ShadowAtlas::SHADOW_INVALID; + uint32_t old_quadrant = ShadowAtlas::SHADOW_INVALID; + uint32_t old_shadow = ShadowAtlas::SHADOW_INVALID; + int old_subdivision = -1; + + bool should_realloc = false; + bool should_redraw = false; if (shadow_atlas->shadow_owners.has(p_light_intance)) { - //it does! - uint32_t key = shadow_atlas->shadow_owners[p_light_intance]; - uint32_t q = (key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; - uint32_t s = key & ShadowAtlas::SHADOW_INDEX_MASK; + old_key = shadow_atlas->shadow_owners[p_light_intance]; + old_quadrant = (old_key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; + old_shadow = old_key & ShadowAtlas::SHADOW_INDEX_MASK; - bool should_realloc = shadow_atlas->quadrants[q].subdivision != (uint32_t)best_subdiv && (shadow_atlas->quadrants[q].shadows[s].alloc_tick - tick > shadow_atlas_realloc_tolerance_msec); - bool should_redraw = shadow_atlas->quadrants[q].shadows[s].version != p_light_version; + should_realloc = shadow_atlas->quadrants[old_quadrant].subdivision != (uint32_t)best_subdiv && (shadow_atlas->quadrants[old_quadrant].shadows[old_shadow].alloc_tick - tick > shadow_atlas_realloc_tolerance_msec); + should_redraw = shadow_atlas->quadrants[old_quadrant].shadows[old_shadow].version != p_light_version; if (!should_realloc) { - shadow_atlas->quadrants[q].shadows.write[s].version = p_light_version; + shadow_atlas->quadrants[old_quadrant].shadows.write[old_shadow].version = p_light_version; //already existing, see if it should redraw or it's just OK return should_redraw; } - int new_quadrant, new_shadow; + old_subdivision = shadow_atlas->quadrants[old_quadrant].subdivision; + } - //find a better place - if (_shadow_atlas_find_shadow(shadow_atlas, valid_quadrants, valid_quadrant_count, shadow_atlas->quadrants[q].subdivision, tick, new_quadrant, new_shadow)) { - //found a better place! - ShadowAtlas::Quadrant::Shadow *sh = &shadow_atlas->quadrants[new_quadrant].shadows.write[new_shadow]; - if (sh->owner.is_valid()) { - //is taken, but is invalid, erasing it - shadow_atlas->shadow_owners.erase(sh->owner); - LightInstance *sli = light_instance_owner.getornull(sh->owner); - sli->shadow_atlases.erase(p_atlas); - } + bool is_omni = li->light_type == RS::LIGHT_OMNI; + bool found_shadow = false; + int new_quadrant = -1; + int new_shadow = -1; - //erase previous - shadow_atlas->quadrants[q].shadows.write[s].version = 0; - shadow_atlas->quadrants[q].shadows.write[s].owner = RID(); + if (is_omni) { + found_shadow = _shadow_atlas_find_omni_shadows(shadow_atlas, valid_quadrants, valid_quadrant_count, old_subdivision, tick, new_quadrant, new_shadow); + } else { + found_shadow = _shadow_atlas_find_shadow(shadow_atlas, valid_quadrants, valid_quadrant_count, old_subdivision, tick, new_quadrant, new_shadow); + } - sh->owner = p_light_intance; - sh->alloc_tick = tick; - sh->version = p_light_version; - li->shadow_atlases.insert(p_atlas); - - //make new key - key = new_quadrant << ShadowAtlas::QUADRANT_SHIFT; - key |= new_shadow; - //update it in map - shadow_atlas->shadow_owners[p_light_intance] = key; - //make it dirty, as it should redraw anyway - return true; + if (found_shadow) { + if (old_quadrant != ShadowAtlas::SHADOW_INVALID) { + shadow_atlas->quadrants[old_quadrant].shadows.write[old_shadow].version = 0; + shadow_atlas->quadrants[old_quadrant].shadows.write[old_shadow].owner = RID(); + + if (old_key & ShadowAtlas::OMNI_LIGHT_FLAG) { + shadow_atlas->quadrants[old_quadrant].shadows.write[old_shadow + 1].version = 0; + shadow_atlas->quadrants[old_quadrant].shadows.write[old_shadow + 1].owner = RID(); + } } - //no better place for this shadow found, keep current + uint32_t new_key = new_quadrant << ShadowAtlas::QUADRANT_SHIFT; + new_key |= new_shadow; - //already existing, see if it should redraw or it's just OK + ShadowAtlas::Quadrant::Shadow *sh = &shadow_atlas->quadrants[new_quadrant].shadows.write[new_shadow]; + _shadow_atlas_invalidate_shadow(sh, p_atlas, shadow_atlas, new_quadrant, new_shadow); - shadow_atlas->quadrants[q].shadows.write[s].version = p_light_version; + sh->owner = p_light_intance; + sh->alloc_tick = tick; + sh->version = p_light_version; - return should_redraw; - } + if (is_omni) { + new_key |= ShadowAtlas::OMNI_LIGHT_FLAG; - int new_quadrant, new_shadow; + int new_omni_shadow = new_shadow + 1; + ShadowAtlas::Quadrant::Shadow *extra_sh = &shadow_atlas->quadrants[new_quadrant].shadows.write[new_omni_shadow]; + _shadow_atlas_invalidate_shadow(extra_sh, p_atlas, shadow_atlas, new_quadrant, new_omni_shadow); - //find a better place - if (_shadow_atlas_find_shadow(shadow_atlas, valid_quadrants, valid_quadrant_count, -1, tick, new_quadrant, new_shadow)) { - //found a better place! - ShadowAtlas::Quadrant::Shadow *sh = &shadow_atlas->quadrants[new_quadrant].shadows.write[new_shadow]; - if (sh->owner.is_valid()) { - //is taken, but is invalid, erasing it - shadow_atlas->shadow_owners.erase(sh->owner); - LightInstance *sli = light_instance_owner.getornull(sh->owner); - sli->shadow_atlases.erase(p_atlas); + extra_sh->owner = p_light_intance; + extra_sh->alloc_tick = tick; + extra_sh->version = p_light_version; } - sh->owner = p_light_intance; - sh->alloc_tick = tick; - sh->version = p_light_version; li->shadow_atlases.insert(p_atlas); - //make new key - uint32_t key = new_quadrant << ShadowAtlas::QUADRANT_SHIFT; - key |= new_shadow; //update it in map - shadow_atlas->shadow_owners[p_light_intance] = key; + shadow_atlas->shadow_owners[p_light_intance] = new_key; //make it dirty, as it should redraw anyway - return true; } - //no place to allocate this light, apologies + return should_redraw; +} - return false; +void RendererSceneRenderRD::_shadow_atlas_invalidate_shadow(RendererSceneRenderRD::ShadowAtlas::Quadrant::Shadow *p_shadow, RID p_atlas, RendererSceneRenderRD::ShadowAtlas *p_shadow_atlas, uint32_t p_quadrant, uint32_t p_shadow_idx) { + if (p_shadow->owner.is_valid()) { + LightInstance *sli = light_instance_owner.get_or_null(p_shadow->owner); + uint32_t old_key = p_shadow_atlas->shadow_owners[p_shadow->owner]; + + if (old_key & ShadowAtlas::OMNI_LIGHT_FLAG) { + uint32_t s = old_key & ShadowAtlas::SHADOW_INDEX_MASK; + uint32_t omni_shadow_idx = p_shadow_idx + (s == (uint32_t)p_shadow_idx ? 1 : -1); + RendererSceneRenderRD::ShadowAtlas::Quadrant::Shadow *omni_shadow = &p_shadow_atlas->quadrants[p_quadrant].shadows.write[omni_shadow_idx]; + omni_shadow->version = 0; + omni_shadow->owner = RID(); + } + + p_shadow->version = 0; + p_shadow->owner = RID(); + sli->shadow_atlases.erase(p_atlas); + p_shadow_atlas->shadow_owners.erase(p_shadow->owner); + } } void RendererSceneRenderRD::_update_directional_shadow_atlas() { @@ -1129,6 +1213,7 @@ void RendererSceneRenderRD::directional_shadow_atlas_set_size(int p_size, bool p } directional_shadow.size = p_size; + directional_shadow.use_16_bits = p_16_bits; if (directional_shadow.depth.is_valid()) { RD::get_singleton()->free(directional_shadow.depth); @@ -1169,7 +1254,7 @@ int RendererSceneRenderRD::get_directional_light_shadow_size(RID p_light_intance Rect2i r = _get_directional_shadow_rect(directional_shadow.size, directional_shadow.light_count, 0); - LightInstance *light_instance = light_instance_owner.getornull(p_light_intance); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light_intance); ERR_FAIL_COND_V(!light_instance, 0); switch (storage->light_directional_get_shadow_mode(light_instance->light)) { @@ -1205,7 +1290,7 @@ void RendererSceneRenderRD::camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokeh } void RendererSceneRenderRD::camera_effects_set_dof_blur(RID p_camera_effects, bool p_far_enable, float p_far_distance, float p_far_transition, bool p_near_enable, float p_near_distance, float p_near_transition, float p_amount) { - CameraEffects *camfx = camera_effects_owner.getornull(p_camera_effects); + CameraEffects *camfx = camera_effects_owner.get_or_null(p_camera_effects); ERR_FAIL_COND(!camfx); camfx->dof_blur_far_enabled = p_far_enable; @@ -1220,7 +1305,7 @@ void RendererSceneRenderRD::camera_effects_set_dof_blur(RID p_camera_effects, bo } void RendererSceneRenderRD::camera_effects_set_custom_exposure(RID p_camera_effects, bool p_enable, float p_exposure) { - CameraEffects *camfx = camera_effects_owner.getornull(p_camera_effects); + CameraEffects *camfx = camera_effects_owner.get_or_null(p_camera_effects); ERR_FAIL_COND(!camfx); camfx->override_exposure_enabled = p_enable; @@ -1230,7 +1315,7 @@ void RendererSceneRenderRD::camera_effects_set_custom_exposure(RID p_camera_effe RID RendererSceneRenderRD::light_instance_create(RID p_light) { RID li = light_instance_owner.make_rid(LightInstance()); - LightInstance *light_instance = light_instance_owner.getornull(li); + LightInstance *light_instance = light_instance_owner.get_or_null(li); light_instance->self = li; light_instance->light = p_light; @@ -1243,21 +1328,21 @@ RID RendererSceneRenderRD::light_instance_create(RID p_light) { } void RendererSceneRenderRD::light_instance_set_transform(RID p_light_instance, const Transform3D &p_transform) { - LightInstance *light_instance = light_instance_owner.getornull(p_light_instance); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light_instance); ERR_FAIL_COND(!light_instance); light_instance->transform = p_transform; } void RendererSceneRenderRD::light_instance_set_aabb(RID p_light_instance, const AABB &p_aabb) { - LightInstance *light_instance = light_instance_owner.getornull(p_light_instance); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light_instance); ERR_FAIL_COND(!light_instance); light_instance->aabb = p_aabb; } void RendererSceneRenderRD::light_instance_set_shadow_transform(RID p_light_instance, const CameraMatrix &p_projection, const Transform3D &p_transform, float p_far, float p_split, int p_pass, float p_shadow_texel_size, float p_bias_scale, float p_range_begin, const Vector2 &p_uv_scale) { - LightInstance *light_instance = light_instance_owner.getornull(p_light_instance); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light_instance); ERR_FAIL_COND(!light_instance); ERR_FAIL_INDEX(p_pass, 6); @@ -1273,7 +1358,7 @@ void RendererSceneRenderRD::light_instance_set_shadow_transform(RID p_light_inst } void RendererSceneRenderRD::light_instance_mark_visible(RID p_light_instance) { - LightInstance *light_instance = light_instance_owner.getornull(p_light_instance); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light_instance); ERR_FAIL_COND(!light_instance); light_instance->last_scene_pass = scene_pass; @@ -1316,7 +1401,7 @@ RID RendererSceneRenderRD::decal_instance_create(RID p_decal) { } void RendererSceneRenderRD::decal_instance_set_transform(RID p_decal, const Transform3D &p_transform) { - DecalInstance *di = decal_instance_owner.getornull(p_decal); + DecalInstance *di = decal_instance_owner.get_or_null(p_decal); ERR_FAIL_COND(!di); di->transform = p_transform; } @@ -1329,7 +1414,7 @@ RID RendererSceneRenderRD::lightmap_instance_create(RID p_lightmap) { return lightmap_instance_owner.make_rid(li); } void RendererSceneRenderRD::lightmap_instance_set_transform(RID p_lightmap, const Transform3D &p_transform) { - LightmapInstance *li = lightmap_instance_owner.getornull(p_lightmap); + LightmapInstance *li = lightmap_instance_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!li); li->transform = p_transform; } @@ -1361,7 +1446,7 @@ void RendererSceneRenderRD::voxel_gi_update(RID p_probe, bool p_update_light_ins } void RendererSceneRenderRD::_debug_sdfgi_probes(RID p_render_buffers, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); if (!rb->sdfgi) { @@ -1383,12 +1468,20 @@ void RendererSceneRenderRD::_allocate_blur_textures(RenderBuffers *rb) { uint32_t mipmaps_required = Image::get_image_required_mipmaps(rb->width, rb->height, Image::FORMAT_RGBAH); + // TODO make sure texture_create_shared_from_slice works for multiview + RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = _render_buffers_get_color_format(); // RD::DATA_FORMAT_R16G16B16A16_SFLOAT; tf.width = rb->width; tf.height = rb->height; - tf.texture_type = RD::TEXTURE_TYPE_2D; - tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; + tf.texture_type = rb->view_count > 1 ? RD::TEXTURE_TYPE_2D_ARRAY : RD::TEXTURE_TYPE_2D; + tf.array_layers = rb->view_count; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; + if (_render_buffers_can_be_storage()) { + tf.usage_bits += RD::TEXTURE_USAGE_STORAGE_BIT; + } else { + tf.usage_bits += RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + } tf.mipmaps = mipmaps_required; rb->blur[0].texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); @@ -1408,17 +1501,125 @@ void RendererSceneRenderRD::_allocate_blur_textures(RenderBuffers *rb) { mm.width = base_width; mm.height = base_height; + if (!_render_buffers_can_be_storage()) { + Vector<RID> fb; + fb.push_back(mm.texture); + + mm.fb = RD::get_singleton()->framebuffer_create(fb); + } + + if (!_render_buffers_can_be_storage()) { + // and half texture, this is an intermediate result so just allocate a texture, is this good enough? + tf.width = MAX(1, base_width >> 1); + tf.height = base_height; + tf.mipmaps = 1; // 1 or 0? + + mm.half_texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + Vector<RID> half_fb; + half_fb.push_back(mm.half_texture); + mm.half_fb = RD::get_singleton()->framebuffer_create(half_fb); + } + rb->blur[0].mipmaps.push_back(mm); if (i > 0) { mm.texture = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), rb->blur[1].texture, 0, i - 1); + if (!_render_buffers_can_be_storage()) { + Vector<RID> fb; + fb.push_back(mm.texture); + + mm.fb = RD::get_singleton()->framebuffer_create(fb); + + // We can re-use the half texture here as it is an intermediate result + } + rb->blur[1].mipmaps.push_back(mm); } base_width = MAX(1, base_width >> 1); base_height = MAX(1, base_height >> 1); } + + if (!_render_buffers_can_be_storage()) { + // create 4 weight textures, 2 full size, 2 half size + + tf.format = RD::DATA_FORMAT_R16_SFLOAT; // We could probably use DATA_FORMAT_R8_SNORM if we don't pre-multiply by blur_size but that depends on whether we can remove DEPTH_GAP + tf.width = rb->width; + tf.height = rb->height; + tf.texture_type = rb->view_count > 1 ? RD::TEXTURE_TYPE_2D_ARRAY : RD::TEXTURE_TYPE_2D; + tf.array_layers = rb->view_count; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tf.mipmaps = 1; + for (uint32_t i = 0; i < 4; i++) { + // associated blur texture + RID texture; + if (i == 0) { + texture = rb->texture; + } else if (i == 1) { + texture = rb->blur[0].mipmaps[0].texture; + } else if (i == 2) { + texture = rb->blur[1].mipmaps[0].texture; + } else if (i == 3) { + texture = rb->blur[0].mipmaps[1].texture; + } + + // create weight texture + rb->weight_buffers[i].weight = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + // create frame buffer + Vector<RID> fb; + fb.push_back(texture); + fb.push_back(rb->weight_buffers[i].weight); + rb->weight_buffers[i].fb = RD::get_singleton()->framebuffer_create(fb); + + if (i == 1) { + // next 2 are half size + tf.width = MAX(1, tf.width >> 1); + tf.height = MAX(1, tf.height >> 1); + } + } + + { + // and finally an FB for just our base weights + Vector<RID> fb; + fb.push_back(rb->weight_buffers[0].weight); + rb->base_weight_fb = RD::get_singleton()->framebuffer_create(fb); + } + } +} + +void RendererSceneRenderRD::_allocate_depth_backbuffer_textures(RenderBuffers *rb) { + ERR_FAIL_COND(!rb->depth_back_texture.is_null()); + + { + RD::TextureFormat tf; + if (rb->view_count > 1) { + tf.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; + } + // We're not using this as a depth stencil, just copying our data into this. May need to look into using a different format on mobile, maybe R16? + tf.format = RD::DATA_FORMAT_R32_SFLOAT; + + tf.width = rb->width; + tf.height = rb->height; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT; + tf.array_layers = rb->view_count; // create a layer for every view + + tf.usage_bits |= RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + tf.usage_bits |= RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; // set this as color attachment because we're copying data into it, it's not actually used as a depth buffer + + rb->depth_back_texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); + } + + if (!_render_buffers_can_be_storage()) { + // create framebuffer so we can write into this... + + Vector<RID> fb; + fb.push_back(rb->depth_back_texture); + + rb->depth_back_fb = RD::get_singleton()->framebuffer_create(fb, RD::INVALID_ID, rb->view_count); + } } void RendererSceneRenderRD::_allocate_luminance_textures(RenderBuffers *rb) { @@ -1435,26 +1636,48 @@ void RendererSceneRenderRD::_allocate_luminance_textures(RenderBuffers *rb) { tf.format = RD::DATA_FORMAT_R32_SFLOAT; tf.width = w; tf.height = h; - tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; bool final = w == 1 && h == 1; - if (final) { - tf.usage_bits |= RD::TEXTURE_USAGE_SAMPLING_BIT; + if (_render_buffers_can_be_storage()) { + tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; + if (final) { + tf.usage_bits |= RD::TEXTURE_USAGE_SAMPLING_BIT; + } + } else { + tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT; } RID texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); rb->luminance.reduce.push_back(texture); + if (!_render_buffers_can_be_storage()) { + Vector<RID> fb; + fb.push_back(texture); + + rb->luminance.fb.push_back(RD::get_singleton()->framebuffer_create(fb)); + } if (final) { rb->luminance.current = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + if (!_render_buffers_can_be_storage()) { + Vector<RID> fb; + fb.push_back(rb->luminance.current); + + rb->luminance.current_fb = RD::get_singleton()->framebuffer_create(fb); + } break; } } } void RendererSceneRenderRD::_free_render_buffer_data(RenderBuffers *rb) { + if (rb->texture_fb.is_valid()) { + RD::get_singleton()->free(rb->texture_fb); + rb->texture_fb = RID(); + } + if (rb->texture.is_valid()) { RD::get_singleton()->free(rb->texture); rb->texture = RID(); @@ -1465,20 +1688,54 @@ void RendererSceneRenderRD::_free_render_buffer_data(RenderBuffers *rb) { rb->depth_texture = RID(); } + if (rb->depth_back_fb.is_valid()) { + RD::get_singleton()->free(rb->depth_back_fb); + rb->depth_back_fb = RID(); + } + + if (rb->depth_back_texture.is_valid()) { + RD::get_singleton()->free(rb->depth_back_texture); + rb->depth_back_texture = RID(); + } + for (int i = 0; i < 2; i++) { + for (int m = 0; m < rb->blur[i].mipmaps.size(); m++) { + // do we free the texture slice here? or is it enough to free the main texture? + + // do free the mobile extra stuff + if (rb->blur[i].mipmaps[m].fb.is_valid()) { + RD::get_singleton()->free(rb->blur[i].mipmaps[m].fb); + } + if (rb->blur[i].mipmaps[m].half_fb.is_valid()) { + RD::get_singleton()->free(rb->blur[i].mipmaps[m].half_fb); + } + if (rb->blur[i].mipmaps[m].half_texture.is_valid()) { + RD::get_singleton()->free(rb->blur[i].mipmaps[m].half_texture); + } + } + rb->blur[i].mipmaps.clear(); + if (rb->blur[i].texture.is_valid()) { RD::get_singleton()->free(rb->blur[i].texture); rb->blur[i].texture = RID(); - rb->blur[i].mipmaps.clear(); } } + for (int i = 0; i < rb->luminance.fb.size(); i++) { + RD::get_singleton()->free(rb->luminance.fb[i]); + } + rb->luminance.fb.clear(); + for (int i = 0; i < rb->luminance.reduce.size(); i++) { RD::get_singleton()->free(rb->luminance.reduce[i]); } - rb->luminance.reduce.clear(); + if (rb->luminance.current_fb.is_valid()) { + RD::get_singleton()->free(rb->luminance.current_fb); + rb->luminance.current_fb = RID(); + } + if (rb->luminance.current.is_valid()) { RD::get_singleton()->free(rb->luminance.current); rb->luminance.current = RID(); @@ -1527,7 +1784,7 @@ void RendererSceneRenderRD::_free_render_buffer_data(RenderBuffers *rb) { } void RendererSceneRenderRD::_process_sss(RID p_render_buffers, const CameraMatrix &p_camera) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); bool can_use_effects = rb->width >= 8 && rb->height >= 8; @@ -1545,7 +1802,7 @@ void RendererSceneRenderRD::_process_sss(RID p_render_buffers, const CameraMatri } void RendererSceneRenderRD::_process_ssr(RID p_render_buffers, RID p_dest_framebuffer, RID p_normal_buffer, RID p_specular_buffer, RID p_metallic, const Color &p_metallic_mask, RID p_environment, const CameraMatrix &p_projection, bool p_use_additive) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); bool can_use_effects = rb->width >= 8 && rb->height >= 8; @@ -1556,7 +1813,7 @@ void RendererSceneRenderRD::_process_ssr(RID p_render_buffers, RID p_dest_frameb return; } - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_environment); ERR_FAIL_COND(!env); ERR_FAIL_COND(!env->ssr_enabled); @@ -1597,10 +1854,10 @@ void RendererSceneRenderRD::_process_ssr(RID p_render_buffers, RID p_dest_frameb } void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environment, RID p_normal_buffer, const CameraMatrix &p_projection) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_environment); ERR_FAIL_COND(!env); RENDER_TIMESTAMP("Process SSAO"); @@ -1741,26 +1998,111 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen storage->get_effects()->generate_ssao(rb->depth_texture, p_normal_buffer, rb->ssao.depth, rb->ssao.depth_slices, rb->ssao.ao_deinterleaved, rb->ssao.ao_deinterleaved_slices, rb->ssao.ao_pong, rb->ssao.ao_pong_slices, rb->ssao.ao_final, rb->ssao.importance_map[0], rb->ssao.importance_map[1], p_projection, settings, uniform_sets_are_invalid, rb->ssao.downsample_uniform_set, rb->ssao.gather_uniform_set, rb->ssao.importance_map_uniform_set); } +void RendererSceneRenderRD::_render_buffers_copy_screen_texture(const RenderDataRD *p_render_data) { + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + + RD::get_singleton()->draw_command_begin_label("Copy screen texture"); + + if (rb->blur[0].texture.is_null()) { + _allocate_blur_textures(rb); + } + + // @TODO IMPLEMENT MULTIVIEW, all effects need to support stereo buffers or effects are only applied to the left eye + + bool can_use_storage = _render_buffers_can_be_storage(); + + if (can_use_storage) { + storage->get_effects()->copy_to_rect(rb->texture, rb->blur[0].mipmaps[0].texture, Rect2i(0, 0, rb->width, rb->height)); + for (int i = 1; i < rb->blur[0].mipmaps.size(); i++) { + storage->get_effects()->make_mipmap(rb->blur[0].mipmaps[i - 1].texture, rb->blur[0].mipmaps[i].texture, Size2i(rb->blur[0].mipmaps[i].width, rb->blur[0].mipmaps[i].height)); + } + } else { + storage->get_effects()->copy_to_fb_rect(rb->texture, rb->blur[0].mipmaps[0].fb, Rect2i(0, 0, rb->width, rb->height)); + for (int i = 1; i < rb->blur[0].mipmaps.size(); i++) { + storage->get_effects()->make_mipmap_raster(rb->blur[0].mipmaps[i - 1].texture, rb->blur[0].mipmaps[i].fb, Size2i(rb->blur[0].mipmaps[i].width, rb->blur[0].mipmaps[i].height)); + } + } + + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneRenderRD::_render_buffers_copy_depth_texture(const RenderDataRD *p_render_data) { + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + + RD::get_singleton()->draw_command_begin_label("Copy depth texture"); + + if (rb->depth_back_texture.is_null()) { + _allocate_depth_backbuffer_textures(rb); + } + + // @TODO IMPLEMENT MULTIVIEW, all effects need to support stereo buffers or effects are only applied to the left eye + + bool can_use_storage = _render_buffers_can_be_storage(); + + if (can_use_storage) { + storage->get_effects()->copy_to_rect(rb->depth_texture, rb->depth_back_texture, Rect2i(0, 0, rb->width, rb->height)); + } else { + storage->get_effects()->copy_to_fb_rect(rb->depth_texture, rb->depth_back_fb, Rect2i(0, 0, rb->width, rb->height)); + } + + RD::get_singleton()->draw_command_end_label(); +} + void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const RenderDataRD *p_render_data) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); ERR_FAIL_COND(!rb); - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_render_data->environment); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_render_data->environment); //glow (if enabled) - CameraEffects *camfx = camera_effects_owner.getornull(p_render_data->camera_effects); + CameraEffects *camfx = camera_effects_owner.get_or_null(p_render_data->camera_effects); bool can_use_effects = rb->width >= 8 && rb->height >= 8; + bool can_use_storage = _render_buffers_can_be_storage(); + + // @TODO IMPLEMENT MULTIVIEW, all effects need to support stereo buffers or effects are only applied to the left eye if (can_use_effects && camfx && (camfx->dof_blur_near_enabled || camfx->dof_blur_far_enabled) && camfx->dof_blur_amount > 0.0) { + RD::get_singleton()->draw_command_begin_label("DOF"); if (rb->blur[0].texture.is_null()) { _allocate_blur_textures(rb); } + EffectsRD::BokehBuffers buffers; + + // textures we use + buffers.base_texture_size = Size2i(rb->width, rb->height); + buffers.base_texture = rb->texture; + buffers.depth_texture = rb->depth_texture; + buffers.secondary_texture = rb->blur[0].mipmaps[0].texture; + buffers.half_texture[0] = rb->blur[1].mipmaps[0].texture; + buffers.half_texture[1] = rb->blur[0].mipmaps[1].texture; + float bokeh_size = camfx->dof_blur_amount * 64.0; - storage->get_effects()->bokeh_dof(rb->texture, rb->depth_texture, Size2i(rb->width, rb->height), rb->blur[0].mipmaps[0].texture, rb->blur[1].mipmaps[0].texture, rb->blur[0].mipmaps[1].texture, camfx->dof_blur_far_enabled, camfx->dof_blur_far_distance, camfx->dof_blur_far_transition, camfx->dof_blur_near_enabled, camfx->dof_blur_near_distance, camfx->dof_blur_near_transition, bokeh_size, dof_blur_bokeh_shape, dof_blur_quality, dof_blur_use_jitter, p_render_data->z_near, p_render_data->z_far, p_render_data->cam_ortogonal); + if (can_use_storage) { + storage->get_effects()->bokeh_dof(buffers, camfx->dof_blur_far_enabled, camfx->dof_blur_far_distance, camfx->dof_blur_far_transition, camfx->dof_blur_near_enabled, camfx->dof_blur_near_distance, camfx->dof_blur_near_transition, bokeh_size, dof_blur_bokeh_shape, dof_blur_quality, dof_blur_use_jitter, p_render_data->z_near, p_render_data->z_far, p_render_data->cam_ortogonal); + } else { + // set framebuffers + buffers.base_fb = rb->texture_fb; + buffers.secondary_fb = rb->weight_buffers[1].fb; + buffers.half_fb[0] = rb->weight_buffers[2].fb; + buffers.half_fb[1] = rb->weight_buffers[3].fb; + buffers.weight_texture[0] = rb->weight_buffers[0].weight; + buffers.weight_texture[1] = rb->weight_buffers[1].weight; + buffers.weight_texture[2] = rb->weight_buffers[2].weight; + buffers.weight_texture[3] = rb->weight_buffers[3].weight; + + // set weight buffers + buffers.base_weight_fb = rb->base_weight_fb; + + storage->get_effects()->bokeh_dof_raster(buffers, camfx->dof_blur_far_enabled, camfx->dof_blur_far_distance, camfx->dof_blur_far_transition, camfx->dof_blur_near_enabled, camfx->dof_blur_near_distance, camfx->dof_blur_near_transition, bokeh_size, dof_blur_bokeh_shape, dof_blur_quality, p_render_data->z_near, p_render_data->z_far, p_render_data->cam_ortogonal); + } + RD::get_singleton()->draw_command_end_label(); } if (can_use_effects && env && env->auto_exposure) { + RD::get_singleton()->draw_command_begin_label("Auto exposure"); if (rb->luminance.current.is_null()) { _allocate_luminance_textures(rb); } @@ -1769,16 +2111,26 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende rb->auto_exposure_version = env->auto_exposure_version; double step = env->auto_exp_speed * time_step; - storage->get_effects()->luminance_reduction(rb->texture, Size2i(rb->width, rb->height), rb->luminance.reduce, rb->luminance.current, env->min_luminance, env->max_luminance, step, set_immediate); - + if (can_use_storage) { + storage->get_effects()->luminance_reduction(rb->texture, Size2i(rb->width, rb->height), rb->luminance.reduce, rb->luminance.current, env->min_luminance, env->max_luminance, step, set_immediate); + } else { + storage->get_effects()->luminance_reduction_raster(rb->texture, Size2i(rb->width, rb->height), rb->luminance.reduce, rb->luminance.fb, rb->luminance.current, env->min_luminance, env->max_luminance, step, set_immediate); + } //swap final reduce with prev luminance SWAP(rb->luminance.current, rb->luminance.reduce.write[rb->luminance.reduce.size() - 1]); + if (!can_use_storage) { + SWAP(rb->luminance.current_fb, rb->luminance.fb.write[rb->luminance.fb.size() - 1]); + } + RenderingServerDefault::redraw_request(); //redraw all the time if auto exposure rendering is on + RD::get_singleton()->draw_command_end_label(); } int max_glow_level = -1; if (can_use_effects && env && env->glow_enabled) { + RD::get_singleton()->draw_command_begin_label("Gaussian Glow"); + /* see that blur textures are allocated */ if (rb->blur[1].texture.is_null()) { @@ -1804,14 +2156,26 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende if (env->auto_exposure && rb->luminance.current.is_valid()) { luminance_texture = rb->luminance.current; } - storage->get_effects()->gaussian_glow(rb->texture, rb->blur[1].mipmaps[i].texture, Size2i(vp_w, vp_h), env->glow_strength, glow_high_quality, true, env->glow_hdr_luminance_cap, env->exposure, env->glow_bloom, env->glow_hdr_bleed_threshold, env->glow_hdr_bleed_scale, luminance_texture, env->auto_exp_scale); + if (can_use_storage) { + storage->get_effects()->gaussian_glow(rb->texture, rb->blur[1].mipmaps[i].texture, Size2i(vp_w, vp_h), env->glow_strength, glow_high_quality, true, env->glow_hdr_luminance_cap, env->exposure, env->glow_bloom, env->glow_hdr_bleed_threshold, env->glow_hdr_bleed_scale, luminance_texture, env->auto_exp_scale); + } else { + storage->get_effects()->gaussian_glow_raster(rb->texture, rb->blur[1].mipmaps[i].half_fb, rb->blur[1].mipmaps[i].half_texture, rb->blur[1].mipmaps[i].fb, Size2i(vp_w, vp_h), env->glow_strength, glow_high_quality, true, env->glow_hdr_luminance_cap, env->exposure, env->glow_bloom, env->glow_hdr_bleed_threshold, env->glow_hdr_bleed_scale, luminance_texture, env->auto_exp_scale); + } } else { - storage->get_effects()->gaussian_glow(rb->blur[1].mipmaps[i - 1].texture, rb->blur[1].mipmaps[i].texture, Size2i(vp_w, vp_h), env->glow_strength, glow_high_quality); + if (can_use_storage) { + storage->get_effects()->gaussian_glow(rb->blur[1].mipmaps[i - 1].texture, rb->blur[1].mipmaps[i].texture, Size2i(vp_w, vp_h), env->glow_strength, glow_high_quality); + } else { + storage->get_effects()->gaussian_glow_raster(rb->blur[1].mipmaps[i - 1].texture, rb->blur[1].mipmaps[i].half_fb, rb->blur[1].mipmaps[i].half_texture, rb->blur[1].mipmaps[i].fb, Vector2(1.0 / vp_w, 1.0 / vp_h), env->glow_strength, glow_high_quality); + } } } + + RD::get_singleton()->draw_command_end_label(); } { + RD::get_singleton()->draw_command_begin_label("Tonemap"); + //tonemap EffectsRD::TonemapSettings tonemap; @@ -1867,18 +2231,90 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende } } + tonemap.luminance_multiplier = _render_buffers_get_luminance_multiplier(); tonemap.view_count = p_render_data->view_count; storage->get_effects()->tonemapper(rb->texture, storage->render_target_get_rd_framebuffer(rb->render_target), tonemap); + + RD::get_singleton()->draw_command_end_label(); + } + + storage->render_target_disable_clear_request(rb->render_target); +} + +void RendererSceneRenderRD::_post_process_subpass(RID p_source_texture, RID p_framebuffer, const RenderDataRD *p_render_data) { + RD::get_singleton()->draw_command_begin_label("Post Process Subpass"); + + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_render_data->environment); + + bool can_use_effects = rb->width >= 8 && rb->height >= 8; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + + EffectsRD::TonemapSettings tonemap; + + if (env) { + tonemap.tonemap_mode = env->tone_mapper; + tonemap.exposure = env->exposure; + tonemap.white = env->white; } + // We don't support glow or auto exposure here, if they are needed, don't use subpasses! + // The problem is that we need to use the result so far and process them before we can + // apply this to our results. + if (can_use_effects && env && env->glow_enabled) { + ERR_FAIL_MSG("Glow is not supported when using subpasses."); + } + if (can_use_effects && env && env->auto_exposure) { + ERR_FAIL_MSG("Glow is not supported when using subpasses."); + } + + tonemap.use_glow = false; + tonemap.glow_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); + tonemap.use_auto_exposure = false; + tonemap.exposure_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); + + tonemap.use_color_correction = false; + tonemap.use_1d_color_correction = false; + tonemap.color_correction_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); + + if (can_use_effects && env) { + tonemap.use_bcs = env->adjustments_enabled; + tonemap.brightness = env->adjustments_brightness; + tonemap.contrast = env->adjustments_contrast; + tonemap.saturation = env->adjustments_saturation; + if (env->adjustments_enabled && env->color_correction.is_valid()) { + tonemap.use_color_correction = true; + tonemap.use_1d_color_correction = env->use_1d_color_correction; + tonemap.color_correction_texture = storage->texture_get_rd_texture(env->color_correction); + } + } + + tonemap.use_debanding = rb->use_debanding; + tonemap.texture_size = Vector2i(rb->width, rb->height); + + tonemap.luminance_multiplier = _render_buffers_get_luminance_multiplier(); + tonemap.view_count = p_render_data->view_count; + + storage->get_effects()->tonemapper(draw_list, p_source_texture, RD::get_singleton()->framebuffer_get_format(p_framebuffer), tonemap); + + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneRenderRD::_disable_clear_request(const RenderDataRD *p_render_data) { + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + storage->render_target_disable_clear_request(rb->render_target); } void RendererSceneRenderRD::_render_buffers_debug_draw(RID p_render_buffers, RID p_shadow_atlas, RID p_occlusion_buffer) { EffectsRD *effects = storage->get_effects(); - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SHADOW_ATLAS) { @@ -1944,7 +2380,7 @@ void RendererSceneRenderRD::_render_buffers_debug_draw(RID p_render_buffers, RID } void RendererSceneRenderRD::environment_set_adjustment(RID p_env, bool p_enable, float p_brightness, float p_contrast, float p_saturation, bool p_use_1d_color_correction, RID p_color_correction) { - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_env); ERR_FAIL_COND(!env); env->adjustments_enabled = p_enable; @@ -1956,7 +2392,7 @@ void RendererSceneRenderRD::environment_set_adjustment(RID p_env, bool p_enable, } RID RendererSceneRenderRD::render_buffers_get_back_buffer_texture(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); if (!rb->blur[0].texture.is_valid()) { return RID(); //not valid at the moment @@ -1964,15 +2400,31 @@ RID RendererSceneRenderRD::render_buffers_get_back_buffer_texture(RID p_render_b return rb->blur[0].texture; } +RID RendererSceneRenderRD::render_buffers_get_back_depth_texture(RID p_render_buffers) { + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); + ERR_FAIL_COND_V(!rb, RID()); + if (!rb->depth_back_texture.is_valid()) { + return RID(); //not valid at the moment + } + return rb->depth_back_texture; +} + +RID RendererSceneRenderRD::render_buffers_get_depth_texture(RID p_render_buffers) { + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); + ERR_FAIL_COND_V(!rb, RID()); + + return rb->depth_texture; +} + RID RendererSceneRenderRD::render_buffers_get_ao_texture(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); return rb->ssao.ao_final; } RID RendererSceneRenderRD::render_buffers_get_voxel_gi_buffer(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); if (rb->gi.voxel_gi_buffer.is_null()) { rb->gi.voxel_gi_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(RendererSceneGIRD::VoxelGIData) * RendererSceneGIRD::MAX_VOXEL_GI_INSTANCES); @@ -1985,31 +2437,31 @@ RID RendererSceneRenderRD::render_buffers_get_default_voxel_gi_buffer() { } RID RendererSceneRenderRD::render_buffers_get_gi_ambient_texture(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); return rb->ambient_buffer; } RID RendererSceneRenderRD::render_buffers_get_gi_reflection_texture(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); return rb->reflection_buffer; } uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_count(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0); ERR_FAIL_COND_V(!rb->sdfgi, 0); return rb->sdfgi->cascades.size(); } bool RendererSceneRenderRD::render_buffers_is_sdfgi_enabled(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, false); return rb->sdfgi != nullptr; } RID RendererSceneRenderRD::render_buffers_get_sdfgi_irradiance_probes(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); ERR_FAIL_COND_V(!rb->sdfgi, RID()); @@ -2017,7 +2469,7 @@ RID RendererSceneRenderRD::render_buffers_get_sdfgi_irradiance_probes(RID p_rend } Vector3 RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_offset(RID p_render_buffers, uint32_t p_cascade) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, Vector3()); ERR_FAIL_COND_V(!rb->sdfgi, Vector3()); ERR_FAIL_UNSIGNED_INDEX_V(p_cascade, rb->sdfgi->cascades.size(), Vector3()); @@ -2026,7 +2478,7 @@ Vector3 RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_offset(RID p_ren } Vector3i RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_offset(RID p_render_buffers, uint32_t p_cascade) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, Vector3i()); ERR_FAIL_COND_V(!rb->sdfgi, Vector3i()); ERR_FAIL_UNSIGNED_INDEX_V(p_cascade, rb->sdfgi->cascades.size(), Vector3i()); @@ -2036,14 +2488,14 @@ Vector3i RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_offset(RI } float RendererSceneRenderRD::render_buffers_get_sdfgi_normal_bias(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0); ERR_FAIL_COND_V(!rb->sdfgi, 0); return rb->sdfgi->normal_bias; } float RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_size(RID p_render_buffers, uint32_t p_cascade) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0); ERR_FAIL_COND_V(!rb->sdfgi, 0); ERR_FAIL_UNSIGNED_INDEX_V(p_cascade, rb->sdfgi->cascades.size(), 0); @@ -2051,7 +2503,7 @@ float RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_size(RID p_r return float(rb->sdfgi->cascade_size) * rb->sdfgi->cascades[p_cascade].cell_size / float(rb->sdfgi->probe_axis_count - 1); } uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_count(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0); ERR_FAIL_COND_V(!rb->sdfgi, 0); @@ -2059,7 +2511,7 @@ uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_count(RID } uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_size(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0); ERR_FAIL_COND_V(!rb->sdfgi, 0); @@ -2067,7 +2519,7 @@ uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_size(RID p_rend } bool RendererSceneRenderRD::render_buffers_is_sdfgi_using_occlusion(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, false); ERR_FAIL_COND_V(!rb->sdfgi, false); @@ -2075,14 +2527,14 @@ bool RendererSceneRenderRD::render_buffers_is_sdfgi_using_occlusion(RID p_render } float RendererSceneRenderRD::render_buffers_get_sdfgi_energy(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, 0.0); ERR_FAIL_COND_V(!rb->sdfgi, 0.0); return rb->sdfgi->energy; } RID RendererSceneRenderRD::render_buffers_get_sdfgi_occlusion_texture(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); ERR_FAIL_COND_V(!rb->sdfgi, RID()); @@ -2090,20 +2542,20 @@ RID RendererSceneRenderRD::render_buffers_get_sdfgi_occlusion_texture(RID p_rend } bool RendererSceneRenderRD::render_buffers_has_volumetric_fog(RID p_render_buffers) const { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, false); return rb->volumetric_fog != nullptr; } RID RendererSceneRenderRD::render_buffers_get_volumetric_fog_texture(RID p_render_buffers) { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb || !rb->volumetric_fog, RID()); return rb->volumetric_fog->fog_map; } RID RendererSceneRenderRD::render_buffers_get_volumetric_fog_sky_uniform_set(RID p_render_buffers) { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); if (!rb->volumetric_fog) { @@ -2114,16 +2566,20 @@ RID RendererSceneRenderRD::render_buffers_get_volumetric_fog_sky_uniform_set(RID } float RendererSceneRenderRD::render_buffers_get_volumetric_fog_end(RID p_render_buffers) { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb || !rb->volumetric_fog, 0); return rb->volumetric_fog->length; } float RendererSceneRenderRD::render_buffers_get_volumetric_fog_detail_spread(RID p_render_buffers) { - const RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + const RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb || !rb->volumetric_fog, 0); return rb->volumetric_fog->spread; } +float RendererSceneRenderRD::_render_buffers_get_luminance_multiplier() { + return 1.0; +} + RD::DataFormat RendererSceneRenderRD::_render_buffers_get_color_format() { return RD::DATA_FORMAT_R16G16B16A16_SFLOAT; } @@ -2133,9 +2589,11 @@ bool RendererSceneRenderRD::_render_buffers_can_be_storage() { } void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p_render_target, int p_width, int p_height, RS::ViewportMSAA p_msaa, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_debanding, uint32_t p_view_count) { - ERR_FAIL_COND_MSG(p_view_count == 0, "Must have atleast 1 view"); + ERR_FAIL_COND_MSG(p_view_count == 0, "Must have at least 1 view"); - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); + + // Should we add an overrule per viewport? rb->width = p_width; rb->height = p_height; rb->render_target = p_render_target; @@ -2162,12 +2620,11 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p tf.width = rb->width; tf.height = rb->height; tf.array_layers = rb->view_count; // create a layer for every view - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0); + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0) | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; if (rb->msaa != RS::VIEWPORT_MSAA_DISABLED) { - tf.usage_bits |= RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0); - } else { - tf.usage_bits |= RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tf.usage_bits |= RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; } + tf.usage_bits |= RD::TEXTURE_USAGE_INPUT_ATTACHMENT_BIT; // only needed when using subpasses in the mobile renderer rb->texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); } @@ -2183,8 +2640,8 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p tf.format = RD::DATA_FORMAT_R32_SFLOAT; } - tf.width = p_width; - tf.height = p_height; + tf.width = rb->width; + tf.height = rb->height; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT; tf.array_layers = rb->view_count; // create a layer for every view @@ -2197,10 +2654,19 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p rb->depth_texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); } - rb->data->configure(rb->texture, rb->depth_texture, p_width, p_height, p_msaa, p_view_count); + if (!_render_buffers_can_be_storage()) { + // ONLY USED ON MOBILE RENDERER, ONLY USED FOR POST EFFECTS! + Vector<RID> fb; + fb.push_back(rb->texture); + + rb->texture_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, rb->view_count); + } + + RID target_texture = storage->render_target_get_rd_texture(rb->render_target); + rb->data->configure(rb->texture, rb->depth_texture, target_texture, rb->width, rb->height, p_msaa, p_view_count); if (is_clustered_enabled()) { - rb->cluster_builder->setup(Size2i(p_width, p_height), max_cluster_elements, rb->depth_texture, storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED), rb->texture); + rb->cluster_builder->setup(Size2i(rb->width, rb->height), max_cluster_elements, rb->depth_texture, storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED), rb->texture); } } @@ -2329,7 +2795,7 @@ bool RendererSceneRenderRD::is_using_radiance_cubemap_array() const { } RendererSceneRenderRD::RenderBufferData *RendererSceneRenderRD::render_buffers_get_data(RID p_render_buffers) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND_V(!rb, nullptr); return rb->data; } @@ -2342,7 +2808,7 @@ void RendererSceneRenderRD::_setup_reflections(const PagedArray<RID> &p_reflecti break; } - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_reflections[i]); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_reflections[i]); if (!rpi) { continue; } @@ -2420,7 +2886,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const r_positional_light_count = 0; sky.sky_scene_state.ubo.directional_light_count = 0; - Plane camera_plane(p_camera_transform.origin, -p_camera_transform.basis.get_axis(Vector3::AXIS_Z).normalized()); + Plane camera_plane(-p_camera_transform.basis.get_axis(Vector3::AXIS_Z).normalized(), p_camera_transform.origin); cluster.omni_light_count = 0; cluster.spot_light_count = 0; @@ -2428,7 +2894,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const r_directional_light_soft_shadows = false; for (int i = 0; i < (int)p_lights.size(); i++) { - LightInstance *li = light_instance_owner.getornull(p_lights[i]); + LightInstance *li = light_instance_owner.get_or_null(p_lights[i]); if (!li) { continue; } @@ -2560,7 +3026,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const RS::LightDirectionalShadowMode smode = storage->light_directional_get_shadow_mode(base); int limit = smode == RS::LIGHT_DIRECTIONAL_SHADOW_ORTHOGONAL ? 0 : (smode == RS::LIGHT_DIRECTIONAL_SHADOW_PARALLEL_2_SPLITS ? 1 : 3); - light_data.blend_splits = storage->light_directional_get_blend_splits(base); + light_data.blend_splits = (smode != RS::LIGHT_DIRECTIONAL_SHADOW_ORTHOGONAL) && storage->light_directional_get_blend_splits(base); for (int j = 0; j < 4; j++) { Rect2 atlas_rect = li->shadow_transform[j].atlas_rect; CameraMatrix matrix = li->shadow_transform[j].camera; @@ -2576,7 +3042,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const CameraMatrix shadow_mtx = rectm * bias * matrix * modelview; light_data.shadow_split_offsets[j] = split; float bias_scale = li->shadow_transform[j].bias_scale; - light_data.shadow_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * bias_scale; + light_data.shadow_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) / 100.0 * bias_scale; light_data.shadow_normal_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * li->shadow_transform[j].shadow_texel_size; light_data.shadow_transmittance_bias[j] = storage->light_get_transmittance_bias(base) * bias_scale; light_data.shadow_z_range[j] = li->shadow_transform[j].farplane; @@ -2657,7 +3123,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const ShadowAtlas *shadow_atlas = nullptr; if (p_shadow_atlas.is_valid() && p_using_shadows) { - shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + shadow_atlas = shadow_atlas_owner.get_or_null(p_shadow_atlas); } bool using_forward_ids = _uses_forward_ids(); @@ -2746,22 +3212,19 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const light_data.shadow_enabled = true; - if (type == RS::LIGHT_SPOT) { - light_data.shadow_bias = (storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0); - float shadow_texel_size = Math::tan(Math::deg2rad(spot_angle)) * radius * 2.0; - shadow_texel_size *= light_instance_get_shadow_texel_size(li->self, p_shadow_atlas); - - light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size; + float shadow_texel_size = light_instance_get_shadow_texel_size(li->self, p_shadow_atlas); + light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size * 10.0; + if (type == RS::LIGHT_SPOT) { + light_data.shadow_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) / 100.0; } else { //omni - light_data.shadow_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0; - float shadow_texel_size = light_instance_get_shadow_texel_size(li->self, p_shadow_atlas); - light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size * 2.0; // applied in -1 .. 1 space + light_data.shadow_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS); } light_data.transmittance_bias = storage->light_get_transmittance_bias(base); - Rect2 rect = light_instance_get_shadow_atlas_rect(li->self, p_shadow_atlas); + Vector2i omni_offset; + Rect2 rect = light_instance_get_shadow_atlas_rect(li->self, p_shadow_atlas, omni_offset); light_data.atlas_rect[0] = rect.position.x; light_data.atlas_rect[1] = rect.position.y; @@ -2772,7 +3235,6 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const light_data.shadow_volumetric_fog_fade = 1.0 / storage->light_get_shadow_volumetric_fog_fade(base); if (type == RS::LIGHT_OMNI) { - light_data.atlas_rect[3] *= 0.5; //one paraboloid on top of another Transform3D proj = (inverse_transform * light_transform).inverse(); RendererStorageRD::store_transform(proj, light_data.shadow_matrix); @@ -2784,6 +3246,8 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const light_data.soft_shadow_scale *= shadows_quality_radius_get(); // Only use quality radius for PCF } + light_data.direction[0] = omni_offset.x * float(rect.size.width); + light_data.direction[1] = omni_offset.y * float(rect.size.height); } else if (type == RS::LIGHT_SPOT) { Transform3D modelview = (inverse_transform * light_transform).inverse(); CameraMatrix bias; @@ -2842,7 +3306,7 @@ void RendererSceneRenderRD::_setup_decals(const PagedArray<RID> &p_decals, const break; } - DecalInstance *di = decal_instance_owner.getornull(p_decals[i]); + DecalInstance *di = decal_instance_owner.get_or_null(p_decals[i]); if (!di) { continue; } @@ -2902,7 +3366,7 @@ void RendererSceneRenderRD::_setup_decals(const PagedArray<RID> &p_decals, const Vector3 decal_extents = storage->decal_get_extents(decal); Transform3D scale_xform; - scale_xform.basis.scale(Vector3(decal_extents.x, decal_extents.y, decal_extents.z)); + scale_xform.basis.scale(decal_extents); Transform3D to_decal_xform = (p_camera_inverse_xform * di->transform * scale_xform * uv_xform).affine_inverse(); RendererStorageRD::store_transform(to_decal_xform, dd.xform); @@ -3025,9 +3489,9 @@ void RendererSceneRenderRD::_volumetric_fog_erase(RenderBuffers *rb) { void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_environment, const CameraMatrix &p_cam_projection, const Transform3D &p_cam_transform, RID p_shadow_atlas, int p_directional_light_count, bool p_use_directional_shadows, int p_positional_light_count, int p_voxel_gi_count) { ERR_FAIL_COND(!is_clustered_enabled()); // can't use volumetric fog without clustered - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_environment); float ratio = float(rb->width) / float((rb->width + rb->height) / 2); uint32_t target_width = uint32_t(float(volumetric_fog_size) * ratio); @@ -3096,7 +3560,7 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 1; - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_shadow_atlas); if (shadow_atlas == nullptr || shadow_atlas->depth.is_null()) { u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK)); } else { @@ -3418,7 +3882,7 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e bool RendererSceneRenderRD::_needs_post_prepass_render(RenderDataRD *p_render_data, bool p_use_gi) { if (p_render_data->render_buffers.is_valid()) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); if (rb->sdfgi != nullptr) { return true; } @@ -3429,14 +3893,14 @@ bool RendererSceneRenderRD::_needs_post_prepass_render(RenderDataRD *p_render_da void RendererSceneRenderRD::_post_prepass_render(RenderDataRD *p_render_data, bool p_use_gi) { if (p_render_data->render_buffers.is_valid()) { if (p_use_gi) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); ERR_FAIL_COND(rb == nullptr); if (rb->sdfgi == nullptr) { return; } - RendererSceneEnvironmentRD *env = environment_owner.getornull(p_render_data->environment); - rb->sdfgi->update_probes(env, sky.sky_owner.getornull(env->sky)); + RendererSceneEnvironmentRD *env = environment_owner.get_or_null(p_render_data->environment); + rb->sdfgi->update_probes(env, sky.sky_owner.get_or_null(env->sky)); } } } @@ -3453,7 +3917,7 @@ void RendererSceneRenderRD::_pre_opaque_render(RenderDataRD *p_render_data, bool // Render shadows while GI is rendering, due to how barriers are handled, this should happen at the same time if (p_render_data->render_buffers.is_valid() && p_use_gi) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_render_data->render_buffers); ERR_FAIL_COND(rb == nullptr); if (rb->sdfgi != nullptr) { rb->sdfgi->store_probes(); @@ -3464,11 +3928,11 @@ void RendererSceneRenderRD::_pre_opaque_render(RenderDataRD *p_render_data, bool render_state.shadows.clear(); render_state.directional_shadows.clear(); - Plane camera_plane(p_render_data->cam_transform.origin, -p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z)); + Plane camera_plane(-p_render_data->cam_transform.basis.get_axis(Vector3::AXIS_Z), p_render_data->cam_transform.origin); float lod_distance_multiplier = p_render_data->cam_projection.get_lod_multiplier(); { for (int i = 0; i < render_state.render_shadow_count; i++) { - LightInstance *li = light_instance_owner.getornull(render_state.render_shadows[i].light); + LightInstance *li = light_instance_owner.get_or_null(render_state.render_shadows[i].light); if (storage->light_get_type(li->light) == RS::LIGHT_DIRECTIONAL) { render_state.directional_shadows.push_back(i); @@ -3588,7 +4052,7 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData // getting this here now so we can direct call a bunch of things more easily RenderBuffers *rb = nullptr; if (p_render_buffers.is_valid()) { - rb = render_buffers_owner.getornull(p_render_buffers); + rb = render_buffers_owner.get_or_null(p_render_buffers); ERR_FAIL_COND(!rb); } @@ -3626,7 +4090,7 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData // this should be the same for all cameras.. render_data.lod_distance_multiplier = p_camera_data->main_projection.get_lod_multiplier(); - render_data.lod_camera_plane = Plane(p_camera_data->main_transform.get_origin(), -p_camera_data->main_transform.basis.get_axis(Vector3::AXIS_Z)); + render_data.lod_camera_plane = Plane(-p_camera_data->main_transform.basis.get_axis(Vector3::AXIS_Z), p_camera_data->main_transform.get_origin()); if (get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_DISABLE_LOD) { render_data.screen_lod_threshold = 0.0; @@ -3670,7 +4134,7 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData //assign render indices to voxel_gi_instances if (is_dynamic_gi_supported()) { for (uint32_t i = 0; i < (uint32_t)p_voxel_gi_instances.size(); i++) { - RendererSceneGIRD::VoxelGIInstance *voxel_gi_inst = gi.voxel_gi_instance_owner.getornull(p_voxel_gi_instances[i]); + RendererSceneGIRD::VoxelGIInstance *voxel_gi_inst = gi.voxel_gi_instance_owner.get_or_null(p_voxel_gi_instances[i]); if (voxel_gi_inst) { voxel_gi_inst->render_index = i; } @@ -3681,8 +4145,8 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData // render_data.render_buffers == p_render_buffers so we can use our already retrieved rb current_cluster_builder = rb->cluster_builder; } else if (reflection_probe_instance_owner.owns(render_data.reflection_probe)) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(render_data.reflection_probe); - ReflectionAtlas *ra = reflection_atlas_owner.getornull(rpi->atlas); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(render_data.reflection_probe); + ReflectionAtlas *ra = reflection_atlas_owner.get_or_null(rpi->atlas); if (!ra) { ERR_PRINT("reflection probe has no reflection atlas! Bug?"); current_cluster_builder = nullptr; @@ -3696,7 +4160,7 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData render_state.voxel_gi_count = 0; - if (rb != nullptr) { + if (rb != nullptr && is_dynamic_gi_supported()) { if (rb->sdfgi) { rb->sdfgi->update_cascades(); rb->sdfgi->pre_process_gi(render_data.cam_transform, &render_data, this); @@ -3717,9 +4181,28 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData _render_scene(&render_data, clear_color); if (p_render_buffers.is_valid()) { - if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES) { + /* + _debug_draw_cluster(p_render_buffers); + + RENDER_TIMESTAMP("Tonemap"); + + _render_buffers_post_process_and_tonemap(&render_data); + */ + + _render_buffers_debug_draw(p_render_buffers, p_shadow_atlas, p_occluder_debug_tex); + if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI && rb != nullptr && rb->sdfgi != nullptr) { + rb->sdfgi->debug_draw(render_data.cam_projection, render_data.cam_transform, rb->width, rb->height, rb->render_target, rb->texture); + } + } +} + +void RendererSceneRenderRD::_debug_draw_cluster(RID p_render_buffers) { + if (p_render_buffers.is_valid() && current_cluster_builder != nullptr) { + RS::ViewportDebugDraw dd = get_debug_draw_mode(); + + if (dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES) { ClusterBuilderRD::ElementType elem_type = ClusterBuilderRD::ELEMENT_TYPE_MAX; - switch (debug_draw) { + switch (dd) { case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS: elem_type = ClusterBuilderRD::ELEMENT_TYPE_OMNI_LIGHT; break; @@ -3735,23 +4218,13 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData default: { } } - if (current_cluster_builder != nullptr) { - current_cluster_builder->debug(elem_type); - } - } - - RENDER_TIMESTAMP("Tonemap"); - - _render_buffers_post_process_and_tonemap(&render_data); - _render_buffers_debug_draw(p_render_buffers, p_shadow_atlas, p_occluder_debug_tex); - if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI && rb != nullptr && rb->sdfgi != nullptr) { - rb->sdfgi->debug_draw(render_data.cam_projection, render_data.cam_transform, rb->width, rb->height, rb->render_target, rb->texture); + current_cluster_builder->debug(elem_type); } } } void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold, bool p_open_pass, bool p_close_pass, bool p_clear_region, RendererScene::RenderInfo *p_render_info) { - LightInstance *light_instance = light_instance_owner.getornull(p_light); + LightInstance *light_instance = light_instance_owner.get_or_null(p_light); ERR_FAIL_COND(!light_instance); Rect2i atlas_rect; @@ -3760,6 +4233,7 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, bool using_dual_paraboloid = false; bool using_dual_paraboloid_flip = false; + Vector2i dual_paraboloid_offset; RID render_fb; RID render_texture; float zfar; @@ -3785,10 +4259,7 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, light_projection = light_instance->shadow_transform[p_pass].camera; light_transform = light_instance->shadow_transform[p_pass].transform; - atlas_rect.position.x = light_instance->directional_rect.position.x; - atlas_rect.position.y = light_instance->directional_rect.position.y; - atlas_rect.size.width = light_instance->directional_rect.size.x; - atlas_rect.size.height = light_instance->directional_rect.size.y; + atlas_rect = light_instance->directional_rect; if (storage->light_directional_get_shadow_mode(light_instance->light) == RS::LIGHT_DIRECTIONAL_SHADOW_PARALLEL_4_SPLITS) { atlas_rect.size.width /= 2; @@ -3799,8 +4270,7 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, } else if (p_pass == 2) { atlas_rect.position.y += atlas_rect.size.height; } else if (p_pass == 3) { - atlas_rect.position.x += atlas_rect.size.width; - atlas_rect.position.y += atlas_rect.size.height; + atlas_rect.position += atlas_rect.size; } } else if (storage->light_directional_get_shadow_mode(light_instance->light) == RS::LIGHT_DIRECTIONAL_SHADOW_PARALLEL_2_SPLITS) { atlas_rect.size.height /= 2; @@ -3825,7 +4295,7 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, } else { //set from shadow atlas - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_shadow_atlas); ERR_FAIL_COND(!shadow_atlas); ERR_FAIL_COND(!shadow_atlas->shadow_owners.has(p_light)); @@ -3853,6 +4323,9 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, zfar = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_RANGE); if (storage->light_get_type(light_instance->light) == RS::LIGHT_OMNI) { + bool wrap = (shadow + 1) % shadow_atlas->quadrants[quadrant].subdivision == 0; + dual_paraboloid_offset = wrap ? Vector2i(1 - shadow_atlas->quadrants[quadrant].subdivision, 1) : Vector2i(1, 0); + if (storage->light_omni_get_shadow_mode(light_instance->light) == RS::LIGHT_OMNI_SHADOW_CUBE) { ShadowCubemap *cubemap = _get_shadow_cubemap(shadow_size / 2); @@ -3872,12 +4345,16 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, } } else { + atlas_rect.position.x += 1; + atlas_rect.position.y += 1; + atlas_rect.size.x -= 2; + atlas_rect.size.y -= 2; + + atlas_rect.position += p_pass * atlas_rect.size * dual_paraboloid_offset; + light_projection = light_instance->shadow_transform[0].camera; light_transform = light_instance->shadow_transform[0].transform; - atlas_rect.size.height /= 2; - atlas_rect.position.y += p_pass * atlas_rect.size.height; - using_dual_paraboloid = true; using_dual_paraboloid_flip = p_pass == 1; render_fb = shadow_atlas->fb; @@ -3902,14 +4379,11 @@ void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, _render_shadow_end(); //reblit Rect2 atlas_rect_norm = atlas_rect; - atlas_rect_norm.position.x /= float(atlas_size); - atlas_rect_norm.position.y /= float(atlas_size); - atlas_rect_norm.size.x /= float(atlas_size); - atlas_rect_norm.size.y /= float(atlas_size); - atlas_rect_norm.size.height /= 2; - storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, light_projection.get_z_near(), light_projection.get_z_far(), false); - atlas_rect_norm.position.y += atlas_rect_norm.size.height; - storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, light_projection.get_z_near(), light_projection.get_z_far(), true); + atlas_rect_norm.position /= float(atlas_size); + atlas_rect_norm.size /= float(atlas_size); + storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, atlas_rect.size, light_projection.get_z_near(), light_projection.get_z_far(), false); + atlas_rect_norm.position += Vector2(dual_paraboloid_offset) * atlas_rect_norm.size; + storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, atlas_rect.size, light_projection.get_z_near(), light_projection.get_z_far(), true); //restore transform so it can be properly used light_instance_set_shadow_transform(p_light, CameraMatrix(), light_instance->transform, zfar, 0, 0, 0); @@ -3944,7 +4418,7 @@ void RendererSceneRenderRD::render_particle_collider_heightfield(RID p_collider, bool RendererSceneRenderRD::free(RID p_rid) { if (render_buffers_owner.owns(p_rid)) { - RenderBuffers *rb = render_buffers_owner.getornull(p_rid); + RenderBuffers *rb = render_buffers_owner.get_or_null(p_rid); _free_render_buffer_data(rb); memdelete(rb->data); if (rb->sdfgi) { @@ -3967,24 +4441,24 @@ bool RendererSceneRenderRD::free(RID p_rid) { camera_effects_owner.free(p_rid); } else if (reflection_atlas_owner.owns(p_rid)) { reflection_atlas_set_size(p_rid, 0, 0); - ReflectionAtlas *ra = reflection_atlas_owner.getornull(p_rid); + ReflectionAtlas *ra = reflection_atlas_owner.get_or_null(p_rid); if (ra->cluster_builder) { memdelete(ra->cluster_builder); } reflection_atlas_owner.free(p_rid); } else if (reflection_probe_instance_owner.owns(p_rid)) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_rid); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_rid); _free_forward_id(FORWARD_ID_TYPE_REFLECTION_PROBE, rpi->forward_id); reflection_probe_release_atlas_index(p_rid); reflection_probe_instance_owner.free(p_rid); } else if (decal_instance_owner.owns(p_rid)) { - DecalInstance *di = decal_instance_owner.getornull(p_rid); + DecalInstance *di = decal_instance_owner.get_or_null(p_rid); _free_forward_id(FORWARD_ID_TYPE_DECAL, di->forward_id); decal_instance_owner.free(p_rid); } else if (lightmap_instance_owner.owns(p_rid)) { lightmap_instance_owner.free(p_rid); } else if (gi.voxel_gi_instance_owner.owns(p_rid)) { - RendererSceneGIRD::VoxelGIInstance *voxel_gi = gi.voxel_gi_instance_owner.getornull(p_rid); + RendererSceneGIRD::VoxelGIInstance *voxel_gi = gi.voxel_gi_instance_owner.get_or_null(p_rid); if (voxel_gi->texture.is_valid()) { RD::get_singleton()->free(voxel_gi->texture); RD::get_singleton()->free(voxel_gi->write_buffer); @@ -4000,17 +4474,23 @@ bool RendererSceneRenderRD::free(RID p_rid) { sky.update_dirty_skys(); sky.free_sky(p_rid); } else if (light_instance_owner.owns(p_rid)) { - LightInstance *light_instance = light_instance_owner.getornull(p_rid); + LightInstance *light_instance = light_instance_owner.get_or_null(p_rid); //remove from shadow atlases.. for (Set<RID>::Element *E = light_instance->shadow_atlases.front(); E; E = E->next()) { - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(E->get()); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(E->get()); ERR_CONTINUE(!shadow_atlas->shadow_owners.has(p_rid)); uint32_t key = shadow_atlas->shadow_owners[p_rid]; uint32_t q = (key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; uint32_t s = key & ShadowAtlas::SHADOW_INDEX_MASK; shadow_atlas->quadrants[q].shadows.write[s].owner = RID(); + + if (key & ShadowAtlas::OMNI_LIGHT_FLAG) { + // Omni lights use two atlas spots, make sure to clear the other as well + shadow_atlas->quadrants[q].shadows.write[s + 1].owner = RID(); + } + shadow_atlas->shadow_owners.erase(p_rid); } @@ -4208,10 +4688,12 @@ uint32_t RendererSceneRenderRD::get_max_elements() const { } RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { - max_cluster_elements = get_max_elements(); - storage = p_storage; singleton = this; +} + +void RendererSceneRenderRD::init() { + max_cluster_elements = get_max_elements(); directional_shadow.size = GLOBAL_GET("rendering/shadows/directional_shadow/size"); directional_shadow.use_16_bits = GLOBAL_GET("rendering/shadows/directional_shadow/16_bits"); @@ -4313,8 +4795,8 @@ RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { } RendererSceneRenderRD::~RendererSceneRenderRD() { - for (Map<int, ShadowCubemap>::Element *E = shadow_cubemaps.front(); E; E = E->next()) { - RD::get_singleton()->free(E->get().cubemap); + for (const KeyValue<int, ShadowCubemap> &E : shadow_cubemaps) { + RD::get_singleton()->free(E.value.cubemap); } if (sky.sky_scene_state.uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(sky.sky_scene_state.uniform_set)) { diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.h b/servers/rendering/renderer_rd/renderer_scene_render_rd.h index 5e0281002d..d74848f0af 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.h @@ -95,7 +95,7 @@ protected: double time_step = 0; struct RenderBufferData { - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) = 0; + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) = 0; virtual ~RenderBufferData() {} }; virtual RenderBufferData *_create_render_buffer_data() = 0; @@ -117,6 +117,7 @@ protected: virtual void _render_particle_collider_heightfield(RID p_fb, const Transform3D &p_cam_transform, const CameraMatrix &p_cam_projection, const PagedArray<GeometryInstance *> &p_instances) = 0; void _debug_sdfgi_probes(RID p_render_buffers, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform); + void _debug_draw_cluster(RID p_render_buffers); RenderBufferData *render_buffers_get_data(RID p_render_buffers); @@ -133,6 +134,12 @@ protected: void _pre_opaque_render(RenderDataRD *p_render_data, bool p_use_ssao, bool p_use_gi, RID p_normal_roughness_buffer, RID p_voxel_gi_buffer); + void _render_buffers_copy_screen_texture(const RenderDataRD *p_render_data); + void _render_buffers_copy_depth_texture(const RenderDataRD *p_render_data); + void _render_buffers_post_process_and_tonemap(const RenderDataRD *p_render_data); + void _post_process_subpass(RID p_source_texture, RID p_framebuffer, const RenderDataRD *p_render_data); + void _disable_clear_request(const RenderDataRD *p_render_data); + // needed for a single argument calls (material and uv2) PagedArrayPool<GeometryInstance *> cull_argument_pool; PagedArray<GeometryInstance *> cull_argument; //need this to exist @@ -142,11 +149,11 @@ protected: RendererSceneEnvironmentRD *get_environment(RID p_environment) { if (p_environment.is_valid()) { - return environment_owner.getornull(p_environment); + return environment_owner.get_or_null(p_environment); } else { return nullptr; } - } + }; //used for mobile renderer mostly @@ -248,7 +255,8 @@ private: struct ShadowAtlas { enum { QUADRANT_SHIFT = 27, - SHADOW_INDEX_MASK = (1 << QUADRANT_SHIFT) - 1, + OMNI_LIGHT_FLAG = 1 << 26, + SHADOW_INDEX_MASK = OMNI_LIGHT_FLAG - 1, SHADOW_INVALID = 0xFFFFFFFF }; @@ -292,7 +300,9 @@ private: void _update_shadow_atlas(ShadowAtlas *shadow_atlas); + void _shadow_atlas_invalidate_shadow(RendererSceneRenderRD::ShadowAtlas::Quadrant::Shadow *p_shadow, RID p_atlas, RendererSceneRenderRD::ShadowAtlas *p_shadow_atlas, uint32_t p_quadrant, uint32_t p_shadow_idx); bool _shadow_atlas_find_shadow(ShadowAtlas *shadow_atlas, int *p_in_quadrants, int p_quadrant_count, int p_current_subdiv, uint64_t p_tick, int &r_quadrant, int &r_shadow); + bool _shadow_atlas_find_omni_shadows(ShadowAtlas *shadow_atlas, int *p_in_quadrants, int p_quadrant_count, int p_current_subdiv, uint64_t p_tick, int &r_quadrant, int &r_shadow); RS::ShadowQuality shadows_quality = RS::SHADOW_QUALITY_MAX; //So it always updates when first set RS::ShadowQuality directional_shadow_quality = RS::SHADOW_QUALITY_MAX; @@ -372,10 +382,6 @@ private: uint32_t cull_mask = 0; uint32_t light_directional_index = 0; - uint32_t current_shadow_atlas_key = 0; - - Vector2 dp; - Rect2 directional_rect; Set<RID> shadow_atlases; //shadow atlases where this light is registered @@ -450,6 +456,7 @@ private: RID texture; //main texture for rendering to, must be filled after done rendering RID depth_texture; //main depth texture + RID texture_fb; // framebuffer for the main texture, ONLY USED FOR MOBILE RENDERER POST EFFECTS, DO NOT USE FOR RENDERING 3D!!! RendererSceneGIRD::SDFGI *sdfgi = nullptr; VolumetricFog *volumetric_fog = nullptr; @@ -465,6 +472,11 @@ private: RID texture; int width; int height; + + // only used on mobile renderer + RID fb; + RID half_texture; + RID half_fb; }; Vector<Mipmap> mipmaps; @@ -472,9 +484,25 @@ private: Blur blur[2]; //the second one starts from the first mipmap + struct WeightBuffers { + RID weight; + RID fb; // FB with both texture and weight + }; + + // 2 full size, 2 half size + WeightBuffers weight_buffers[4]; // Only used in raster + RID base_weight_fb; // base buffer for weight + + RID depth_back_texture; + RID depth_back_fb; // only used on mobile + struct Luminance { Vector<RID> reduce; RID current; + + // used only on mobile renderer + Vector<RID> fb; + RID current_fb; } luminance; struct SSAO { @@ -511,10 +539,10 @@ private: void _free_render_buffer_data(RenderBuffers *rb); void _allocate_blur_textures(RenderBuffers *rb); + void _allocate_depth_backbuffer_textures(RenderBuffers *rb); void _allocate_luminance_textures(RenderBuffers *rb); void _render_buffers_debug_draw(RID p_render_buffers, RID p_shadow_atlas, RID p_occlusion_buffer); - void _render_buffers_post_process_and_tonemap(const RenderDataRD *p_render_data); /* Cluster */ @@ -546,7 +574,7 @@ private: struct LightData { float position[3]; float inv_radius; - float direction[3]; + float direction[3]; // in omni, x and y are used for dual paraboloid offset float size; float color[3]; @@ -786,19 +814,19 @@ public: virtual void shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision) override; virtual bool shadow_atlas_update_light(RID p_atlas, RID p_light_intance, float p_coverage, uint64_t p_light_version) override; _FORCE_INLINE_ bool shadow_atlas_owns_light_instance(RID p_atlas, RID p_light_intance) { - ShadowAtlas *atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND_V(!atlas, false); return atlas->shadow_owners.has(p_light_intance); } _FORCE_INLINE_ RID shadow_atlas_get_texture(RID p_atlas) { - ShadowAtlas *atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND_V(!atlas, RID()); return atlas->depth; } _FORCE_INLINE_ Size2i shadow_atlas_get_size(RID p_atlas) { - ShadowAtlas *atlas = shadow_atlas_owner.getornull(p_atlas); + ShadowAtlas *atlas = shadow_atlas_owner.get_or_null(p_atlas); ERR_FAIL_COND_V(!atlas, Size2i()); return Size2(atlas->size, atlas->size); } @@ -845,7 +873,7 @@ public: virtual void environment_set_bg_color(RID p_env, const Color &p_color) override; virtual void environment_set_bg_energy(RID p_env, float p_energy) override; virtual void environment_set_canvas_max_layer(RID p_env, int p_max_layer) override; - virtual void environment_set_ambient_light(RID p_env, const Color &p_color, RS::EnvironmentAmbientSource p_ambient = RS::ENV_AMBIENT_SOURCE_BG, float p_energy = 1.0, float p_sky_contribution = 0.0, RS::EnvironmentReflectionSource p_reflection_source = RS::ENV_REFLECTION_SOURCE_BG, const Color &p_ao_color = Color()) override; + virtual void environment_set_ambient_light(RID p_env, const Color &p_color, RS::EnvironmentAmbientSource p_ambient = RS::ENV_AMBIENT_SOURCE_BG, float p_energy = 1.0, float p_sky_contribution = 0.0, RS::EnvironmentReflectionSource p_reflection_source = RS::ENV_REFLECTION_SOURCE_BG) override; virtual RS::EnvironmentBG environment_get_background(RID p_env) const override; RID environment_get_sky(RID p_env) const; @@ -859,7 +887,6 @@ public: float environment_get_ambient_light_energy(RID p_env) const; float environment_get_ambient_sky_contribution(RID p_env) const; RS::EnvironmentReflectionSource environment_get_reflection_source(RID p_env) const; - Color environment_get_ao_color(RID p_env) const; virtual bool is_environment(RID p_env) const override; @@ -913,6 +940,12 @@ public: virtual void camera_effects_set_dof_blur(RID p_camera_effects, bool p_far_enable, float p_far_distance, float p_far_transition, bool p_near_enable, float p_near_distance, float p_near_transition, float p_amount) override; virtual void camera_effects_set_custom_exposure(RID p_camera_effects, bool p_enable, float p_exposure) override; + bool camera_effects_uses_dof(RID p_camera_effects) { + CameraEffects *camfx = camera_effects_owner.get_or_null(p_camera_effects); + + return camfx && (camfx->dof_blur_near_enabled || camfx->dof_blur_far_enabled) && camfx->dof_blur_amount > 0.0; + } + virtual RID light_instance_create(RID p_light) override; virtual void light_instance_set_transform(RID p_light_instance, const Transform3D &p_transform) override; virtual void light_instance_set_aabb(RID p_light_instance, const AABB &p_aabb) override; @@ -920,18 +953,18 @@ public: virtual void light_instance_mark_visible(RID p_light_instance) override; _FORCE_INLINE_ RID light_instance_get_base_light(RID p_light_instance) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->light; } _FORCE_INLINE_ Transform3D light_instance_get_base_transform(RID p_light_instance) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->transform; } - _FORCE_INLINE_ Rect2 light_instance_get_shadow_atlas_rect(RID p_light_instance, RID p_shadow_atlas) { - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); - LightInstance *li = light_instance_owner.getornull(p_light_instance); + _FORCE_INLINE_ Rect2 light_instance_get_shadow_atlas_rect(RID p_light_instance, RID p_shadow_atlas, Vector2i &r_omni_offset) { + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_shadow_atlas); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); uint32_t key = shadow_atlas->shadow_owners[li->self]; uint32_t quadrant = (key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; @@ -949,6 +982,16 @@ public: x += (shadow % shadow_atlas->quadrants[quadrant].subdivision) * shadow_size; y += (shadow / shadow_atlas->quadrants[quadrant].subdivision) * shadow_size; + if (key & ShadowAtlas::OMNI_LIGHT_FLAG) { + if (((shadow + 1) % shadow_atlas->quadrants[quadrant].subdivision) == 0) { + r_omni_offset.x = 1 - int(shadow_atlas->quadrants[quadrant].subdivision); + r_omni_offset.y = 1; + } else { + r_omni_offset.x = 1; + r_omni_offset.y = 0; + } + } + uint32_t width = shadow_size; uint32_t height = shadow_size; @@ -956,16 +999,16 @@ public: } _FORCE_INLINE_ CameraMatrix light_instance_get_shadow_camera(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].camera; } _FORCE_INLINE_ float light_instance_get_shadow_texel_size(RID p_light_instance, RID p_shadow_atlas) { #ifdef DEBUG_ENABLED - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); ERR_FAIL_COND_V(!li->shadow_atlases.has(p_shadow_atlas), 0); #endif - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + ShadowAtlas *shadow_atlas = shadow_atlas_owner.get_or_null(p_shadow_atlas); ERR_FAIL_COND_V(!shadow_atlas, 0); #ifdef DEBUG_ENABLED ERR_FAIL_COND_V(!shadow_atlas->shadow_owners.has(p_light_instance), 0); @@ -983,59 +1026,59 @@ public: _FORCE_INLINE_ Transform3D light_instance_get_shadow_transform(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].transform; } _FORCE_INLINE_ float light_instance_get_shadow_bias_scale(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].bias_scale; } _FORCE_INLINE_ float light_instance_get_shadow_range(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].farplane; } _FORCE_INLINE_ float light_instance_get_shadow_range_begin(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].range_begin; } _FORCE_INLINE_ Vector2 light_instance_get_shadow_uv_scale(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].uv_scale; } _FORCE_INLINE_ Rect2 light_instance_get_directional_shadow_atlas_rect(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].atlas_rect; } _FORCE_INLINE_ float light_instance_get_directional_shadow_split(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].split; } _FORCE_INLINE_ float light_instance_get_directional_shadow_texel_size(RID p_light_instance, int p_index) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->shadow_transform[p_index].shadow_texel_size; } _FORCE_INLINE_ void light_instance_set_render_pass(RID p_light_instance, uint64_t p_pass) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); li->last_pass = p_pass; } _FORCE_INLINE_ uint64_t light_instance_get_render_pass(RID p_light_instance) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->last_pass; } _FORCE_INLINE_ ForwardID light_instance_get_forward_id(RID p_light_instance) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->forward_id; } _FORCE_INLINE_ RS::LightType light_instance_get_type(RID p_light_instance) { - LightInstance *li = light_instance_owner.getornull(p_light_instance); + LightInstance *li = light_instance_owner.get_or_null(p_light_instance); return li->light_type; } @@ -1044,7 +1087,7 @@ public: virtual int reflection_atlas_get_size(RID p_ref_atlas) const override; _FORCE_INLINE_ RID reflection_atlas_get_texture(RID p_ref_atlas) { - ReflectionAtlas *atlas = reflection_atlas_owner.getornull(p_ref_atlas); + ReflectionAtlas *atlas = reflection_atlas_owner.get_or_null(p_ref_atlas); ERR_FAIL_COND_V(!atlas, RID()); return atlas->reflection; } @@ -1055,6 +1098,7 @@ public: virtual bool reflection_probe_instance_needs_redraw(RID p_instance) override; virtual bool reflection_probe_instance_has_reflection(RID p_instance) override; virtual bool reflection_probe_instance_begin_render(RID p_instance, RID p_reflection_atlas) override; + virtual RID reflection_probe_create_framebuffer(RID p_color, RID p_depth); virtual bool reflection_probe_instance_postprocess_step(RID p_instance) override; uint32_t reflection_probe_instance_get_resolution(RID p_instance); @@ -1062,41 +1106,41 @@ public: RID reflection_probe_instance_get_depth_framebuffer(RID p_instance, int p_index); _FORCE_INLINE_ RID reflection_probe_instance_get_probe(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, RID()); return rpi->probe; } _FORCE_INLINE_ ForwardID reflection_probe_instance_get_forward_id(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, 0); return rpi->forward_id; } _FORCE_INLINE_ void reflection_probe_instance_set_render_pass(RID p_instance, uint32_t p_render_pass) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND(!rpi); rpi->last_pass = p_render_pass; } _FORCE_INLINE_ uint32_t reflection_probe_instance_get_render_pass(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, 0); return rpi->last_pass; } _FORCE_INLINE_ Transform3D reflection_probe_instance_get_transform(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, Transform3D()); return rpi->transform; } _FORCE_INLINE_ int reflection_probe_instance_get_atlas_index(RID p_instance) { - ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.get_or_null(p_instance); ERR_FAIL_COND_V(!rpi, -1); return rpi->atlas_index; @@ -1106,32 +1150,32 @@ public: virtual void decal_instance_set_transform(RID p_decal, const Transform3D &p_transform) override; _FORCE_INLINE_ RID decal_instance_get_base(RID p_decal) const { - DecalInstance *decal = decal_instance_owner.getornull(p_decal); + DecalInstance *decal = decal_instance_owner.get_or_null(p_decal); return decal->decal; } _FORCE_INLINE_ ForwardID decal_instance_get_forward_id(RID p_decal) const { - DecalInstance *decal = decal_instance_owner.getornull(p_decal); + DecalInstance *decal = decal_instance_owner.get_or_null(p_decal); return decal->forward_id; } _FORCE_INLINE_ Transform3D decal_instance_get_transform(RID p_decal) const { - DecalInstance *decal = decal_instance_owner.getornull(p_decal); + DecalInstance *decal = decal_instance_owner.get_or_null(p_decal); return decal->transform; } virtual RID lightmap_instance_create(RID p_lightmap) override; virtual void lightmap_instance_set_transform(RID p_lightmap, const Transform3D &p_transform) override; _FORCE_INLINE_ bool lightmap_instance_is_valid(RID p_lightmap_instance) { - return lightmap_instance_owner.getornull(p_lightmap_instance) != nullptr; + return lightmap_instance_owner.get_or_null(p_lightmap_instance) != nullptr; } _FORCE_INLINE_ RID lightmap_instance_get_lightmap(RID p_lightmap_instance) { - LightmapInstance *li = lightmap_instance_owner.getornull(p_lightmap_instance); + LightmapInstance *li = lightmap_instance_owner.get_or_null(p_lightmap_instance); return li->lightmap; } _FORCE_INLINE_ Transform3D lightmap_instance_get_transform(RID p_lightmap_instance) { - LightmapInstance *li = lightmap_instance_owner.getornull(p_lightmap_instance); + LightmapInstance *li = lightmap_instance_owner.get_or_null(p_lightmap_instance); return li->transform; } @@ -1145,14 +1189,17 @@ public: /* render buffers */ + virtual float _render_buffers_get_luminance_multiplier(); virtual RD::DataFormat _render_buffers_get_color_format(); virtual bool _render_buffers_can_be_storage(); virtual RID render_buffers_create() override; virtual void render_buffers_configure(RID p_render_buffers, RID p_render_target, int p_width, int p_height, RS::ViewportMSAA p_msaa, RS::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_debanding, uint32_t p_view_count) override; virtual void gi_set_use_half_resolution(bool p_enable) override; + RID render_buffers_get_depth_texture(RID p_render_buffers); RID render_buffers_get_ao_texture(RID p_render_buffers); RID render_buffers_get_back_buffer_texture(RID p_render_buffers); + RID render_buffers_get_back_depth_texture(RID p_render_buffers); RID render_buffers_get_voxel_gi_buffer(RID p_render_buffers); RID render_buffers_get_default_voxel_gi_buffer(); RID render_buffers_get_gi_ambient_texture(RID p_render_buffers); @@ -1253,6 +1300,8 @@ public: virtual bool is_volumetric_supported() const; virtual uint32_t get_max_elements() const; + void init(); + RendererSceneRenderRD(RendererStorageRD *p_storage); ~RendererSceneRenderRD(); }; diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp index bc1603a219..5814c164cc 100644 --- a/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp @@ -92,8 +92,7 @@ void RendererSceneSkyRD::SkyShaderData::set_code(const String &p_code) { RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; Error err = scene_singleton->sky.sky_shader.compiler.compile(RS::SHADER_SKY, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); + ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); if (version.is_null()) { version = scene_singleton->sky.sky_shader.shader.version_create(); @@ -149,36 +148,36 @@ void RendererSceneSkyRD::SkyShaderData::set_default_texture_param(const StringNa void RendererSceneSkyRD::SkyShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { Map<int, StringName> order; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); + if (E.value.texture_order >= 0) { + order[E.value.texture_order + 100000] = E.key; } else { - order[E->get().order] = E->key(); + order[E.value.order] = E.key; } } - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); + for (const KeyValue<int, StringName> &E : order) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E.value]); + pi.name = E.value; p_param_list->push_back(pi); } } void RendererSceneSkyRD::SkyShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p.info = ShaderLanguage::uniform_to_property_info(E.value); + p.info.name = E.key; //supply name + p.index = E.value.instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E.value.default_value, E.value.type, E.value.array_size, E.value.hint); p_param_list->push_back(p); } } @@ -203,7 +202,7 @@ Variant RendererSceneSkyRD::SkyShaderData::get_default_parameter(const StringNam if (uniforms.has(p_parameter)) { ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.array_size, uniform.hint); } return Variant(); } @@ -260,7 +259,7 @@ static _FORCE_INLINE_ void store_transform_3x3(const Basis &p_basis, float *p_ar p_array[11] = 0; } -void RendererSceneSkyRD::_render_sky(RD::DrawListID p_list, float p_time, RID p_fb, PipelineCacheRD *p_pipeline, RID p_uniform_set, RID p_texture_set, uint32_t p_view_count, const CameraMatrix *p_projections, const Basis &p_orientation, float p_multiplier, const Vector3 &p_position) { +void RendererSceneSkyRD::_render_sky(RD::DrawListID p_list, float p_time, RID p_fb, PipelineCacheRD *p_pipeline, RID p_uniform_set, RID p_texture_set, uint32_t p_view_count, const CameraMatrix *p_projections, const Basis &p_orientation, float p_multiplier, const Vector3 &p_position, float p_luminance_multiplier) { SkyPushConstant sky_push_constant; memset(&sky_push_constant, 0, sizeof(SkyPushConstant)); @@ -277,20 +276,24 @@ void RendererSceneSkyRD::_render_sky(RD::DrawListID p_list, float p_time, RID p_ sky_push_constant.position[2] = p_position.z; sky_push_constant.multiplier = p_multiplier; sky_push_constant.time = p_time; + sky_push_constant.luminance_multiplier = p_luminance_multiplier; store_transform_3x3(p_orientation, sky_push_constant.orientation); RenderingDevice::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(p_fb); RD::DrawListID draw_list = p_list; - RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, p_pipeline->get_render_pipeline(RD::INVALID_ID, fb_format)); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, p_pipeline->get_render_pipeline(RD::INVALID_ID, fb_format, false, RD::get_singleton()->draw_list_get_current_pass())); - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.uniform_set, 0); - if (p_uniform_set.is_valid()) { //material may not have uniform set - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_uniform_set, 1); + // Update uniform sets. + { + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.uniform_set, 0); + if (p_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(p_uniform_set)) { // Material may not have a uniform set. + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_uniform_set, 1); + } + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_texture_set, 2); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.fog_uniform_set, 3); } - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_texture_set, 2); - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.fog_uniform_set, 3); RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); @@ -313,12 +316,16 @@ void RendererSceneSkyRD::ReflectionData::clear_reflection_data() { coefficient_buffer = RID(); } -void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers) { +void RendererSceneSkyRD::ReflectionData::update_reflection_data(RendererStorageRD *p_storage, int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers, RD::DataFormat p_texture_format) { //recreate radiance and all data int mipmaps = p_mipmaps; uint32_t w = p_size, h = p_size; + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + if (p_use_array) { int num_layers = p_low_quality ? 8 : p_roughness_layers; @@ -377,9 +384,9 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int } radiance_base_cubemap = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer, 0, RD::TEXTURE_SLICE_CUBEMAP); - + RD::get_singleton()->set_resource_name(radiance_base_cubemap, "radiance base cubemap"); RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = p_texture_format; tf.width = 64; // Always 64x64 tf.height = 64; tf.texture_type = RD::TEXTURE_TYPE_CUBE; @@ -388,6 +395,7 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; downsampled_radiance_cubemap = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(downsampled_radiance_cubemap, "downsampled radiance cubemap"); { uint32_t mmw = 64; uint32_t mmh = 64; @@ -397,6 +405,18 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int mm.size.width = mmw; mm.size.height = mmh; mm.view = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), downsampled_radiance_cubemap, 0, j, RD::TEXTURE_SLICE_CUBEMAP); + RD::get_singleton()->set_resource_name(mm.view, "Downsampled Radiance Cubemap Mip " + itos(j) + " "); + if (prefer_raster_effects) { + // we need a framebuffer for each side of our cubemap + + for (int k = 0; k < 6; k++) { + mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), downsampled_radiance_cubemap, k, j); + RD::get_singleton()->set_resource_name(mm.view, "Downsampled Radiance Cubemap Mip: " + itos(j) + " Face: " + itos(k) + " "); + Vector<RID> fbtex; + fbtex.push_back(mm.views[k]); + mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); + } + } mmw = MAX(1, mmw >> 1); mmh = MAX(1, mmh >> 1); @@ -405,50 +425,128 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int } void RendererSceneSkyRD::ReflectionData::create_reflection_fast_filter(RendererStorageRD *p_storage, bool p_use_arrays) { - p_storage->get_effects()->cubemap_downsample(radiance_base_cubemap, downsampled_layer.mipmaps[0].view, downsampled_layer.mipmaps[0].size); + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + if (prefer_raster_effects) { + RD::get_singleton()->draw_command_begin_label("Downsample radiance map"); + for (int k = 0; k < 6; k++) { + effects->cubemap_downsample_raster(radiance_base_cubemap, downsampled_layer.mipmaps[0].framebuffers[k], k, downsampled_layer.mipmaps[0].size); + } - for (int i = 1; i < downsampled_layer.mipmaps.size(); i++) { - p_storage->get_effects()->cubemap_downsample(downsampled_layer.mipmaps[i - 1].view, downsampled_layer.mipmaps[i].view, downsampled_layer.mipmaps[i].size); - } + for (int i = 1; i < downsampled_layer.mipmaps.size(); i++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_downsample_raster(downsampled_layer.mipmaps[i - 1].view, downsampled_layer.mipmaps[i].framebuffers[k], k, downsampled_layer.mipmaps[i].size); + } + } + RD::get_singleton()->draw_command_end_label(); // Downsample Radiance - Vector<RID> views; - if (p_use_arrays) { - for (int i = 1; i < layers.size(); i++) { - views.push_back(layers[i].views[0]); + if (p_use_arrays) { + RD::get_singleton()->draw_command_begin_label("filter radiance map into array heads"); + for (int i = 0; i < layers.size(); i++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_filter_raster(downsampled_radiance_cubemap, layers[i].mipmaps[0].framebuffers[k], k, i); + } + } + } else { + RD::get_singleton()->draw_command_begin_label("filter radiance map into mipmaps directly"); + for (int j = 0; j < layers[0].mipmaps.size(); j++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_filter_raster(downsampled_radiance_cubemap, layers[0].mipmaps[j].framebuffers[k], k, j); + } + } } + RD::get_singleton()->draw_command_end_label(); // Filter radiance } else { - for (int i = 1; i < layers[0].views.size(); i++) { - views.push_back(layers[0].views[i]); + effects->cubemap_downsample(radiance_base_cubemap, downsampled_layer.mipmaps[0].view, downsampled_layer.mipmaps[0].size); + + for (int i = 1; i < downsampled_layer.mipmaps.size(); i++) { + effects->cubemap_downsample(downsampled_layer.mipmaps[i - 1].view, downsampled_layer.mipmaps[i].view, downsampled_layer.mipmaps[i].size); + } + + Vector<RID> views; + if (p_use_arrays) { + for (int i = 1; i < layers.size(); i++) { + views.push_back(layers[i].views[0]); + } + } else { + for (int i = 1; i < layers[0].views.size(); i++) { + views.push_back(layers[0].views[i]); + } } - } - p_storage->get_effects()->cubemap_filter(downsampled_radiance_cubemap, views, p_use_arrays); + effects->cubemap_filter(downsampled_radiance_cubemap, views, p_use_arrays); + } } void RendererSceneSkyRD::ReflectionData::create_reflection_importance_sample(RendererStorageRD *p_storage, bool p_use_arrays, int p_cube_side, int p_base_layer, uint32_t p_sky_ggx_samples_quality) { - if (p_use_arrays) { - //render directly to the layers - p_storage->get_effects()->cubemap_roughness(radiance_base_cubemap, layers[p_base_layer].views[0], p_cube_side, p_sky_ggx_samples_quality, float(p_base_layer) / (layers.size() - 1.0), layers[p_base_layer].mipmaps[0].size.x); + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + if (prefer_raster_effects) { + // Need to ask clayjohn but p_cube_side is set to 10, looks like in the compute shader we're doing all 6 sides in one call + // here we need to do them one by one so ignoring p_cube_side + if (p_use_arrays) { + for (int k = 0; k < 6; k++) { + effects->cubemap_roughness_raster( + radiance_base_cubemap, + layers[p_base_layer].mipmaps[0].framebuffers[k], + k, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers.size() - 1.0), + layers[p_base_layer].mipmaps[0].size.x); + } + } else { + for (int k = 0; k < 6; k++) { + effects->cubemap_roughness_raster( + layers[0].views[p_base_layer - 1], + layers[0].mipmaps[p_base_layer].framebuffers[k], + k, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers[0].mipmaps.size() - 1.0), + layers[0].mipmaps[p_base_layer].size.x); + } + } } else { - p_storage->get_effects()->cubemap_roughness( - layers[0].views[p_base_layer - 1], - layers[0].views[p_base_layer], - p_cube_side, - p_sky_ggx_samples_quality, - float(p_base_layer) / (layers[0].mipmaps.size() - 1.0), - layers[0].mipmaps[p_base_layer].size.x); + if (p_use_arrays) { + //render directly to the layers + effects->cubemap_roughness(radiance_base_cubemap, layers[p_base_layer].views[0], p_cube_side, p_sky_ggx_samples_quality, float(p_base_layer) / (layers.size() - 1.0), layers[p_base_layer].mipmaps[0].size.x); + } else { + effects->cubemap_roughness( + layers[0].views[p_base_layer - 1], + layers[0].views[p_base_layer], + p_cube_side, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers[0].mipmaps.size() - 1.0), + layers[0].mipmaps[p_base_layer].size.x); + } } } void RendererSceneSkyRD::ReflectionData::update_reflection_mipmaps(RendererStorageRD *p_storage, int p_start, int p_end) { + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + RD::get_singleton()->draw_command_begin_label("Update Radiance Cubemap Array Mipmaps"); for (int i = p_start; i < p_end; i++) { for (int j = 0; j < layers[i].views.size() - 1; j++) { RID view = layers[i].views[j]; - RID texture = layers[i].views[j + 1]; Size2i size = layers[i].mipmaps[j + 1].size; - p_storage->get_effects()->cubemap_downsample(view, texture, size); + if (prefer_raster_effects) { + for (int k = 0; k < 6; k++) { + RID framebuffer = layers[i].mipmaps[j + 1].framebuffers[k]; + effects->cubemap_downsample_raster(view, framebuffer, k, size); + } + } else { + RID texture = layers[i].views[j + 1]; + effects->cubemap_downsample(view, texture, size); + } } } + RD::get_singleton()->draw_command_end_label(); } //////////////////////////////////////////////////////////////////////////////// @@ -758,6 +856,8 @@ void RendererSceneSkyRD::init(RendererStorageRD *p_storage) { storage->shader_initialize(sky_shader.default_shader); storage->shader_set_code(sky_shader.default_shader, R"( +// Default sky shader. + shader_type sky; void sky() { @@ -845,6 +945,8 @@ void sky() { storage->shader_initialize(sky_scene_state.fog_shader); storage->shader_set_code(sky_scene_state.fog_shader, R"( +// Default clear color sky shader. + shader_type sky; uniform vec4 clear_color; @@ -902,6 +1004,10 @@ void sky() { } } +void RendererSceneSkyRD::set_texture_format(RD::DataFormat p_texture_format) { + texture_format = p_texture_format; +} + RendererSceneSkyRD::~RendererSceneSkyRD() { // TODO cleanup anything created in init... @@ -930,11 +1036,7 @@ void RendererSceneSkyRD::setup(RendererSceneEnvironmentRD *p_env, RID p_render_b SkyShaderData *shader_data = nullptr; - RS::EnvironmentBG background = p_env->background; - - if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { - // !BAS! Possibly silently fail here, we now get error spam when you select sky as the background but haven't setup the sky yet. - ERR_FAIL_COND(!sky); + if (sky) { sky_material = sky_get_material(p_env->sky); if (sky_material.is_valid()) { @@ -954,9 +1056,7 @@ void RendererSceneSkyRD::setup(RendererSceneEnvironmentRD *p_env, RID p_render_b shader_data = material->shader_data; ERR_FAIL_COND(!shader_data); - } - if (sky) { // Invalidate supbass buffers if screen size changes if (sky->screen_size != p_screen_size) { sky->screen_size = p_screen_size; @@ -1090,7 +1190,7 @@ void RendererSceneSkyRD::setup(RendererSceneEnvironmentRD *p_env, RID p_render_b RD::get_singleton()->buffer_update(sky_scene_state.uniform_buffer, 0, sizeof(SkySceneState::UBO), &sky_scene_state.ubo); } -void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform3D &p_transform, double p_time) { +void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform3D &p_transform, double p_time, float p_luminance_multiplier) { ERR_FAIL_COND(!p_env); Sky *sky = get_sky(p_env->sky); @@ -1170,6 +1270,7 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM cm = correction * cm; if (shader_data->uses_quarter_res) { + RD::get_singleton()->draw_command_begin_label("Render Sky to Quarter Res Cubemap"); PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_QUARTER_RES]; Vector<Color> clear_colors; @@ -1177,17 +1278,18 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM RD::DrawListID cubemap_draw_list; for (int i = 0; i < 6; i++) { - Transform3D local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + Basis local_view = Basis::looking_at(view_normals[i], view_up[i]); RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES, sky_shader.default_shader_rd); cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[2].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[2].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view.basis, multiplier, p_transform.origin); + _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[2].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin, p_luminance_multiplier); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); } if (shader_data->uses_half_res) { + RD::get_singleton()->draw_command_begin_label("Render Sky to Half Res Cubemap"); PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_HALF_RES]; Vector<Color> clear_colors; @@ -1195,28 +1297,29 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM RD::DrawListID cubemap_draw_list; for (int i = 0; i < 6; i++) { - Transform3D local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + Basis local_view = Basis::looking_at(view_normals[i], view_up[i]); RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP_HALF_RES, sky_shader.default_shader_rd); cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[1].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[1].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view.basis, multiplier, p_transform.origin); + _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[1].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin, p_luminance_multiplier); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); } RD::DrawListID cubemap_draw_list; PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP]; + RD::get_singleton()->draw_command_begin_label("Render Sky Cubemap"); for (int i = 0; i < 6; i++) { - Transform3D local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + Basis local_view = Basis::looking_at(view_normals[i], view_up[i]); RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP, sky_shader.default_shader_rd); cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[0].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[0].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view.basis, multiplier, p_transform.origin); + _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[0].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin, p_luminance_multiplier); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); if (sky_mode == RS::SKY_MODE_REALTIME) { sky->reflection.create_reflection_fast_filter(storage, sky_use_cubemap_array); @@ -1262,7 +1365,6 @@ void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_cont ERR_FAIL_COND(p_view_count > RendererSceneRender::MAX_RENDER_VIEWS); Sky *sky = get_sky(p_env->sky); - ERR_FAIL_COND(!sky); SkyMaterialData *material = nullptr; RID sky_material; @@ -1331,7 +1433,7 @@ void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_cont clear_colors.push_back(Color(0.0, 0.0, 0.0)); RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->quarter_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); - _render_sky(draw_list, p_time, sky->quarter_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); + _render_sky(draw_list, p_time, sky->quarter_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, 1.0); RD::get_singleton()->draw_list_end(); } @@ -1344,7 +1446,7 @@ void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_cont clear_colors.push_back(Color(0.0, 0.0, 0.0)); RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->half_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); - _render_sky(draw_list, p_time, sky->half_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); + _render_sky(draw_list, p_time, sky->half_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, 1.0); RD::get_singleton()->draw_list_end(); } @@ -1358,10 +1460,172 @@ void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_cont } RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_fb, RD::INITIAL_ACTION_CONTINUE, p_can_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, p_can_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ); - _render_sky(draw_list, p_time, p_fb, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); + _render_sky(draw_list, p_time, p_fb, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, 1.0); RD::get_singleton()->draw_list_end(); } +void RendererSceneSkyRD::update_res_buffers(RendererSceneEnvironmentRD *p_env, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time, float p_luminance_multiplier) { + ERR_FAIL_COND(!p_env); + + ERR_FAIL_COND(p_view_count == 0); + ERR_FAIL_COND(p_view_count > RendererSceneRender::MAX_RENDER_VIEWS); + + Sky *sky = get_sky(p_env->sky); + ERR_FAIL_COND(!sky); + + SkyMaterialData *material = nullptr; + RID sky_material; + + sky_material = sky_get_material(p_env->sky); + + if (sky_material.is_valid()) { + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + if (!material || !material->shader_data->valid) { + material = nullptr; + } + } + + if (!material) { + sky_material = sky_shader.default_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + + ERR_FAIL_COND(!material); + + SkyShaderData *shader_data = material->shader_data; + + ERR_FAIL_COND(!shader_data); + + Basis sky_transform = p_env->sky_orientation; + sky_transform.invert(); + + float multiplier = p_env->bg_energy; + float custom_fov = p_env->sky_custom_fov; + + // Camera + CameraMatrix camera; + uint32_t view_count = p_view_count; + const CameraMatrix *projections = p_projections; + + if (custom_fov) { + // With custom fov we don't support stereo... + float near_plane = p_projections[0].get_z_near(); + float far_plane = p_projections[0].get_z_far(); + float aspect = p_projections[0].get_aspect(); + + camera.set_perspective(custom_fov, aspect, near_plane, far_plane); + + view_count = 1; + projections = &camera; + } + + sky_transform = p_transform.basis * sky_transform; + + if (shader_data->uses_quarter_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_QUARTER_RES_MULTIVIEW : SKY_VERSION_QUARTER_RES]; + + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_QUARTER_RES, sky_shader.default_shader_rd); + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->quarter_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); + _render_sky(draw_list, p_time, sky->quarter_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, p_luminance_multiplier); + RD::get_singleton()->draw_list_end(); + } + + if (shader_data->uses_half_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_HALF_RES_MULTIVIEW : SKY_VERSION_HALF_RES]; + + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_HALF_RES, sky_shader.default_shader_rd); + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->half_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); + _render_sky(draw_list, p_time, sky->half_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, p_luminance_multiplier); + RD::get_singleton()->draw_list_end(); + } +} + +void RendererSceneSkyRD::draw(RD::DrawListID p_draw_list, RendererSceneEnvironmentRD *p_env, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time, float p_luminance_multiplier) { + ERR_FAIL_COND(!p_env); + + ERR_FAIL_COND(p_view_count == 0); + ERR_FAIL_COND(p_view_count > RendererSceneRender::MAX_RENDER_VIEWS); + + Sky *sky = get_sky(p_env->sky); + + SkyMaterialData *material = nullptr; + RID sky_material; + + RS::EnvironmentBG background = p_env->background; + + if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { + ERR_FAIL_COND(!sky); + sky_material = sky_get_material(p_env->sky); + + if (sky_material.is_valid()) { + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + if (!material || !material->shader_data->valid) { + material = nullptr; + } + } + + if (!material) { + sky_material = sky_shader.default_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + } + + if (background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) { + sky_material = sky_scene_state.fog_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + + ERR_FAIL_COND(!material); + + SkyShaderData *shader_data = material->shader_data; + + ERR_FAIL_COND(!shader_data); + + Basis sky_transform = p_env->sky_orientation; + sky_transform.invert(); + + float multiplier = p_env->bg_energy; + float custom_fov = p_env->sky_custom_fov; + + // Camera + CameraMatrix camera; + uint32_t view_count = p_view_count; + const CameraMatrix *projections = p_projections; + + if (custom_fov) { + // With custom fov we don't support stereo... + float near_plane = p_projections[0].get_z_near(); + float far_plane = p_projections[0].get_z_far(); + float aspect = p_projections[0].get_aspect(); + + camera.set_perspective(custom_fov, aspect, near_plane, far_plane); + + view_count = 1; + projections = &camera; + } + + sky_transform = p_transform.basis * sky_transform; + + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_BACKGROUND_MULTIVIEW : SKY_VERSION_BACKGROUND]; + + RID texture_uniform_set; + if (sky) { + texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_BACKGROUND, sky_shader.default_shader_rd); + } else { + texture_uniform_set = sky_scene_state.fog_only_texture_uniform_set; + } + + _render_sky(p_draw_list, p_time, p_fb, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin, p_luminance_multiplier); +} + void RendererSceneSkyRD::invalidate_sky(Sky *p_sky) { if (!p_sky->dirty) { p_sky->dirty = true; @@ -1393,7 +1657,7 @@ void RendererSceneSkyRD::update_dirty_skys() { //array (higher quality, 6 times more memory) RD::TextureFormat tf; tf.array_layers = layers * 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = texture_format; tf.texture_type = RD::TEXTURE_TYPE_CUBE_ARRAY; tf.mipmaps = mipmaps; tf.width = w; @@ -1402,13 +1666,13 @@ void RendererSceneSkyRD::update_dirty_skys() { sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); - sky->reflection.update_reflection_data(sky->radiance_size, mipmaps, true, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers); + sky->reflection.update_reflection_data(storage, sky->radiance_size, mipmaps, true, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers, texture_format); } else { //regular cubemap, lower quality (aliasing, less memory) RD::TextureFormat tf; tf.array_layers = 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = texture_format; tf.texture_type = RD::TEXTURE_TYPE_CUBE; tf.mipmaps = MIN(mipmaps, layers); tf.width = w; @@ -1417,7 +1681,7 @@ void RendererSceneSkyRD::update_dirty_skys() { sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); - sky->reflection.update_reflection_data(sky->radiance_size, MIN(mipmaps, layers), false, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers); + sky->reflection.update_reflection_data(storage, sky->radiance_size, MIN(mipmaps, layers), false, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers, texture_format); } texture_set_dirty = true; } @@ -1425,7 +1689,7 @@ void RendererSceneSkyRD::update_dirty_skys() { // Create subpass buffers if they haven't been created already if (sky->half_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->half_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { RD::TextureFormat tformat; - tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tformat.format = texture_format; tformat.width = sky->screen_size.x / 2; tformat.height = sky->screen_size.y / 2; tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; @@ -1440,7 +1704,7 @@ void RendererSceneSkyRD::update_dirty_skys() { if (sky->quarter_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->quarter_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { RD::TextureFormat tformat; - tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tformat.format = texture_format; tformat.width = sky->screen_size.x / 4; tformat.height = sky->screen_size.y / 4; tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; @@ -1490,7 +1754,7 @@ void RendererSceneSkyRD::initialize_sky_rid(RID p_rid) { } RendererSceneSkyRD::Sky *RendererSceneSkyRD::get_sky(RID p_sky) const { - return sky_owner.getornull(p_sky); + return sky_owner.get_or_null(p_sky); } void RendererSceneSkyRD::free_sky(RID p_sky) { diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.h b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h index 4f852e55a7..7f563c9bc4 100644 --- a/servers/rendering/renderer_rd/renderer_scene_sky_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h @@ -64,6 +64,7 @@ public: private: RendererStorageRD *storage; + RD::DataFormat texture_format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; RID index_buffer; RID index_array; @@ -99,7 +100,8 @@ private: float position[3]; // 12 - 92 float multiplier; // 4 - 96 float time; // 4 - 100 - float pad[3]; // 12 - 112 // Using pad to align on 16 bytes + float luminance_multiplier; // 4 - 104 + float pad[2]; // 8 - 112 // Using pad to align on 16 bytes // 128 is the max size of a push constant. We can replace "pad" but we can't add any more. }; @@ -137,7 +139,7 @@ private: virtual ~SkyShaderData(); }; - void _render_sky(RD::DrawListID p_list, float p_time, RID p_fb, PipelineCacheRD *p_pipeline, RID p_uniform_set, RID p_texture_set, uint32_t p_view_count, const CameraMatrix *p_projections, const Basis &p_orientation, float p_multiplier, const Vector3 &p_position); + void _render_sky(RD::DrawListID p_list, float p_time, RID p_fb, PipelineCacheRD *p_pipeline, RID p_uniform_set, RID p_texture_set, uint32_t p_view_count, const CameraMatrix *p_projections, const Basis &p_orientation, float p_multiplier, const Vector3 &p_position, float p_luminance_multiplier); public: struct SkySceneState { @@ -190,6 +192,10 @@ public: struct Mipmap { RID view; Size2i size; + + // for mobile only + RID views[6]; + RID framebuffers[6]; }; Vector<Mipmap> mipmaps; }; @@ -204,7 +210,7 @@ public: Vector<Layer> layers; void clear_reflection_data(); - void update_reflection_data(int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers); + void update_reflection_data(RendererStorageRD *p_storage, int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers, RD::DataFormat p_texture_format); void create_reflection_fast_filter(RendererStorageRD *p_storage, bool p_use_arrays); void create_reflection_importance_sample(RendererStorageRD *p_storage, bool p_use_arrays, int p_cube_side, int p_base_layer, uint32_t p_sky_ggx_samples_quality); void update_reflection_mipmaps(RendererStorageRD *p_storage, int p_start, int p_end); @@ -284,11 +290,14 @@ public: RendererSceneSkyRD(); void init(RendererStorageRD *p_storage); + void set_texture_format(RD::DataFormat p_texture_format); ~RendererSceneSkyRD(); void setup(RendererSceneEnvironmentRD *p_env, RID p_render_buffers, const CameraMatrix &p_projection, const Transform3D &p_transform, const Size2i p_screen_size, RendererSceneRenderRD *p_scene_render); - void update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform3D &p_transform, double p_time); - void draw(RendererSceneEnvironmentRD *p_env, bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time); + void update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform3D &p_transform, double p_time, float p_luminance_multiplier = 1.0); + void draw(RendererSceneEnvironmentRD *p_env, bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time); // only called by clustered renderer + void update_res_buffers(RendererSceneEnvironmentRD *p_env, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time, float p_luminance_multiplier = 1.0); + void draw(RD::DrawListID p_draw_list, RendererSceneEnvironmentRD *p_env, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time, float p_luminance_multiplier = 1.0); void invalidate_sky(Sky *p_sky); void update_dirty_skys(); diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.cpp b/servers/rendering/renderer_rd/renderer_storage_rd.cpp index 2657fa3ae8..a7bfea455d 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_storage_rd.cpp @@ -842,7 +842,7 @@ void RendererStorageRD::texture_3d_initialize(RID p_texture, Image::Format p_for } void RendererStorageRD::texture_proxy_initialize(RID p_texture, RID p_base) { - Texture *tex = texture_owner.getornull(p_base); + Texture *tex = texture_owner.get_or_null(p_base); ERR_FAIL_COND(!tex); Texture proxy_tex = *tex; @@ -865,7 +865,7 @@ void RendererStorageRD::texture_proxy_initialize(RID p_texture, RID p_base) { void RendererStorageRD::_texture_2d_update(RID p_texture, const Ref<Image> &p_image, int p_layer, bool p_immediate) { ERR_FAIL_COND(p_image.is_null() || p_image->is_empty()); - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); ERR_FAIL_COND(tex->is_render_target); ERR_FAIL_COND(p_image->get_width() != tex->width || p_image->get_height() != tex->height); @@ -889,7 +889,7 @@ void RendererStorageRD::texture_2d_update(RID p_texture, const Ref<Image> &p_ima } void RendererStorageRD::texture_3d_update(RID p_texture, const Vector<Ref<Image>> &p_data) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); ERR_FAIL_COND(tex->type != Texture::TYPE_3D); Image::Image3DValidateError verr = Image::validate_3d_image(tex->format, tex->width, tex->height, tex->depth, tex->mipmaps > 1, p_data); @@ -926,10 +926,10 @@ void RendererStorageRD::texture_3d_update(RID p_texture, const Vector<Ref<Image> } void RendererStorageRD::texture_proxy_update(RID p_texture, RID p_proxy_to) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); ERR_FAIL_COND(!tex->is_proxy); - Texture *proxy_to = texture_owner.getornull(p_proxy_to); + Texture *proxy_to = texture_owner.get_or_null(p_proxy_to); ERR_FAIL_COND(!proxy_to); ERR_FAIL_COND(proxy_to->is_proxy); @@ -943,7 +943,7 @@ void RendererStorageRD::texture_proxy_update(RID p_texture, RID p_proxy_to) { RD::get_singleton()->free(tex->rd_texture_srgb); tex->rd_texture_srgb = RID(); } - Texture *prev_tex = texture_owner.getornull(tex->proxy_to); + Texture *prev_tex = texture_owner.get_or_null(tex->proxy_to); ERR_FAIL_COND(!prev_tex); prev_tex->proxies.erase(p_texture); } @@ -1030,7 +1030,7 @@ void RendererStorageRD::texture_3d_placeholder_initialize(RID p_texture) { } Ref<Image> RendererStorageRD::texture_2d_get(RID p_texture) const { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND_V(!tex, Ref<Image>()); #ifdef TOOLS_ENABLED @@ -1058,7 +1058,7 @@ Ref<Image> RendererStorageRD::texture_2d_get(RID p_texture) const { } Ref<Image> RendererStorageRD::texture_2d_layer_get(RID p_texture, int p_layer) const { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND_V(!tex, Ref<Image>()); Vector<uint8_t> data = RD::get_singleton()->texture_get_data(tex->rd_texture, p_layer); @@ -1075,7 +1075,7 @@ Ref<Image> RendererStorageRD::texture_2d_layer_get(RID p_texture, int p_layer) c } Vector<Ref<Image>> RendererStorageRD::texture_3d_get(RID p_texture) const { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND_V(!tex, Vector<Ref<Image>>()); ERR_FAIL_COND_V(tex->type != Texture::TYPE_3D, Vector<Ref<Image>>()); @@ -1106,10 +1106,10 @@ Vector<Ref<Image>> RendererStorageRD::texture_3d_get(RID p_texture) const { } void RendererStorageRD::texture_replace(RID p_texture, RID p_by_texture) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); ERR_FAIL_COND(tex->proxy_to.is_valid()); //can't replace proxy - Texture *by_tex = texture_owner.getornull(p_by_texture); + Texture *by_tex = texture_owner.get_or_null(p_by_texture); ERR_FAIL_COND(!by_tex); ERR_FAIL_COND(by_tex->proxy_to.is_valid()); //can't replace proxy @@ -1155,7 +1155,7 @@ void RendererStorageRD::texture_replace(RID p_texture, RID p_by_texture) { } void RendererStorageRD::texture_set_size_override(RID p_texture, int p_width, int p_height) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); ERR_FAIL_COND(tex->type != Texture::TYPE_2D); tex->width_2d = p_width; @@ -1163,7 +1163,7 @@ void RendererStorageRD::texture_set_size_override(RID p_texture, int p_width, in } void RendererStorageRD::texture_set_path(RID p_texture, const String &p_path) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); tex->path = p_path; } @@ -1173,21 +1173,21 @@ String RendererStorageRD::texture_get_path(RID p_texture) const { } void RendererStorageRD::texture_set_detect_3d_callback(RID p_texture, RS::TextureDetectCallback p_callback, void *p_userdata) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); tex->detect_3d_callback_ud = p_userdata; tex->detect_3d_callback = p_callback; } void RendererStorageRD::texture_set_detect_normal_callback(RID p_texture, RS::TextureDetectCallback p_callback, void *p_userdata) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); tex->detect_normal_callback_ud = p_userdata; tex->detect_normal_callback = p_callback; } void RendererStorageRD::texture_set_detect_roughness_callback(RID p_texture, RS::TextureDetectRoughnessCallback p_callback, void *p_userdata) { - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); ERR_FAIL_COND(!tex); tex->detect_roughness_callback_ud = p_userdata; tex->detect_roughness_callback = p_callback; @@ -1235,7 +1235,7 @@ void RendererStorageRD::canvas_texture_initialize(RID p_rid) { } void RendererStorageRD::canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) { - CanvasTexture *ct = canvas_texture_owner.getornull(p_canvas_texture); + CanvasTexture *ct = canvas_texture_owner.get_or_null(p_canvas_texture); switch (p_channel) { case RS::CANVAS_TEXTURE_CHANNEL_DIFFUSE: { ct->diffuse = p_texture; @@ -1252,7 +1252,7 @@ void RendererStorageRD::canvas_texture_set_channel(RID p_canvas_texture, RS::Can } void RendererStorageRD::canvas_texture_set_shading_parameters(RID p_canvas_texture, const Color &p_specular_color, float p_shininess) { - CanvasTexture *ct = canvas_texture_owner.getornull(p_canvas_texture); + CanvasTexture *ct = canvas_texture_owner.get_or_null(p_canvas_texture); ct->specular_color.r = p_specular_color.r; ct->specular_color.g = p_specular_color.g; ct->specular_color.b = p_specular_color.b; @@ -1261,13 +1261,13 @@ void RendererStorageRD::canvas_texture_set_shading_parameters(RID p_canvas_textu } void RendererStorageRD::canvas_texture_set_texture_filter(RID p_canvas_texture, RS::CanvasItemTextureFilter p_filter) { - CanvasTexture *ct = canvas_texture_owner.getornull(p_canvas_texture); + CanvasTexture *ct = canvas_texture_owner.get_or_null(p_canvas_texture); ct->texture_filter = p_filter; ct->clear_sets(); } void RendererStorageRD::canvas_texture_set_texture_repeat(RID p_canvas_texture, RS::CanvasItemTextureRepeat p_repeat) { - CanvasTexture *ct = canvas_texture_owner.getornull(p_canvas_texture); + CanvasTexture *ct = canvas_texture_owner.get_or_null(p_canvas_texture); ct->texture_repeat = p_repeat; ct->clear_sets(); } @@ -1275,7 +1275,7 @@ void RendererStorageRD::canvas_texture_set_texture_repeat(RID p_canvas_texture, bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::CanvasItemTextureFilter p_base_filter, RS::CanvasItemTextureRepeat p_base_repeat, RID p_base_shader, int p_base_set, RID &r_uniform_set, Size2i &r_size, Color &r_specular_shininess, bool &r_use_normal, bool &r_use_specular) { CanvasTexture *ct = nullptr; - Texture *t = texture_owner.getornull(p_texture); + Texture *t = texture_owner.get_or_null(p_texture); if (t) { //regular texture @@ -1286,7 +1286,7 @@ bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::Canvas ct = t->canvas_texture; } else { - ct = canvas_texture_owner.getornull(p_texture); + ct = canvas_texture_owner.get_or_null(p_texture); } if (!ct) { @@ -1308,7 +1308,7 @@ bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::Canvas u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 0; - t = texture_owner.getornull(ct->diffuse); + t = texture_owner.get_or_null(ct->diffuse); if (!t) { u.ids.push_back(texture_rd_get_default(DEFAULT_RD_TEXTURE_WHITE)); ct->size_cache = Size2i(1, 1); @@ -1323,7 +1323,7 @@ bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::Canvas u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 1; - t = texture_owner.getornull(ct->normal_map); + t = texture_owner.get_or_null(ct->normal_map); if (!t) { u.ids.push_back(texture_rd_get_default(DEFAULT_RD_TEXTURE_NORMAL)); ct->use_normal_cache = false; @@ -1338,7 +1338,7 @@ bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::Canvas u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 2; - t = texture_owner.getornull(ct->specular); + t = texture_owner.get_or_null(ct->specular); if (!t) { u.ids.push_back(texture_rd_get_default(DEFAULT_RD_TEXTURE_WHITE)); ct->use_specular_cache = false; @@ -1384,7 +1384,7 @@ void RendererStorageRD::shader_initialize(RID p_rid) { } void RendererStorageRD::shader_set_code(RID p_shader, const String &p_code) { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND(!shader); shader->code = p_code; @@ -1438,8 +1438,8 @@ void RendererStorageRD::shader_set_code(RID p_shader, const String &p_code) { } if (shader->data) { - for (Map<StringName, RID>::Element *E = shader->default_texture_parameter.front(); E; E = E->next()) { - shader->data->set_default_texture_param(E->key(), E->get()); + for (const KeyValue<StringName, RID> &E : shader->default_texture_parameter) { + shader->data->set_default_texture_param(E.key, E.value); } } } @@ -1456,13 +1456,13 @@ void RendererStorageRD::shader_set_code(RID p_shader, const String &p_code) { } String RendererStorageRD::shader_get_code(RID p_shader) const { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND_V(!shader, String()); return shader->code; } void RendererStorageRD::shader_get_param_list(RID p_shader, List<PropertyInfo> *p_param_list) const { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND(!shader); if (shader->data) { return shader->data->get_param_list(p_param_list); @@ -1470,7 +1470,7 @@ void RendererStorageRD::shader_get_param_list(RID p_shader, List<PropertyInfo> * } void RendererStorageRD::shader_set_default_texture_param(RID p_shader, const StringName &p_name, RID p_texture) { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND(!shader); if (p_texture.is_valid() && texture_owner.owns(p_texture)) { @@ -1488,7 +1488,7 @@ void RendererStorageRD::shader_set_default_texture_param(RID p_shader, const Str } RID RendererStorageRD::shader_get_default_texture_param(RID p_shader, const StringName &p_name) const { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND_V(!shader, RID()); if (shader->default_texture_parameter.has(p_name)) { return shader->default_texture_parameter[p_name]; @@ -1498,7 +1498,7 @@ RID RendererStorageRD::shader_get_default_texture_param(RID p_shader, const Stri } Variant RendererStorageRD::shader_get_param_default(RID p_shader, const StringName &p_param) const { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND_V(!shader, Variant()); if (shader->data) { return shader->data->get_default_parameter(p_param); @@ -1512,7 +1512,7 @@ void RendererStorageRD::shader_set_data_request_function(ShaderType p_shader_typ } RS::ShaderNativeSourceCode RendererStorageRD::shader_get_native_source_code(RID p_shader) const { - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND_V(!shader, RS::ShaderNativeSourceCode()); if (shader->data) { return shader->data->get_native_source_code(); @@ -1527,7 +1527,7 @@ RID RendererStorageRD::material_allocate() { } void RendererStorageRD::material_initialize(RID p_rid) { material_owner.initialize_rid(p_rid); - Material *material = material_owner.getornull(p_rid); + Material *material = material_owner.get_or_null(p_rid); material->self = p_rid; } @@ -1543,7 +1543,7 @@ void RendererStorageRD::_material_queue_update(Material *material, bool p_unifor } void RendererStorageRD::material_set_shader(RID p_material, RID p_shader) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); if (material->data) { @@ -1563,7 +1563,7 @@ void RendererStorageRD::material_set_shader(RID p_material, RID p_shader) { return; } - Shader *shader = shader_owner.getornull(p_shader); + Shader *shader = shader_owner.get_or_null(p_shader); ERR_FAIL_COND(!shader); material->shader = shader; material->shader_type = shader->type; @@ -1586,7 +1586,7 @@ void RendererStorageRD::material_set_shader(RID p_material, RID p_shader) { } void RendererStorageRD::material_set_param(RID p_material, const StringName &p_param, const Variant &p_value) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); if (p_value.get_type() == Variant::NIL) { @@ -1605,7 +1605,7 @@ void RendererStorageRD::material_set_param(RID p_material, const StringName &p_p } Variant RendererStorageRD::material_get_param(RID p_material, const StringName &p_param) const { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND_V(!material, Variant()); if (material->params.has(p_param)) { return material->params[p_param]; @@ -1615,7 +1615,7 @@ Variant RendererStorageRD::material_get_param(RID p_material, const StringName & } void RendererStorageRD::material_set_next_pass(RID p_material, RID p_next_material) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); if (material->next_pass == p_next_material) { @@ -1631,7 +1631,7 @@ void RendererStorageRD::material_set_next_pass(RID p_material, RID p_next_materi } void RendererStorageRD::material_set_render_priority(RID p_material, int priority) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); material->priority = priority; if (material->data) { @@ -1640,7 +1640,7 @@ void RendererStorageRD::material_set_render_priority(RID p_material, int priorit } bool RendererStorageRD::material_is_animated(RID p_material) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND_V(!material, false); if (material->shader && material->shader->data) { if (material->shader->data->is_animated()) { @@ -1653,7 +1653,7 @@ bool RendererStorageRD::material_is_animated(RID p_material) { } bool RendererStorageRD::material_casts_shadows(RID p_material) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND_V(!material, true); if (material->shader && material->shader->data) { if (material->shader->data->casts_shadows()) { @@ -1666,7 +1666,7 @@ bool RendererStorageRD::material_casts_shadows(RID p_material) { } void RendererStorageRD::material_get_instance_shader_parameters(RID p_material, List<InstanceShaderParam> *r_parameters) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); if (material->shader && material->shader->data) { material->shader->data->get_instance_param_list(r_parameters); @@ -1678,7 +1678,7 @@ void RendererStorageRD::material_get_instance_shader_parameters(RID p_material, } void RendererStorageRD::material_update_dependency(RID p_material, DependencyTracker *p_instance) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); ERR_FAIL_COND(!material); p_instance->update_dependency(&material->dependency); if (material->next_pass.is_valid()) { @@ -1691,73 +1691,183 @@ void RendererStorageRD::material_set_data_request_function(ShaderType p_shader_t material_data_request_func[p_shader_type] = p_function; } -_FORCE_INLINE_ static void _fill_std140_variant_ubo_value(ShaderLanguage::DataType type, const Variant &value, uint8_t *data, bool p_linear_color) { +_FORCE_INLINE_ static void _fill_std140_variant_ubo_value(ShaderLanguage::DataType type, int p_array_size, const Variant &value, uint8_t *data, bool p_linear_color) { switch (type) { case ShaderLanguage::TYPE_BOOL: { - bool v = value; - uint32_t *gui = (uint32_t *)data; - *gui = v ? 1 : 0; + + if (p_array_size > 0) { + const PackedInt32Array &ba = value; + int s = ba.size(); + const int *r = ba.ptr(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = (r[i] != 0) ? 1 : 0; + } else { + gui[j] = 0; + } + gui[j + 1] = 0; // ignored + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + bool v = value; + gui[0] = v ? 1 : 0; + } } break; case ShaderLanguage::TYPE_BVEC2: { - int v = value; uint32_t *gui = (uint32_t *)data; - gui[0] = v & 1 ? 1 : 0; - gui[1] = v & 2 ? 1 : 0; + if (p_array_size > 0) { + const PackedInt32Array &ba = value; + int s = ba.size(); + const int *r = ba.ptr(); + int count = 2 * p_array_size; + + for (int i = 0, j = 0; i < count; i += 2, j += 4) { + if (i < s) { + gui[j] = r[i] ? 1 : 0; + gui[j + 1] = r[i + 1] ? 1 : 0; + } else { + gui[j] = 0; + gui[j + 1] = 0; + } + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + int v = value; + gui[0] = v & 1 ? 1 : 0; + gui[1] = v & 2 ? 1 : 0; + } } break; case ShaderLanguage::TYPE_BVEC3: { - int v = value; uint32_t *gui = (uint32_t *)data; - gui[0] = (v & 1) ? 1 : 0; - gui[1] = (v & 2) ? 1 : 0; - gui[2] = (v & 4) ? 1 : 0; + if (p_array_size > 0) { + const PackedInt32Array &ba = value; + int s = ba.size(); + const int *r = ba.ptr(); + int count = 3 * p_array_size; + + for (int i = 0, j = 0; i < count; i += 3, j += 4) { + if (i < s) { + gui[j] = r[i] ? 1 : 0; + gui[j + 1] = r[i + 1] ? 1 : 0; + gui[j + 2] = r[i + 2] ? 1 : 0; + } else { + gui[j] = 0; + gui[j + 1] = 0; + gui[j + 2] = 0; + } + gui[j + 3] = 0; // ignored + } + } else { + int v = value; + gui[0] = (v & 1) ? 1 : 0; + gui[1] = (v & 2) ? 1 : 0; + gui[2] = (v & 4) ? 1 : 0; + } } break; case ShaderLanguage::TYPE_BVEC4: { - int v = value; uint32_t *gui = (uint32_t *)data; - gui[0] = (v & 1) ? 1 : 0; - gui[1] = (v & 2) ? 1 : 0; - gui[2] = (v & 4) ? 1 : 0; - gui[3] = (v & 8) ? 1 : 0; + if (p_array_size > 0) { + const PackedInt32Array &ba = value; + int s = ba.size(); + const int *r = ba.ptr(); + int count = 4 * p_array_size; + + for (int i = 0; i < count; i += 4) { + if (i < s) { + gui[i] = r[i] ? 1 : 0; + gui[i + 1] = r[i + 1] ? 1 : 0; + gui[i + 2] = r[i + 2] ? 1 : 0; + gui[i + 3] = r[i + 3] ? 1 : 0; + } else { + gui[i] = 0; + gui[i + 1] = 0; + gui[i + 2] = 0; + gui[i + 3] = 0; + } + } + } else { + int v = value; + gui[0] = (v & 1) ? 1 : 0; + gui[1] = (v & 2) ? 1 : 0; + gui[2] = (v & 4) ? 1 : 0; + gui[3] = (v & 8) ? 1 : 0; + } } break; case ShaderLanguage::TYPE_INT: { - int v = value; int32_t *gui = (int32_t *)data; - gui[0] = v; + if (p_array_size > 0) { + Vector<int> iv = value; + int s = iv.size(); + const int *r = iv.ptr(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = r[i]; + } else { + gui[j] = 0; + } + gui[j + 1] = 0; // ignored + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + int v = value; + gui[0] = v; + } } break; case ShaderLanguage::TYPE_IVEC2: { Vector<int> iv = value; int s = iv.size(); int32_t *gui = (int32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 2 * p_array_size; - for (int i = 0; i < 2; i++) { + const int *r = iv.ptr(); + for (int i = 0, j = 0; i < count; i += 2, j += 4) { if (i < s) { - gui[i] = r[i]; + gui[j] = r[i]; + gui[j + 1] = r[i + 1]; } else { - gui[i] = 0; + gui[j] = 0; + gui[j + 1] = 0; } + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored } - } break; case ShaderLanguage::TYPE_IVEC3: { Vector<int> iv = value; int s = iv.size(); int32_t *gui = (int32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 3 * p_array_size; - for (int i = 0; i < 3; i++) { + const int *r = iv.ptr(); + for (int i = 0, j = 0; i < count; i += 3, j += 4) { if (i < s) { - gui[i] = r[i]; + gui[j] = r[i]; + gui[j + 1] = r[i + 1]; + gui[j + 2] = r[i + 2]; } else { - gui[i] = 0; + gui[j] = 0; + gui[j + 1] = 0; + gui[j + 2] = 0; } + gui[j + 3] = 0; // ignored } } break; case ShaderLanguage::TYPE_IVEC4: { @@ -1765,35 +1875,70 @@ _FORCE_INLINE_ static void _fill_std140_variant_ubo_value(ShaderLanguage::DataTy int s = iv.size(); int32_t *gui = (int32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 4 * p_array_size; - for (int i = 0; i < 4; i++) { + const int *r = iv.ptr(); + for (int i = 0; i < count; i += 4) { if (i < s) { gui[i] = r[i]; + gui[i + 1] = r[i + 1]; + gui[i + 2] = r[i + 2]; + gui[i + 3] = r[i + 3]; } else { gui[i] = 0; + gui[i + 1] = 0; + gui[i + 2] = 0; + gui[i + 3] = 0; } } } break; case ShaderLanguage::TYPE_UINT: { - int v = value; uint32_t *gui = (uint32_t *)data; - gui[0] = v; + if (p_array_size > 0) { + Vector<int> iv = value; + int s = iv.size(); + const int *r = iv.ptr(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = r[i]; + } else { + gui[j] = 0; + } + gui[j + 1] = 0; // ignored + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + int v = value; + gui[0] = v; + } } break; case ShaderLanguage::TYPE_UVEC2: { Vector<int> iv = value; int s = iv.size(); uint32_t *gui = (uint32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 2 * p_array_size; - for (int i = 0; i < 2; i++) { + const int *r = iv.ptr(); + for (int i = 0, j = 0; i < count; i += 2, j += 4) { if (i < s) { - gui[i] = r[i]; + gui[j] = r[i]; + gui[j + 1] = r[i + 1]; } else { - gui[i] = 0; + gui[j] = 0; + gui[j + 1] = 0; } + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored } } break; case ShaderLanguage::TYPE_UVEC3: { @@ -1801,141 +1946,370 @@ _FORCE_INLINE_ static void _fill_std140_variant_ubo_value(ShaderLanguage::DataTy int s = iv.size(); uint32_t *gui = (uint32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 3 * p_array_size; - for (int i = 0; i < 3; i++) { + const int *r = iv.ptr(); + for (int i = 0, j = 0; i < count; i += 3, j += 4) { if (i < s) { - gui[i] = r[i]; + gui[j] = r[i]; + gui[j + 1] = r[i + 1]; + gui[j + 2] = r[i + 2]; } else { - gui[i] = 0; + gui[j] = 0; + gui[j + 1] = 0; + gui[j + 2] = 0; } + gui[j + 3] = 0; // ignored } - } break; case ShaderLanguage::TYPE_UVEC4: { Vector<int> iv = value; int s = iv.size(); uint32_t *gui = (uint32_t *)data; - const int *r = iv.ptr(); + if (p_array_size <= 0) { + p_array_size = 1; + } + int count = 4 * p_array_size; - for (int i = 0; i < 4; i++) { + const int *r = iv.ptr(); + for (int i = 0; i < count; i++) { if (i < s) { gui[i] = r[i]; + gui[i + 1] = r[i + 1]; + gui[i + 2] = r[i + 2]; + gui[i + 3] = r[i + 3]; } else { gui[i] = 0; + gui[i + 1] = 0; + gui[i + 2] = 0; + gui[i + 3] = 0; } } } break; case ShaderLanguage::TYPE_FLOAT: { - float v = value; float *gui = (float *)data; - gui[0] = v; + if (p_array_size > 0) { + const PackedFloat32Array &a = value; + int s = a.size(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = a[i]; + } else { + gui[j] = 0; + } + gui[j + 1] = 0; // ignored + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + float v = value; + gui[0] = v; + } } break; case ShaderLanguage::TYPE_VEC2: { - Vector2 v = value; float *gui = (float *)data; - gui[0] = v.x; - gui[1] = v.y; + if (p_array_size > 0) { + const PackedVector2Array &a = value; + int s = a.size(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = a[i].x; + gui[j + 1] = a[i].y; + } else { + gui[j] = 0; + gui[j + 1] = 0; + } + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + } + } else { + Vector2 v = value; + gui[0] = v.x; + gui[1] = v.y; + } } break; case ShaderLanguage::TYPE_VEC3: { - Vector3 v = value; float *gui = (float *)data; - gui[0] = v.x; - gui[1] = v.y; - gui[2] = v.z; + if (p_array_size > 0) { + const PackedVector3Array &a = value; + int s = a.size(); + + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + gui[j] = a[i].x; + gui[j + 1] = a[i].y; + gui[j + 2] = a[i].z; + } else { + gui[j] = 0; + gui[j + 1] = 0; + gui[j + 2] = 0; + } + gui[j + 3] = 0; // ignored + } + } else { + Vector3 v = value; + gui[0] = v.x; + gui[1] = v.y; + gui[2] = v.z; + } } break; case ShaderLanguage::TYPE_VEC4: { float *gui = (float *)data; - if (value.get_type() == Variant::COLOR) { - Color v = value; + if (p_array_size > 0) { + if (value.get_type() == Variant::PACKED_COLOR_ARRAY) { + const PackedColorArray &a = value; + int s = a.size(); - if (p_linear_color) { - v = v.to_linear(); + for (int i = 0, j = 0; i < p_array_size; i++, j += 4) { + if (i < s) { + Color color = a[i]; + if (p_linear_color) { + color = color.to_linear(); + } + gui[j] = color.r; + gui[j + 1] = color.g; + gui[j + 2] = color.b; + gui[j + 3] = color.a; + } else { + gui[j] = 0; + gui[j + 1] = 0; + gui[j + 2] = 0; + gui[j + 3] = 0; + } + } + } else { + const PackedFloat32Array &a = value; + int s = a.size(); + int count = 4 * p_array_size; + + for (int i = 0; i < count; i += 4) { + if (i + 3 < s) { + gui[i] = a[i]; + gui[i + 1] = a[i + 1]; + gui[i + 2] = a[i + 2]; + gui[i + 3] = a[i + 3]; + } else { + gui[i] = 0; + gui[i + 1] = 0; + gui[i + 2] = 0; + gui[i + 3] = 0; + } + } } + } else { + if (value.get_type() == Variant::COLOR) { + Color v = value; - gui[0] = v.r; - gui[1] = v.g; - gui[2] = v.b; - gui[3] = v.a; - } else if (value.get_type() == Variant::RECT2) { - Rect2 v = value; - - gui[0] = v.position.x; - gui[1] = v.position.y; - gui[2] = v.size.x; - gui[3] = v.size.y; - } else if (value.get_type() == Variant::QUATERNION) { - Quaternion v = value; + if (p_linear_color) { + v = v.to_linear(); + } - gui[0] = v.x; - gui[1] = v.y; - gui[2] = v.z; - gui[3] = v.w; - } else { - Plane v = value; + gui[0] = v.r; + gui[1] = v.g; + gui[2] = v.b; + gui[3] = v.a; + } else if (value.get_type() == Variant::RECT2) { + Rect2 v = value; + + gui[0] = v.position.x; + gui[1] = v.position.y; + gui[2] = v.size.x; + gui[3] = v.size.y; + } else if (value.get_type() == Variant::QUATERNION) { + Quaternion v = value; + + gui[0] = v.x; + gui[1] = v.y; + gui[2] = v.z; + gui[3] = v.w; + } else { + Plane v = value; - gui[0] = v.normal.x; - gui[1] = v.normal.y; - gui[2] = v.normal.z; - gui[3] = v.d; + gui[0] = v.normal.x; + gui[1] = v.normal.y; + gui[2] = v.normal.z; + gui[3] = v.d; + } } } break; case ShaderLanguage::TYPE_MAT2: { - Transform2D v = value; float *gui = (float *)data; - //in std140 members of mat2 are treated as vec4s - gui[0] = v.elements[0][0]; - gui[1] = v.elements[0][1]; - gui[2] = 0; - gui[3] = 0; - gui[4] = v.elements[1][0]; - gui[5] = v.elements[1][1]; - gui[6] = 0; - gui[7] = 0; + if (p_array_size > 0) { + const PackedFloat32Array &a = value; + int s = a.size(); + + for (int i = 0, j = 0; i < p_array_size * 4; i += 4, j += 8) { + if (i + 3 < s) { + gui[j] = a[i]; + gui[j + 1] = a[i + 1]; + + gui[j + 4] = a[i + 2]; + gui[j + 5] = a[i + 3]; + } else { + gui[j] = 1; + gui[j + 1] = 0; + + gui[j + 4] = 0; + gui[j + 5] = 1; + } + gui[j + 2] = 0; // ignored + gui[j + 3] = 0; // ignored + gui[j + 6] = 0; // ignored + gui[j + 7] = 0; // ignored + } + } else { + Transform2D v = value; + + //in std140 members of mat2 are treated as vec4s + gui[0] = v.elements[0][0]; + gui[1] = v.elements[0][1]; + gui[2] = 0; // ignored + gui[3] = 0; // ignored + + gui[4] = v.elements[1][0]; + gui[5] = v.elements[1][1]; + gui[6] = 0; // ignored + gui[7] = 0; // ignored + } } break; case ShaderLanguage::TYPE_MAT3: { - Basis v = value; float *gui = (float *)data; - gui[0] = v.elements[0][0]; - gui[1] = v.elements[1][0]; - gui[2] = v.elements[2][0]; - gui[3] = 0; - gui[4] = v.elements[0][1]; - gui[5] = v.elements[1][1]; - gui[6] = v.elements[2][1]; - gui[7] = 0; - gui[8] = v.elements[0][2]; - gui[9] = v.elements[1][2]; - gui[10] = v.elements[2][2]; - gui[11] = 0; + if (p_array_size > 0) { + const PackedFloat32Array &a = value; + int s = a.size(); + + for (int i = 0, j = 0; i < p_array_size * 9; i += 9, j += 12) { + if (i + 8 < s) { + gui[j] = a[i]; + gui[j + 1] = a[i + 1]; + gui[j + 2] = a[i + 2]; + + gui[j + 4] = a[i + 3]; + gui[j + 5] = a[i + 4]; + gui[j + 6] = a[i + 5]; + + gui[j + 8] = a[i + 6]; + gui[j + 9] = a[i + 7]; + gui[j + 10] = a[i + 8]; + } else { + gui[j] = 1; + gui[j + 1] = 0; + gui[j + 2] = 0; + + gui[j + 4] = 0; + gui[j + 5] = 1; + gui[j + 6] = 0; + + gui[j + 8] = 0; + gui[j + 9] = 0; + gui[j + 10] = 1; + } + gui[j + 3] = 0; // ignored + gui[j + 7] = 0; // ignored + gui[j + 11] = 0; // ignored + } + } else { + Basis v = value; + gui[0] = v.elements[0][0]; + gui[1] = v.elements[1][0]; + gui[2] = v.elements[2][0]; + gui[3] = 0; // ignored + + gui[4] = v.elements[0][1]; + gui[5] = v.elements[1][1]; + gui[6] = v.elements[2][1]; + gui[7] = 0; // ignored + + gui[8] = v.elements[0][2]; + gui[9] = v.elements[1][2]; + gui[10] = v.elements[2][2]; + gui[11] = 0; // ignored + } } break; case ShaderLanguage::TYPE_MAT4: { - Transform3D v = value; float *gui = (float *)data; - gui[0] = v.basis.elements[0][0]; - gui[1] = v.basis.elements[1][0]; - gui[2] = v.basis.elements[2][0]; - gui[3] = 0; - gui[4] = v.basis.elements[0][1]; - gui[5] = v.basis.elements[1][1]; - gui[6] = v.basis.elements[2][1]; - gui[7] = 0; - gui[8] = v.basis.elements[0][2]; - gui[9] = v.basis.elements[1][2]; - gui[10] = v.basis.elements[2][2]; - gui[11] = 0; - gui[12] = v.origin.x; - gui[13] = v.origin.y; - gui[14] = v.origin.z; - gui[15] = 1; + if (p_array_size > 0) { + const PackedFloat32Array &a = value; + int s = a.size(); + + for (int i = 0; i < p_array_size * 16; i += 16) { + if (i + 15 < s) { + gui[i] = a[i]; + gui[i + 1] = a[i + 1]; + gui[i + 2] = a[i + 2]; + gui[i + 3] = a[i + 3]; + + gui[i + 4] = a[i + 4]; + gui[i + 5] = a[i + 5]; + gui[i + 6] = a[i + 6]; + gui[i + 7] = a[i + 7]; + + gui[i + 8] = a[i + 8]; + gui[i + 9] = a[i + 9]; + gui[i + 10] = a[i + 10]; + gui[i + 11] = a[i + 11]; + + gui[i + 12] = a[i + 12]; + gui[i + 13] = a[i + 13]; + gui[i + 14] = a[i + 14]; + gui[i + 15] = a[i + 15]; + } else { + gui[i] = 1; + gui[i + 1] = 0; + gui[i + 2] = 0; + gui[i + 3] = 0; + + gui[i + 4] = 0; + gui[i + 5] = 1; + gui[i + 6] = 0; + gui[i + 7] = 0; + + gui[i + 8] = 0; + gui[i + 9] = 0; + gui[i + 10] = 1; + gui[i + 11] = 0; + + gui[i + 12] = 0; + gui[i + 13] = 0; + gui[i + 14] = 0; + gui[i + 15] = 1; + } + } + } else { + Transform3D v = value; + gui[0] = v.basis.elements[0][0]; + gui[1] = v.basis.elements[1][0]; + gui[2] = v.basis.elements[2][0]; + gui[3] = 0; + + gui[4] = v.basis.elements[0][1]; + gui[5] = v.basis.elements[1][1]; + gui[6] = v.basis.elements[2][1]; + gui[7] = 0; + + gui[8] = v.basis.elements[0][2]; + gui[9] = v.basis.elements[1][2]; + gui[10] = v.basis.elements[2][2]; + gui[11] = 0; + + gui[12] = v.origin.x; + gui[13] = v.origin.y; + gui[14] = v.origin.z; + gui[15] = 1; + } } break; default: { } @@ -2094,19 +2468,23 @@ _FORCE_INLINE_ static void _fill_std140_ubo_value(ShaderLanguage::DataType type, } } -_FORCE_INLINE_ static void _fill_std140_ubo_empty(ShaderLanguage::DataType type, uint8_t *data) { +_FORCE_INLINE_ static void _fill_std140_ubo_empty(ShaderLanguage::DataType type, int p_array_size, uint8_t *data) { + if (p_array_size <= 0) { + p_array_size = 1; + } + switch (type) { case ShaderLanguage::TYPE_BOOL: case ShaderLanguage::TYPE_INT: case ShaderLanguage::TYPE_UINT: case ShaderLanguage::TYPE_FLOAT: { - memset(data, 0, 4); + memset(data, 0, 4 * p_array_size); } break; case ShaderLanguage::TYPE_BVEC2: case ShaderLanguage::TYPE_IVEC2: case ShaderLanguage::TYPE_UVEC2: case ShaderLanguage::TYPE_VEC2: { - memset(data, 0, 8); + memset(data, 0, 8 * p_array_size); } break; case ShaderLanguage::TYPE_BVEC3: case ShaderLanguage::TYPE_IVEC3: @@ -2116,16 +2494,16 @@ _FORCE_INLINE_ static void _fill_std140_ubo_empty(ShaderLanguage::DataType type, case ShaderLanguage::TYPE_IVEC4: case ShaderLanguage::TYPE_UVEC4: case ShaderLanguage::TYPE_VEC4: { - memset(data, 0, 16); + memset(data, 0, 16 * p_array_size); } break; case ShaderLanguage::TYPE_MAT2: { - memset(data, 0, 32); + memset(data, 0, 32 * p_array_size); } break; case ShaderLanguage::TYPE_MAT3: { - memset(data, 0, 48); + memset(data, 0, 48 * p_array_size); } break; case ShaderLanguage::TYPE_MAT4: { - memset(data, 0, 64); + memset(data, 0, 64 * p_array_size); } break; default: { @@ -2136,28 +2514,28 @@ _FORCE_INLINE_ static void _fill_std140_ubo_empty(ShaderLanguage::DataType type, void RendererStorageRD::MaterialData::update_uniform_buffer(const Map<StringName, ShaderLanguage::ShaderNode::Uniform> &p_uniforms, const uint32_t *p_uniform_offsets, const Map<StringName, Variant> &p_parameters, uint8_t *p_buffer, uint32_t p_buffer_size, bool p_use_linear_color) { bool uses_global_buffer = false; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = p_uniforms.front(); E; E = E->next()) { - if (E->get().order < 0) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : p_uniforms) { + if (E.value.order < 0) { continue; // texture, does not go here } - if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + if (E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; //instance uniforms don't appear in the bufferr } - if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL) { + if (E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL) { //this is a global variable, get the index to it RendererStorageRD *rs = base_singleton; - GlobalVariables::Variable *gv = rs->global_variables.variables.getptr(E->key()); + GlobalVariables::Variable *gv = rs->global_variables.variables.getptr(E.key); uint32_t index = 0; if (gv) { index = gv->buffer_index; } else { - WARN_PRINT("Shader uses global uniform '" + E->key() + "', but it was removed at some point. Material will not display correctly."); + WARN_PRINT("Shader uses global uniform '" + E.key + "', but it was removed at some point. Material will not display correctly."); } - uint32_t offset = p_uniform_offsets[E->get().order]; + uint32_t offset = p_uniform_offsets[E.value.order]; uint32_t *intptr = (uint32_t *)&p_buffer[offset]; *intptr = index; uses_global_buffer = true; @@ -2165,30 +2543,30 @@ void RendererStorageRD::MaterialData::update_uniform_buffer(const Map<StringName } //regular uniform - uint32_t offset = p_uniform_offsets[E->get().order]; + uint32_t offset = p_uniform_offsets[E.value.order]; #ifdef DEBUG_ENABLED - uint32_t size = ShaderLanguage::get_type_size(E->get().type); + uint32_t size = ShaderLanguage::get_type_size(E.value.type); ERR_CONTINUE(offset + size > p_buffer_size); #endif uint8_t *data = &p_buffer[offset]; - const Map<StringName, Variant>::Element *V = p_parameters.find(E->key()); + const Map<StringName, Variant>::Element *V = p_parameters.find(E.key); if (V) { //user provided - _fill_std140_variant_ubo_value(E->get().type, V->get(), data, p_use_linear_color); + _fill_std140_variant_ubo_value(E.value.type, E.value.array_size, V->get(), data, p_use_linear_color); - } else if (E->get().default_value.size()) { + } else if (E.value.default_value.size()) { //default value - _fill_std140_ubo_value(E->get().type, E->get().default_value, data); - //value=E->get().default_value; + _fill_std140_ubo_value(E.value.type, E.value.default_value, data); + //value=E.value.default_value; } else { //zero because it was not provided - if (E->get().type == ShaderLanguage::TYPE_VEC4 && E->get().hint == ShaderLanguage::ShaderNode::Uniform::HINT_COLOR) { + if (E.value.type == ShaderLanguage::TYPE_VEC4 && E.value.hint == ShaderLanguage::ShaderNode::Uniform::HINT_COLOR) { //colors must be set as black, with alpha as 1.0 - _fill_std140_variant_ubo_value(E->get().type, Color(0, 0, 0, 1), data, p_use_linear_color); + _fill_std140_variant_ubo_value(E.value.type, E.value.array_size, Color(0, 0, 0, 1), data, p_use_linear_color); } else { //else just zero it out - _fill_std140_ubo_empty(E->get().type, data); + _fill_std140_ubo_empty(E.value.type, E.value.array_size, data); } } } @@ -2215,8 +2593,8 @@ RendererStorageRD::MaterialData::~MaterialData() { //unregister global textures RendererStorageRD *rs = base_singleton; - for (Map<StringName, uint64_t>::Element *E = used_global_textures.front(); E; E = E->next()) { - GlobalVariables::Variable *v = rs->global_variables.variables.getptr(E->key()); + for (const KeyValue<StringName, uint64_t> &E : used_global_textures) { + GlobalVariables::Variable *v = rs->global_variables.variables.getptr(E.key); if (v) { v->texture_materials.erase(self); } @@ -2241,10 +2619,11 @@ void RendererStorageRD::MaterialData::update_textures(const Map<StringName, Vari bool uses_global_textures = false; global_textures_pass++; - for (int i = 0; i < p_texture_uniforms.size(); i++) { + for (int i = 0, k = 0; i < p_texture_uniforms.size(); i++) { const StringName &uniform_name = p_texture_uniforms[i].name; + int uniform_array_size = p_texture_uniforms[i].array_size; - RID texture; + Vector<RID> textures; if (p_texture_uniforms[i].global) { RendererStorageRD *rs = base_singleton; @@ -2265,31 +2644,51 @@ void RendererStorageRD::MaterialData::update_textures(const Map<StringName, Vari E->get() = global_textures_pass; } - texture = v->override.get_type() != Variant::NIL ? v->override : v->value; + textures.push_back(v->override.get_type() != Variant::NIL ? v->override : v->value); } } else { WARN_PRINT("Shader uses global uniform texture '" + String(uniform_name) + "', but it was removed at some point. Material will not display correctly."); } } else { - if (!texture.is_valid()) { - const Map<StringName, Variant>::Element *V = p_parameters.find(uniform_name); - if (V) { - texture = V->get(); + const Map<StringName, Variant>::Element *V = p_parameters.find(uniform_name); + if (V) { + if (V->get().is_array()) { + Array array = (Array)V->get(); + if (uniform_array_size > 0) { + for (int j = 0; j < array.size(); j++) { + textures.push_back(array[j]); + } + } else { + if (array.size() > 0) { + textures.push_back(array[0]); + } + } + } else { + textures.push_back(V->get()); } } - if (!texture.is_valid()) { + if (uniform_array_size > 0) { + if (textures.size() < uniform_array_size) { + const Map<StringName, RID>::Element *W = p_default_textures.find(uniform_name); + if (W) { + for (int j = textures.size(); j < uniform_array_size; j++) { + textures.push_back(W->get()); + } + } + } + } else if (textures.is_empty()) { const Map<StringName, RID>::Element *W = p_default_textures.find(uniform_name); if (W) { - texture = W->get(); + textures.push_back(W->get()); } } } RID rd_texture; - if (texture.is_null()) { + if (textures.is_empty()) { //check default usage switch (p_texture_uniforms[i].hint) { case ShaderLanguage::ShaderNode::Uniform::HINT_BLACK: @@ -2306,45 +2705,55 @@ void RendererStorageRD::MaterialData::update_textures(const Map<StringName, Vari rd_texture = singleton->texture_rd_get_default(DEFAULT_RD_TEXTURE_WHITE); } break; } +#ifdef TOOLS_ENABLED + if (roughness_detect_texture && normal_detect_texture && normal_detect_texture->path != String()) { + roughness_detect_texture->detect_roughness_callback(roughness_detect_texture->detect_roughness_callback_ud, normal_detect_texture->path, roughness_channel); + } +#endif + if (uniform_array_size > 0) { + for (int j = 0; j < uniform_array_size; j++) { + p_textures[k++] = rd_texture; + } + } else { + p_textures[k++] = rd_texture; + } } else { bool srgb = p_use_linear_color && (p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_ALBEDO || p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_BLACK_ALBEDO); - Texture *tex = singleton->texture_owner.getornull(texture); + for (int j = 0; j < textures.size(); j++) { + Texture *tex = singleton->texture_owner.get_or_null(textures[j]); - if (tex) { - rd_texture = (srgb && tex->rd_texture_srgb.is_valid()) ? tex->rd_texture_srgb : tex->rd_texture; + if (tex) { + rd_texture = (srgb && tex->rd_texture_srgb.is_valid()) ? tex->rd_texture_srgb : tex->rd_texture; #ifdef TOOLS_ENABLED - if (tex->detect_3d_callback && p_use_linear_color) { - tex->detect_3d_callback(tex->detect_3d_callback_ud); - } - if (tex->detect_normal_callback && (p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_NORMAL || p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_NORMAL)) { - if (p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_NORMAL) { - normal_detect_texture = tex; + if (tex->detect_3d_callback && p_use_linear_color) { + tex->detect_3d_callback(tex->detect_3d_callback_ud); + } + if (tex->detect_normal_callback && (p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_NORMAL || p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_NORMAL)) { + if (p_texture_uniforms[i].hint == ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_NORMAL) { + normal_detect_texture = tex; + } + tex->detect_normal_callback(tex->detect_normal_callback_ud); + } + if (tex->detect_roughness_callback && (p_texture_uniforms[i].hint >= ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_R || p_texture_uniforms[i].hint <= ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_GRAY)) { + //find the normal texture + roughness_detect_texture = tex; + roughness_channel = RS::TextureDetectRoughnessChannel(p_texture_uniforms[i].hint - ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_R); } - tex->detect_normal_callback(tex->detect_normal_callback_ud); +#endif } - if (tex->detect_roughness_callback && (p_texture_uniforms[i].hint >= ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_R || p_texture_uniforms[i].hint <= ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_GRAY)) { - //find the normal texture - roughness_detect_texture = tex; - roughness_channel = RS::TextureDetectRoughnessChannel(p_texture_uniforms[i].hint - ShaderLanguage::ShaderNode::Uniform::HINT_ROUGHNESS_R); + if (rd_texture.is_null()) { + rd_texture = singleton->texture_rd_get_default(DEFAULT_RD_TEXTURE_WHITE); + } +#ifdef TOOLS_ENABLED + if (roughness_detect_texture && normal_detect_texture && normal_detect_texture->path != String()) { + roughness_detect_texture->detect_roughness_callback(roughness_detect_texture->detect_roughness_callback_ud, normal_detect_texture->path, roughness_channel); } - #endif - } - - if (rd_texture.is_null()) { - //wtf - rd_texture = singleton->texture_rd_get_default(DEFAULT_RD_TEXTURE_WHITE); + p_textures[k++] = rd_texture; } } - - p_textures[i] = rd_texture; } -#ifdef TOOLS_ENABLED - if (roughness_detect_texture && normal_detect_texture && normal_detect_texture->path != String()) { - roughness_detect_texture->detect_roughness_callback(roughness_detect_texture->detect_roughness_callback_ud, normal_detect_texture->path, roughness_channel); - } -#endif { //for textures no longer used, unregister them List<Map<StringName, uint64_t>::Element *> to_delete; @@ -2412,7 +2821,10 @@ bool RendererStorageRD::MaterialData::update_parameters_uniform_set(const Map<St RD::get_singleton()->buffer_update(uniform_buffer, 0, ubo_data.size(), ubo_data.ptrw(), p_barrier); } - uint32_t tex_uniform_count = p_texture_uniforms.size(); + uint32_t tex_uniform_count = 0U; + for (int i = 0; i < p_texture_uniforms.size(); i++) { + tex_uniform_count += uint32_t(p_texture_uniforms[i].array_size > 0 ? p_texture_uniforms[i].array_size : 1); + } if ((uint32_t)texture_cache.size() != tex_uniform_count || p_textures_dirty) { texture_cache.resize(tex_uniform_count); @@ -2452,11 +2864,19 @@ bool RendererStorageRD::MaterialData::update_parameters_uniform_set(const Map<St } const RID *textures = texture_cache.ptrw(); - for (uint32_t i = 0; i < tex_uniform_count; i++) { + for (int i = 0, k = 0; i < p_texture_uniforms.size(); i++) { + const int array_size = p_texture_uniforms[i].array_size; + RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 1 + i; - u.ids.push_back(textures[i]); + u.binding = 1 + k; + if (array_size > 0) { + for (int j = 0; j < array_size; j++) { + u.ids.push_back(textures[k++]); + } + } else { + u.ids.push_back(textures[k++]); + } uniforms.push_back(u); } } @@ -2470,14 +2890,14 @@ bool RendererStorageRD::MaterialData::update_parameters_uniform_set(const Map<St void RendererStorageRD::_material_uniform_set_erased(const RID &p_set, void *p_material) { RID rid = *(RID *)p_material; - Material *material = base_singleton->material_owner.getornull(rid); + Material *material = base_singleton->material_owner.get_or_null(rid); if (material) { material->dependency.changed_notify(DEPENDENCY_CHANGED_MATERIAL); } } void RendererStorageRD::material_force_update_textures(RID p_material, ShaderType p_shader_type) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); if (material->shader_type != p_shader_type) { return; } @@ -2518,7 +2938,7 @@ void RendererStorageRD::mesh_initialize(RID p_rid) { void RendererStorageRD::mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape_count) { ERR_FAIL_COND(p_blend_shape_count < 0); - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_COND(mesh->surface_count > 0); //surfaces already exist @@ -2528,7 +2948,7 @@ void RendererStorageRD::mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape /// Returns stride void RendererStorageRD::mesh_add_surface(RID p_mesh, const RS::SurfaceData &p_surface) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_COND(mesh->surface_count == RS::MAX_MESH_SURFACES); @@ -2716,9 +3136,7 @@ void RendererStorageRD::mesh_add_surface(RID p_mesh, const RS::SurfaceData &p_su mesh->surfaces[mesh->surface_count] = s; mesh->surface_count++; - for (List<MeshInstance *>::Element *E = mesh->instances.front(); E; E = E->next()) { - //update instances - MeshInstance *mi = E->get(); + for (MeshInstance *mi : mesh->instances) { _mesh_instance_add_surface(mi, mesh, mesh->surface_count - 1); } @@ -2734,13 +3152,13 @@ void RendererStorageRD::mesh_add_surface(RID p_mesh, const RS::SurfaceData &p_su } int RendererStorageRD::mesh_get_blend_shape_count(RID p_mesh) const { - const Mesh *mesh = mesh_owner.getornull(p_mesh); + const Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, -1); return mesh->blend_shape_count; } void RendererStorageRD::mesh_set_blend_shape_mode(RID p_mesh, RS::BlendShapeMode p_mode) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_INDEX((int)p_mode, 2); @@ -2748,13 +3166,13 @@ void RendererStorageRD::mesh_set_blend_shape_mode(RID p_mesh, RS::BlendShapeMode } RS::BlendShapeMode RendererStorageRD::mesh_get_blend_shape_mode(RID p_mesh) const { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, RS::BLEND_SHAPE_MODE_NORMALIZED); return mesh->blend_shape_mode; } void RendererStorageRD::mesh_surface_update_vertex_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_UNSIGNED_INDEX((uint32_t)p_surface, mesh->surface_count); ERR_FAIL_COND(p_data.size() == 0); @@ -2765,7 +3183,7 @@ void RendererStorageRD::mesh_surface_update_vertex_region(RID p_mesh, int p_surf } void RendererStorageRD::mesh_surface_update_attribute_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_UNSIGNED_INDEX((uint32_t)p_surface, mesh->surface_count); ERR_FAIL_COND(p_data.size() == 0); @@ -2777,7 +3195,7 @@ void RendererStorageRD::mesh_surface_update_attribute_region(RID p_mesh, int p_s } void RendererStorageRD::mesh_surface_update_skin_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_UNSIGNED_INDEX((uint32_t)p_surface, mesh->surface_count); ERR_FAIL_COND(p_data.size() == 0); @@ -2789,7 +3207,7 @@ void RendererStorageRD::mesh_surface_update_skin_region(RID p_mesh, int p_surfac } void RendererStorageRD::mesh_surface_set_material(RID p_mesh, int p_surface, RID p_material) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); ERR_FAIL_UNSIGNED_INDEX((uint32_t)p_surface, mesh->surface_count); mesh->surfaces[p_surface]->material = p_material; @@ -2799,7 +3217,7 @@ void RendererStorageRD::mesh_surface_set_material(RID p_mesh, int p_surface, RID } RID RendererStorageRD::mesh_surface_get_material(RID p_mesh, int p_surface) const { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, RID()); ERR_FAIL_UNSIGNED_INDEX_V((uint32_t)p_surface, mesh->surface_count, RID()); @@ -2807,7 +3225,7 @@ RID RendererStorageRD::mesh_surface_get_material(RID p_mesh, int p_surface) cons } RS::SurfaceData RendererStorageRD::mesh_get_surface(RID p_mesh, int p_surface) const { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, RS::SurfaceData()); ERR_FAIL_UNSIGNED_INDEX_V((uint32_t)p_surface, mesh->surface_count, RS::SurfaceData()); @@ -2847,32 +3265,32 @@ RS::SurfaceData RendererStorageRD::mesh_get_surface(RID p_mesh, int p_surface) c } int RendererStorageRD::mesh_get_surface_count(RID p_mesh) const { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, 0); return mesh->surface_count; } void RendererStorageRD::mesh_set_custom_aabb(RID p_mesh, const AABB &p_aabb) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); mesh->custom_aabb = p_aabb; } AABB RendererStorageRD::mesh_get_custom_aabb(RID p_mesh) const { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, AABB()); return mesh->custom_aabb; } AABB RendererStorageRD::mesh_get_aabb(RID p_mesh, RID p_skeleton) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, AABB()); if (mesh->custom_aabb != AABB()) { return mesh->custom_aabb; } - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); if (!skeleton || skeleton->size == 0) { return mesh->aabb; @@ -2970,16 +3388,16 @@ AABB RendererStorageRD::mesh_get_aabb(RID p_mesh, RID p_skeleton) { } void RendererStorageRD::mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); - Mesh *shadow_mesh = mesh_owner.getornull(mesh->shadow_mesh); + Mesh *shadow_mesh = mesh_owner.get_or_null(mesh->shadow_mesh); if (shadow_mesh) { shadow_mesh->shadow_owners.erase(mesh); } mesh->shadow_mesh = p_shadow_mesh; - shadow_mesh = mesh_owner.getornull(mesh->shadow_mesh); + shadow_mesh = mesh_owner.get_or_null(mesh->shadow_mesh); if (shadow_mesh) { shadow_mesh->shadow_owners.insert(mesh); @@ -2989,7 +3407,7 @@ void RendererStorageRD::mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) { } void RendererStorageRD::mesh_clear(RID p_mesh) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND(!mesh); for (uint32_t i = 0; i < mesh->surface_count; i++) { Mesh::Surface &s = *mesh->surfaces[i]; @@ -3029,8 +3447,7 @@ void RendererStorageRD::mesh_clear(RID p_mesh) { mesh->surface_count = 0; mesh->material_cache.clear(); //clear instance data - for (List<MeshInstance *>::Element *E = mesh->instances.front(); E; E = E->next()) { - MeshInstance *mi = E->get(); + for (MeshInstance *mi : mesh->instances) { _mesh_instance_clear(mi); } mesh->has_bone_weights = false; @@ -3044,7 +3461,7 @@ void RendererStorageRD::mesh_clear(RID p_mesh) { } bool RendererStorageRD::mesh_needs_instance(RID p_mesh, bool p_has_skeleton) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, false); return mesh->blend_shape_count > 0 || (mesh->has_bone_weights && p_has_skeleton); @@ -3053,11 +3470,11 @@ bool RendererStorageRD::mesh_needs_instance(RID p_mesh, bool p_has_skeleton) { /* MESH INSTANCE */ RID RendererStorageRD::mesh_instance_create(RID p_base) { - Mesh *mesh = mesh_owner.getornull(p_base); + Mesh *mesh = mesh_owner.get_or_null(p_base); ERR_FAIL_COND_V(!mesh, RID()); RID rid = mesh_instance_owner.make_rid(); - MeshInstance *mi = mesh_instance_owner.getornull(rid); + MeshInstance *mi = mesh_instance_owner.get_or_null(rid); mi->mesh = mesh; @@ -3072,7 +3489,7 @@ RID RendererStorageRD::mesh_instance_create(RID p_base) { return rid; } void RendererStorageRD::mesh_instance_set_skeleton(RID p_mesh_instance, RID p_skeleton) { - MeshInstance *mi = mesh_instance_owner.getornull(p_mesh_instance); + MeshInstance *mi = mesh_instance_owner.get_or_null(p_mesh_instance); if (mi->skeleton == p_skeleton) { return; } @@ -3082,7 +3499,7 @@ void RendererStorageRD::mesh_instance_set_skeleton(RID p_mesh_instance, RID p_sk } void RendererStorageRD::mesh_instance_set_blend_shape_weight(RID p_mesh_instance, int p_shape, float p_weight) { - MeshInstance *mi = mesh_instance_owner.getornull(p_mesh_instance); + MeshInstance *mi = mesh_instance_owner.get_or_null(p_mesh_instance); ERR_FAIL_COND(!mi); ERR_FAIL_INDEX(p_shape, (int)mi->blend_weights.size()); mi->blend_weights[p_shape] = p_weight; @@ -3154,7 +3571,7 @@ void RendererStorageRD::_mesh_instance_add_surface(MeshInstance *mi, Mesh *mesh, } void RendererStorageRD::mesh_instance_check_for_update(RID p_mesh_instance) { - MeshInstance *mi = mesh_instance_owner.getornull(p_mesh_instance); + MeshInstance *mi = mesh_instance_owner.get_or_null(p_mesh_instance); bool needs_update = mi->dirty; @@ -3168,7 +3585,7 @@ void RendererStorageRD::mesh_instance_check_for_update(RID p_mesh_instance) { } if (!needs_update && mi->skeleton.is_valid()) { - Skeleton *sk = skeleton_owner.getornull(mi->skeleton); + Skeleton *sk = skeleton_owner.get_or_null(mi->skeleton); if (sk && sk->version != mi->skeleton_version) { needs_update = true; } @@ -3199,7 +3616,7 @@ void RendererStorageRD::update_mesh_instances() { while (dirty_mesh_instance_arrays.first()) { MeshInstance *mi = dirty_mesh_instance_arrays.first()->self(); - Skeleton *sk = skeleton_owner.getornull(mi->skeleton); + Skeleton *sk = skeleton_owner.get_or_null(mi->skeleton); for (uint32_t i = 0; i < mi->surfaces.size(); i++) { if (mi->surfaces[i].uniform_set == RID() || mi->mesh->surfaces[i]->uniform_set == RID()) { @@ -3446,7 +3863,7 @@ void RendererStorageRD::multimesh_initialize(RID p_rid) { } void RendererStorageRD::multimesh_allocate_data(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors, bool p_use_custom_data) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); if (multimesh->instances == p_instances && multimesh->xform_format == p_transform_format && multimesh->uses_colors == p_use_colors && multimesh->uses_custom_data == p_use_custom_data) { @@ -3456,6 +3873,7 @@ void RendererStorageRD::multimesh_allocate_data(RID p_multimesh, int p_instances if (multimesh->buffer.is_valid()) { RD::get_singleton()->free(multimesh->buffer); multimesh->buffer = RID(); + multimesh->uniform_set_2d = RID(); //cleared by dependency multimesh->uniform_set_3d = RID(); //cleared by dependency } @@ -3488,13 +3906,13 @@ void RendererStorageRD::multimesh_allocate_data(RID p_multimesh, int p_instances } int RendererStorageRD::multimesh_get_instance_count(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, 0); return multimesh->instances; } void RendererStorageRD::multimesh_set_mesh(RID p_multimesh, RID p_mesh) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); if (multimesh->mesh == p_mesh) { return; @@ -3541,7 +3959,7 @@ void RendererStorageRD::_multimesh_make_local(MultiMesh *multimesh) const { memcpy(w, r, buffer.size()); } } else { - memset(w, 0, multimesh->instances * multimesh->stride_cache * sizeof(float)); + memset(w, 0, (size_t)multimesh->instances * multimesh->stride_cache * sizeof(float)); } } uint32_t data_cache_dirty_region_count = (multimesh->instances - 1) / MULTIMESH_DIRTY_REGION_SIZE + 1; @@ -3640,7 +4058,7 @@ void RendererStorageRD::_multimesh_re_create_aabb(MultiMesh *multimesh, const fl } void RendererStorageRD::multimesh_instance_set_transform(RID p_multimesh, int p_index, const Transform3D &p_transform) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_INDEX(p_index, multimesh->instances); ERR_FAIL_COND(multimesh->xform_format != RS::MULTIMESH_TRANSFORM_3D); @@ -3670,7 +4088,7 @@ void RendererStorageRD::multimesh_instance_set_transform(RID p_multimesh, int p_ } void RendererStorageRD::multimesh_instance_set_transform_2d(RID p_multimesh, int p_index, const Transform2D &p_transform) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_INDEX(p_index, multimesh->instances); ERR_FAIL_COND(multimesh->xform_format != RS::MULTIMESH_TRANSFORM_2D); @@ -3696,7 +4114,7 @@ void RendererStorageRD::multimesh_instance_set_transform_2d(RID p_multimesh, int } void RendererStorageRD::multimesh_instance_set_color(RID p_multimesh, int p_index, const Color &p_color) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_INDEX(p_index, multimesh->instances); ERR_FAIL_COND(!multimesh->uses_colors); @@ -3718,7 +4136,7 @@ void RendererStorageRD::multimesh_instance_set_color(RID p_multimesh, int p_inde } void RendererStorageRD::multimesh_instance_set_custom_data(RID p_multimesh, int p_index, const Color &p_color) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_INDEX(p_index, multimesh->instances); ERR_FAIL_COND(!multimesh->uses_custom_data); @@ -3740,14 +4158,14 @@ void RendererStorageRD::multimesh_instance_set_custom_data(RID p_multimesh, int } RID RendererStorageRD::multimesh_get_mesh(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, RID()); return multimesh->mesh; } Transform3D RendererStorageRD::multimesh_instance_get_transform(RID p_multimesh, int p_index) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, Transform3D()); ERR_FAIL_INDEX_V(p_index, multimesh->instances, Transform3D()); ERR_FAIL_COND_V(multimesh->xform_format != RS::MULTIMESH_TRANSFORM_3D, Transform3D()); @@ -3778,7 +4196,7 @@ Transform3D RendererStorageRD::multimesh_instance_get_transform(RID p_multimesh, } Transform2D RendererStorageRD::multimesh_instance_get_transform_2d(RID p_multimesh, int p_index) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, Transform2D()); ERR_FAIL_INDEX_V(p_index, multimesh->instances, Transform2D()); ERR_FAIL_COND_V(multimesh->xform_format != RS::MULTIMESH_TRANSFORM_2D, Transform2D()); @@ -3803,7 +4221,7 @@ Transform2D RendererStorageRD::multimesh_instance_get_transform_2d(RID p_multime } Color RendererStorageRD::multimesh_instance_get_color(RID p_multimesh, int p_index) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, Color()); ERR_FAIL_INDEX_V(p_index, multimesh->instances, Color()); ERR_FAIL_COND_V(!multimesh->uses_colors, Color()); @@ -3826,7 +4244,7 @@ Color RendererStorageRD::multimesh_instance_get_color(RID p_multimesh, int p_ind } Color RendererStorageRD::multimesh_instance_get_custom_data(RID p_multimesh, int p_index) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, Color()); ERR_FAIL_INDEX_V(p_index, multimesh->instances, Color()); ERR_FAIL_COND_V(!multimesh->uses_custom_data, Color()); @@ -3849,7 +4267,7 @@ Color RendererStorageRD::multimesh_instance_get_custom_data(RID p_multimesh, int } void RendererStorageRD::multimesh_set_buffer(RID p_multimesh, const Vector<float> &p_buffer) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_COND(p_buffer.size() != (multimesh->instances * (int)multimesh->stride_cache)); @@ -3882,7 +4300,7 @@ void RendererStorageRD::multimesh_set_buffer(RID p_multimesh, const Vector<float } Vector<float> RendererStorageRD::multimesh_get_buffer(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, Vector<float>()); if (multimesh->buffer.is_null()) { return Vector<float>(); @@ -3905,7 +4323,7 @@ Vector<float> RendererStorageRD::multimesh_get_buffer(RID p_multimesh) const { } void RendererStorageRD::multimesh_set_visible_instances(RID p_multimesh, int p_visible) { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND(!multimesh); ERR_FAIL_COND(p_visible < -1 || p_visible > multimesh->instances); if (multimesh->visible_instances == p_visible) { @@ -3923,13 +4341,13 @@ void RendererStorageRD::multimesh_set_visible_instances(RID p_multimesh, int p_v } int RendererStorageRD::multimesh_get_visible_instances(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, 0); return multimesh->visible_instances; } AABB RendererStorageRD::multimesh_get_aabb(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); ERR_FAIL_COND_V(!multimesh, AABB()); if (multimesh->aabb_dirty) { const_cast<RendererStorageRD *>(this)->_update_dirty_multimeshes(); @@ -3954,13 +4372,13 @@ void RendererStorageRD::_update_dirty_multimeshes() { if (multimesh->data_cache_used_dirty_regions > 32 || multimesh->data_cache_used_dirty_regions > visible_region_count / 2) { //if there too many dirty regions, or represent the majority of regions, just copy all, else transfer cost piles up too much - RD::get_singleton()->buffer_update(multimesh->buffer, 0, MIN(visible_region_count * region_size, multimesh->instances * multimesh->stride_cache * sizeof(float)), data); + RD::get_singleton()->buffer_update(multimesh->buffer, 0, MIN(visible_region_count * region_size, multimesh->instances * (uint32_t)multimesh->stride_cache * (uint32_t)sizeof(float)), data); } else { //not that many regions? update them all for (uint32_t i = 0; i < visible_region_count; i++) { if (multimesh->data_cache_dirty_regions[i]) { - uint64_t offset = i * region_size; - uint64_t size = multimesh->stride_cache * multimesh->instances * sizeof(float); + uint32_t offset = i * region_size; + uint32_t size = multimesh->stride_cache * (uint32_t)multimesh->instances * (uint32_t)sizeof(float); RD::get_singleton()->buffer_update(multimesh->buffer, offset, MIN(region_size, size - offset), &data[i * region_size]); } } @@ -4000,7 +4418,7 @@ void RendererStorageRD::particles_initialize(RID p_rid) { } void RendererStorageRD::particles_set_mode(RID p_particles, RS::ParticlesMode p_mode) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); if (particles->mode == p_mode) { return; @@ -4012,7 +4430,7 @@ void RendererStorageRD::particles_set_mode(RID p_particles, RS::ParticlesMode p_ } void RendererStorageRD::particles_set_emitting(RID p_particles, bool p_emitting) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->emitting = p_emitting; @@ -4020,7 +4438,7 @@ void RendererStorageRD::particles_set_emitting(RID p_particles, bool p_emitting) bool RendererStorageRD::particles_get_emitting(RID p_particles) { ERR_FAIL_COND_V_MSG(RSG::threaded, false, "This function should never be used with threaded rendering, as it stalls the renderer."); - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, false); return particles->emitting; @@ -4075,7 +4493,7 @@ void RendererStorageRD::_particles_free_data(Particles *particles) { } void RendererStorageRD::particles_set_amount(RID p_particles, int p_amount) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); if (particles->amount == p_amount) { @@ -4094,56 +4512,57 @@ void RendererStorageRD::particles_set_amount(RID p_particles, int p_amount) { particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); } -void RendererStorageRD::particles_set_lifetime(RID p_particles, float p_lifetime) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_lifetime(RID p_particles, double p_lifetime) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->lifetime = p_lifetime; } void RendererStorageRD::particles_set_one_shot(RID p_particles, bool p_one_shot) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->one_shot = p_one_shot; } -void RendererStorageRD::particles_set_pre_process_time(RID p_particles, float p_time) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_pre_process_time(RID p_particles, double p_time) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->pre_process_time = p_time; } -void RendererStorageRD::particles_set_explosiveness_ratio(RID p_particles, float p_ratio) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_explosiveness_ratio(RID p_particles, real_t p_ratio) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->explosiveness = p_ratio; } -void RendererStorageRD::particles_set_randomness_ratio(RID p_particles, float p_ratio) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_randomness_ratio(RID p_particles, real_t p_ratio) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->randomness = p_ratio; } void RendererStorageRD::particles_set_custom_aabb(RID p_particles, const AABB &p_aabb) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->custom_aabb = p_aabb; particles->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } -void RendererStorageRD::particles_set_speed_scale(RID p_particles, float p_scale) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_speed_scale(RID p_particles, double p_scale) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->speed_scale = p_scale; } void RendererStorageRD::particles_set_use_local_coordinates(RID p_particles, bool p_enable) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->use_local_coords = p_enable; + particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); } void RendererStorageRD::particles_set_fixed_fps(RID p_particles, int p_fps) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->fixed_fps = p_fps; @@ -4159,21 +4578,21 @@ void RendererStorageRD::particles_set_fixed_fps(RID p_particles, int p_fps) { } void RendererStorageRD::particles_set_interpolate(RID p_particles, bool p_enable) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->interpolate = p_enable; } void RendererStorageRD::particles_set_fractional_delta(RID p_particles, bool p_enable) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->fractional_delta = p_enable; } -void RendererStorageRD::particles_set_trails(RID p_particles, bool p_enable, float p_length) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_trails(RID p_particles, bool p_enable, double p_length) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); ERR_FAIL_COND(p_length < 0.1); p_length = MIN(10.0, p_length); @@ -4192,7 +4611,7 @@ void RendererStorageRD::particles_set_trails(RID p_particles, bool p_enable, flo } void RendererStorageRD::particles_set_trail_bind_poses(RID p_particles, const Vector<Transform3D> &p_bind_poses) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); if (particles->trail_bind_pose_buffer.is_valid() && particles->trail_bind_poses.size() != p_bind_poses.size()) { _particles_free_data(particles); @@ -4208,50 +4627,50 @@ void RendererStorageRD::particles_set_trail_bind_poses(RID p_particles, const Ve particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); } -void RendererStorageRD::particles_set_collision_base_size(RID p_particles, float p_size) { - Particles *particles = particles_owner.getornull(p_particles); +void RendererStorageRD::particles_set_collision_base_size(RID p_particles, real_t p_size) { + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->collision_base_size = p_size; } void RendererStorageRD::particles_set_transform_align(RID p_particles, RS::ParticlesTransformAlign p_transform_align) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->transform_align = p_transform_align; } void RendererStorageRD::particles_set_process_material(RID p_particles, RID p_material) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->process_material = p_material; } void RendererStorageRD::particles_set_draw_order(RID p_particles, RS::ParticlesDrawOrder p_order) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->draw_order = p_order; } void RendererStorageRD::particles_set_draw_passes(RID p_particles, int p_passes) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->draw_passes.resize(p_passes); } void RendererStorageRD::particles_set_draw_pass_mesh(RID p_particles, int p_pass, RID p_mesh) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); ERR_FAIL_INDEX(p_pass, particles->draw_passes.size()); particles->draw_passes.write[p_pass] = p_mesh; } void RendererStorageRD::particles_restart(RID p_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->restart_request = true; @@ -4275,7 +4694,7 @@ void RendererStorageRD::_particles_allocate_emission_buffer(Particles *particles } void RendererStorageRD::particles_set_subemitter(RID p_particles, RID p_subemitter_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); ERR_FAIL_COND(p_particles == p_subemitter_particles); @@ -4288,7 +4707,7 @@ void RendererStorageRD::particles_set_subemitter(RID p_particles, RID p_subemitt } void RendererStorageRD::particles_emit(RID p_particles, const Transform3D &p_transform, const Vector3 &p_velocity, const Color &p_color, const Color &p_custom, uint32_t p_emit_flags) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); ERR_FAIL_COND(particles->amount == 0); @@ -4331,7 +4750,7 @@ void RendererStorageRD::particles_emit(RID p_particles, const Transform3D &p_tra } void RendererStorageRD::particles_request_process(RID p_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); if (!particles->dirty) { @@ -4346,7 +4765,7 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { WARN_PRINT_ONCE("Calling this function with threaded rendering enabled stalls the renderer, use with care."); } - const Particles *particles = particles_owner.getornull(p_particles); + const Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, AABB()); int total_amount = particles->amount; @@ -4354,10 +4773,8 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { total_amount *= particles->trail_bind_poses.size(); } - Vector<ParticleData> data; - data.resize(total_amount); - Vector<uint8_t> buffer = RD::get_singleton()->buffer_get_data(particles->particle_buffer); + ERR_FAIL_COND_V(buffer.size() != (int)(total_amount * sizeof(ParticleData)), AABB()); Transform3D inv = particles->emission_transform.affine_inverse(); @@ -4365,7 +4782,7 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { if (buffer.size()) { bool first = true; - const ParticleData *particle_data = (const ParticleData *)data.ptr(); + const ParticleData *particle_data = reinterpret_cast<const ParticleData *>(buffer.ptr()); for (int i = 0; i < total_amount; i++) { if (particle_data[i].active) { Vector3 pos = Vector3(particle_data[i].xform[12], particle_data[i].xform[13], particle_data[i].xform[14]); @@ -4396,28 +4813,28 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { } AABB RendererStorageRD::particles_get_aabb(RID p_particles) const { - const Particles *particles = particles_owner.getornull(p_particles); + const Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, AABB()); return particles->custom_aabb; } void RendererStorageRD::particles_set_emission_transform(RID p_particles, const Transform3D &p_transform) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->emission_transform = p_transform; } int RendererStorageRD::particles_get_draw_passes(RID p_particles) const { - const Particles *particles = particles_owner.getornull(p_particles); + const Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, 0); return particles->draw_passes.size(); } RID RendererStorageRD::particles_get_draw_pass_mesh(RID p_particles, int p_pass) const { - const Particles *particles = particles_owner.getornull(p_particles); + const Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, RID()); ERR_FAIL_INDEX_V(p_pass, particles->draw_passes.size(), RID()); @@ -4425,19 +4842,19 @@ RID RendererStorageRD::particles_get_draw_pass_mesh(RID p_particles, int p_pass) } void RendererStorageRD::particles_add_collision(RID p_particles, RID p_particles_collision_instance) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->collisions.insert(p_particles_collision_instance); } void RendererStorageRD::particles_remove_collision(RID p_particles, RID p_particles_collision_instance) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->collisions.erase(p_particles_collision_instance); } void RendererStorageRD::particles_set_canvas_sdf_collision(RID p_particles, bool p_enable, const Transform2D &p_xform, const Rect2 &p_to_screen, RID p_texture) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); particles->has_sdf_collision = p_enable; particles->sdf_collision_transform = p_xform; @@ -4445,7 +4862,7 @@ void RendererStorageRD::particles_set_canvas_sdf_collision(RID p_particles, bool particles->sdf_collision_texture = p_texture; } -void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta) { +void RendererStorageRD::_particles_process(Particles *p_particles, double p_delta) { if (p_particles->particles_material_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(p_particles->particles_material_uniform_set)) { Vector<RD::Uniform> uniforms; @@ -4479,7 +4896,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.binding = 3; - Particles *sub_emitter = particles_owner.getornull(p_particles->sub_emitter); + Particles *sub_emitter = particles_owner.get_or_null(p_particles->sub_emitter); if (sub_emitter) { if (sub_emitter->emission_buffer == nullptr) { //no emission buffer, allocate emission buffer _particles_allocate_emission_buffer(sub_emitter); @@ -4494,7 +4911,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta p_particles->particles_material_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, particles_shader.default_shader_rd, 1); } - float new_phase = Math::fmod((float)p_particles->phase + (p_delta / p_particles->lifetime) * p_particles->speed_scale, (float)1.0); + double new_phase = Math::fmod((double)p_particles->phase + (p_delta / p_particles->lifetime) * p_particles->speed_scale, 1.0); //move back history (if there is any) for (uint32_t i = p_particles->frame_history.size() - 1; i > 0; i--) { @@ -4596,11 +5013,11 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta uint32_t collision_3d_textures_used = 0; for (const Set<RID>::Element *E = p_particles->collisions.front(); E; E = E->next()) { - ParticlesCollisionInstance *pci = particles_collision_instance_owner.getornull(E->get()); + ParticlesCollisionInstance *pci = particles_collision_instance_owner.get_or_null(E->get()); if (!pci || !pci->active) { continue; } - ParticlesCollision *pc = particles_collision_owner.getornull(pci->collision); + ParticlesCollision *pc = particles_collision_owner.get_or_null(pci->collision); ERR_CONTINUE(!pc); Transform3D to_collider = pci->transform; @@ -4748,7 +5165,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta for (uint32_t i = 0; i < ParticlesFrameParams::MAX_3D_TEXTURES; i++) { RID rd_tex; if (i < collision_3d_textures_used) { - Texture *t = texture_owner.getornull(collision_3d_textures[i]); + Texture *t = texture_owner.get_or_null(collision_3d_textures[i]); if (t && t->type == Texture::TYPE_3D) { rd_tex = t->rd_texture; } @@ -4793,7 +5210,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta p_particles->force_sub_emit = false; //reset - Particles *sub_emitter = particles_owner.getornull(p_particles->sub_emitter); + Particles *sub_emitter = particles_owner.get_or_null(p_particles->sub_emitter); if (sub_emitter && sub_emitter->emission_storage_buffer.is_valid()) { // print_line("updating subemitter buffer"); @@ -4850,7 +5267,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta RD::get_singleton()->compute_list_bind_uniform_set(compute_list, p_particles->particles_material_uniform_set, 1); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, p_particles->collision_textures_uniform_set, 2); - if (m->uniform_set.is_valid()) { + if (m->uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(m->uniform_set)) { RD::get_singleton()->compute_list_bind_uniform_set(compute_list, m->uniform_set, 3); } @@ -4873,7 +5290,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta } void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 &p_axis, const Vector3 &p_up_axis) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); if (particles->draw_order != RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD_Y_TO_VELOCITY) { @@ -4962,7 +5379,7 @@ void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 & RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); RD::get_singleton()->compute_list_end(); - effects.sort_buffer(particles->particles_sort_uniform_set, particles->amount); + effects->sort_buffer(particles->particles_sort_uniform_set, particles->amount); } copy_push_constant.total_particles *= copy_push_constant.total_particles; @@ -5134,14 +5551,14 @@ void RendererStorageRD::update_particles() { bool zero_time_scale = Engine::get_singleton()->get_time_scale() <= 0.0; if (particles->clear && particles->pre_process_time > 0.0) { - float frame_time; + double frame_time; if (fixed_fps > 0) { frame_time = 1.0 / fixed_fps; } else { frame_time = 1.0 / 30.0; } - float todo = particles->pre_process_time; + double todo = particles->pre_process_time; while (todo >= 0) { _particles_process(particles, frame_time); @@ -5150,8 +5567,8 @@ void RendererStorageRD::update_particles() { } if (fixed_fps > 0) { - float frame_time; - float decr; + double frame_time; + double decr; if (zero_time_scale) { frame_time = 0.0; decr = 1.0 / fixed_fps; @@ -5159,13 +5576,13 @@ void RendererStorageRD::update_particles() { frame_time = 1.0 / fixed_fps; decr = frame_time; } - float delta = RendererCompositorRD::singleton->get_frame_delta_time(); + double delta = RendererCompositorRD::singleton->get_frame_delta_time(); if (delta > 0.1) { //avoid recursive stalls if fps goes below 10 delta = 0.1; } else if (delta <= 0.0) { //unlikely but.. delta = 0.001; } - float todo = particles->frame_remainder + delta; + double todo = particles->frame_remainder + delta; while (todo >= frame_time) { _particles_process(particles, frame_time); @@ -5231,7 +5648,7 @@ void RendererStorageRD::update_particles() { bool RendererStorageRD::particles_is_inactive(RID p_particles) const { ERR_FAIL_COND_V_MSG(RSG::threaded, false, "This function should never be used with threaded rendering, as it stalls the renderer."); - const Particles *particles = particles_owner.getornull(p_particles); + const Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, false); return !particles->emitting && particles->inactive; } @@ -5270,8 +5687,7 @@ void RendererStorageRD::ParticlesShaderData::set_code(const String &p_code) { actions.uniforms = &uniforms; Error err = base_singleton->particles_shader.compiler.compile(RS::SHADER_PARTICLES, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); + ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); if (version.is_null()) { version = base_singleton->particles_shader.shader.version_create(); @@ -5302,36 +5718,36 @@ void RendererStorageRD::ParticlesShaderData::set_default_texture_param(const Str void RendererStorageRD::ParticlesShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { Map<int, StringName> order; - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E.value.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); + if (E.value.texture_order >= 0) { + order[E.value.texture_order + 100000] = E.key; } else { - order[E->get().order] = E->key(); + order[E.value.order] = E.key; } } - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); + for (const KeyValue<int, StringName> &E : order) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E.value]); + pi.name = E.value; p_param_list->push_back(pi); } } void RendererStorageRD::ParticlesShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + for (const KeyValue<StringName, ShaderLanguage::ShaderNode::Uniform> &E : uniforms) { + if (E.value.scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { continue; } RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p.info = ShaderLanguage::uniform_to_property_info(E.value); + p.info.name = E.key; //supply name + p.index = E.value.instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E.value.default_value, E.value.type, E.value.array_size, E.value.hint); p_param_list->push_back(p); } } @@ -5356,7 +5772,7 @@ Variant RendererStorageRD::ParticlesShaderData::get_default_parameter(const Stri if (uniforms.has(p_parameter)) { ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.array_size, uniform.hint); } return Variant(); } @@ -5410,7 +5826,7 @@ void RendererStorageRD::particles_collision_initialize(RID p_rid) { } RID RendererStorageRD::particles_collision_get_heightfield_framebuffer(RID p_particles_collision) const { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND_V(!particles_collision, RID()); ERR_FAIL_COND_V(particles_collision->type != RS::PARTICLES_COLLISION_TYPE_HEIGHTFIELD_COLLIDE, RID()); @@ -5445,7 +5861,7 @@ RID RendererStorageRD::particles_collision_get_heightfield_framebuffer(RID p_par } void RendererStorageRD::particles_collision_set_collision_type(RID p_particles_collision, RS::ParticlesCollisionType p_type) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); if (p_type == particles_collision->type) { @@ -5461,13 +5877,13 @@ void RendererStorageRD::particles_collision_set_collision_type(RID p_particles_c } void RendererStorageRD::particles_collision_set_cull_mask(RID p_particles_collision, uint32_t p_cull_mask) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->cull_mask = p_cull_mask; } -void RendererStorageRD::particles_collision_set_sphere_radius(RID p_particles_collision, float p_radius) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); +void RendererStorageRD::particles_collision_set_sphere_radius(RID p_particles_collision, real_t p_radius) { + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->radius = p_radius; @@ -5475,49 +5891,49 @@ void RendererStorageRD::particles_collision_set_sphere_radius(RID p_particles_co } void RendererStorageRD::particles_collision_set_box_extents(RID p_particles_collision, const Vector3 &p_extents) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->extents = p_extents; particles_collision->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } -void RendererStorageRD::particles_collision_set_attractor_strength(RID p_particles_collision, float p_strength) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); +void RendererStorageRD::particles_collision_set_attractor_strength(RID p_particles_collision, real_t p_strength) { + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->attractor_strength = p_strength; } -void RendererStorageRD::particles_collision_set_attractor_directionality(RID p_particles_collision, float p_directionality) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); +void RendererStorageRD::particles_collision_set_attractor_directionality(RID p_particles_collision, real_t p_directionality) { + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->attractor_directionality = p_directionality; } -void RendererStorageRD::particles_collision_set_attractor_attenuation(RID p_particles_collision, float p_curve) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); +void RendererStorageRD::particles_collision_set_attractor_attenuation(RID p_particles_collision, real_t p_curve) { + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->attractor_attenuation = p_curve; } void RendererStorageRD::particles_collision_set_field_texture(RID p_particles_collision, RID p_texture) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->field_texture = p_texture; } void RendererStorageRD::particles_collision_height_field_update(RID p_particles_collision) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); particles_collision->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } void RendererStorageRD::particles_collision_set_height_field_resolution(RID p_particles_collision, RS::ParticlesCollisionHeightfieldResolution p_resolution) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND(!particles_collision); ERR_FAIL_INDEX(p_resolution, RS::PARTICLES_COLLISION_HEIGHTFIELD_RESOLUTION_MAX); @@ -5534,7 +5950,7 @@ void RendererStorageRD::particles_collision_set_height_field_resolution(RID p_pa } AABB RendererStorageRD::particles_collision_get_aabb(RID p_particles_collision) const { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND_V(!particles_collision, AABB()); switch (particles_collision->type) { @@ -5557,13 +5973,13 @@ AABB RendererStorageRD::particles_collision_get_aabb(RID p_particles_collision) } Vector3 RendererStorageRD::particles_collision_get_extents(RID p_particles_collision) const { - const ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + const ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND_V(!particles_collision, Vector3()); return particles_collision->extents; } bool RendererStorageRD::particles_collision_is_heightfield(RID p_particles_collision) const { - const ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); + const ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_particles_collision); ERR_FAIL_COND_V(!particles_collision, false); return particles_collision->type == RS::PARTICLES_COLLISION_TYPE_HEIGHTFIELD_COLLIDE; } @@ -5574,12 +5990,12 @@ RID RendererStorageRD::particles_collision_instance_create(RID p_collision) { return particles_collision_instance_owner.make_rid(pci); } void RendererStorageRD::particles_collision_instance_set_transform(RID p_collision_instance, const Transform3D &p_transform) { - ParticlesCollisionInstance *pci = particles_collision_instance_owner.getornull(p_collision_instance); + ParticlesCollisionInstance *pci = particles_collision_instance_owner.get_or_null(p_collision_instance); ERR_FAIL_COND(!pci); pci->transform = p_transform; } void RendererStorageRD::particles_collision_instance_set_active(RID p_collision_instance, bool p_active) { - ParticlesCollisionInstance *pci = particles_collision_instance_owner.getornull(p_collision_instance); + ParticlesCollisionInstance *pci = particles_collision_instance_owner.get_or_null(p_collision_instance); ERR_FAIL_COND(!pci); pci->active = p_active; } @@ -5593,25 +6009,25 @@ void RendererStorageRD::visibility_notifier_initialize(RID p_notifier) { visibility_notifier_owner.initialize_rid(p_notifier, VisibilityNotifier()); } void RendererStorageRD::visibility_notifier_set_aabb(RID p_notifier, const AABB &p_aabb) { - VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_notifier); + VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_notifier); ERR_FAIL_COND(!vn); vn->aabb = p_aabb; vn->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } void RendererStorageRD::visibility_notifier_set_callbacks(RID p_notifier, const Callable &p_enter_callbable, const Callable &p_exit_callable) { - VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_notifier); + VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_notifier); ERR_FAIL_COND(!vn); vn->enter_callback = p_enter_callbable; vn->exit_callback = p_exit_callable; } AABB RendererStorageRD::visibility_notifier_get_aabb(RID p_notifier) const { - const VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_notifier); + const VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_notifier); ERR_FAIL_COND_V(!vn, AABB()); return vn->aabb; } void RendererStorageRD::visibility_notifier_call(RID p_notifier, bool p_enter, bool p_deferred) { - VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_notifier); + VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_notifier); ERR_FAIL_COND(!vn); if (p_enter) { @@ -5655,7 +6071,7 @@ void RendererStorageRD::_skeleton_make_dirty(Skeleton *skeleton) { } void RendererStorageRD::skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton) { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND(!skeleton); ERR_FAIL_COND(p_bones < 0); @@ -5698,14 +6114,14 @@ void RendererStorageRD::skeleton_allocate_data(RID p_skeleton, int p_bones, bool } int RendererStorageRD::skeleton_get_bone_count(RID p_skeleton) const { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND_V(!skeleton, 0); return skeleton->size; } void RendererStorageRD::skeleton_bone_set_transform(RID p_skeleton, int p_bone, const Transform3D &p_transform) { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND(!skeleton); ERR_FAIL_INDEX(p_bone, skeleton->size); @@ -5730,7 +6146,7 @@ void RendererStorageRD::skeleton_bone_set_transform(RID p_skeleton, int p_bone, } Transform3D RendererStorageRD::skeleton_bone_get_transform(RID p_skeleton, int p_bone) const { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND_V(!skeleton, Transform3D()); ERR_FAIL_INDEX_V(p_bone, skeleton->size, Transform3D()); @@ -5757,7 +6173,7 @@ Transform3D RendererStorageRD::skeleton_bone_get_transform(RID p_skeleton, int p } void RendererStorageRD::skeleton_bone_set_transform_2d(RID p_skeleton, int p_bone, const Transform2D &p_transform) { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND(!skeleton); ERR_FAIL_INDEX(p_bone, skeleton->size); @@ -5778,7 +6194,7 @@ void RendererStorageRD::skeleton_bone_set_transform_2d(RID p_skeleton, int p_bon } Transform2D RendererStorageRD::skeleton_bone_get_transform_2d(RID p_skeleton, int p_bone) const { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND_V(!skeleton, Transform2D()); ERR_FAIL_INDEX_V(p_bone, skeleton->size, Transform2D()); @@ -5798,7 +6214,7 @@ Transform2D RendererStorageRD::skeleton_bone_get_transform_2d(RID p_skeleton, in } void RendererStorageRD::skeleton_set_base_transform_2d(RID p_skeleton, const Transform2D &p_base_transform) { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND(!skeleton->use_2d); @@ -5877,14 +6293,14 @@ void RendererStorageRD::spot_light_initialize(RID p_light) { } void RendererStorageRD::light_set_color(RID p_light, const Color &p_color) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->color = p_color; } void RendererStorageRD::light_set_param(RID p_light, RS::LightParam p_param, float p_value) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); ERR_FAIL_INDEX(p_param, RS::LIGHT_PARAM_MAX); @@ -5919,7 +6335,7 @@ void RendererStorageRD::light_set_param(RID p_light, RS::LightParam p_param, flo } void RendererStorageRD::light_set_shadow(RID p_light, bool p_enabled) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->shadow = p_enabled; @@ -5928,13 +6344,13 @@ void RendererStorageRD::light_set_shadow(RID p_light, bool p_enabled) { } void RendererStorageRD::light_set_shadow_color(RID p_light, const Color &p_color) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->shadow_color = p_color; } void RendererStorageRD::light_set_projector(RID p_light, RID p_texture) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); if (light->projector == p_texture) { @@ -5956,14 +6372,14 @@ void RendererStorageRD::light_set_projector(RID p_light, RID p_texture) { } void RendererStorageRD::light_set_negative(RID p_light, bool p_enable) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->negative = p_enable; } void RendererStorageRD::light_set_cull_mask(RID p_light, uint32_t p_mask) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->cull_mask = p_mask; @@ -5973,7 +6389,7 @@ void RendererStorageRD::light_set_cull_mask(RID p_light, uint32_t p_mask) { } void RendererStorageRD::light_set_reverse_cull_face_mode(RID p_light, bool p_enabled) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->reverse_cull = p_enabled; @@ -5983,7 +6399,7 @@ void RendererStorageRD::light_set_reverse_cull_face_mode(RID p_light, bool p_ena } void RendererStorageRD::light_set_bake_mode(RID p_light, RS::LightBakeMode p_bake_mode) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->bake_mode = p_bake_mode; @@ -5993,7 +6409,7 @@ void RendererStorageRD::light_set_bake_mode(RID p_light, RS::LightBakeMode p_bak } void RendererStorageRD::light_set_max_sdfgi_cascade(RID p_light, uint32_t p_cascade) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->max_sdfgi_cascade = p_cascade; @@ -6003,7 +6419,7 @@ void RendererStorageRD::light_set_max_sdfgi_cascade(RID p_light, uint32_t p_casc } void RendererStorageRD::light_omni_set_shadow_mode(RID p_light, RS::LightOmniShadowMode p_mode) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->omni_shadow_mode = p_mode; @@ -6013,14 +6429,14 @@ void RendererStorageRD::light_omni_set_shadow_mode(RID p_light, RS::LightOmniSha } RS::LightOmniShadowMode RendererStorageRD::light_omni_get_shadow_mode(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_OMNI_SHADOW_CUBE); return light->omni_shadow_mode; } void RendererStorageRD::light_directional_set_shadow_mode(RID p_light, RS::LightDirectionalShadowMode p_mode) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->directional_shadow_mode = p_mode; @@ -6029,7 +6445,7 @@ void RendererStorageRD::light_directional_set_shadow_mode(RID p_light, RS::Light } void RendererStorageRD::light_directional_set_blend_splits(RID p_light, bool p_enable) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->directional_blend_splits = p_enable; @@ -6038,56 +6454,56 @@ void RendererStorageRD::light_directional_set_blend_splits(RID p_light, bool p_e } bool RendererStorageRD::light_directional_get_blend_splits(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, false); return light->directional_blend_splits; } void RendererStorageRD::light_directional_set_sky_only(RID p_light, bool p_sky_only) { - Light *light = light_owner.getornull(p_light); + Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND(!light); light->directional_sky_only = p_sky_only; } bool RendererStorageRD::light_directional_is_sky_only(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, false); return light->directional_sky_only; } RS::LightDirectionalShadowMode RendererStorageRD::light_directional_get_shadow_mode(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_DIRECTIONAL_SHADOW_ORTHOGONAL); return light->directional_shadow_mode; } uint32_t RendererStorageRD::light_get_max_sdfgi_cascade(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0); return light->max_sdfgi_cascade; } RS::LightBakeMode RendererStorageRD::light_get_bake_mode(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_BAKE_DISABLED); return light->bake_mode; } uint64_t RendererStorageRD::light_get_version(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0); return light->version; } AABB RendererStorageRD::light_get_aabb(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, AABB()); switch (light->type) { @@ -6118,7 +6534,7 @@ void RendererStorageRD::reflection_probe_initialize(RID p_reflection_probe) { } void RendererStorageRD::reflection_probe_set_update_mode(RID p_probe, RS::ReflectionProbeUpdateMode p_mode) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->update_mode = p_mode; @@ -6126,35 +6542,35 @@ void RendererStorageRD::reflection_probe_set_update_mode(RID p_probe, RS::Reflec } void RendererStorageRD::reflection_probe_set_intensity(RID p_probe, float p_intensity) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->intensity = p_intensity; } void RendererStorageRD::reflection_probe_set_ambient_mode(RID p_probe, RS::ReflectionProbeAmbientMode p_mode) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->ambient_mode = p_mode; } void RendererStorageRD::reflection_probe_set_ambient_color(RID p_probe, const Color &p_color) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->ambient_color = p_color; } void RendererStorageRD::reflection_probe_set_ambient_energy(RID p_probe, float p_energy) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->ambient_color_energy = p_energy; } void RendererStorageRD::reflection_probe_set_max_distance(RID p_probe, float p_distance) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->max_distance = p_distance; @@ -6163,7 +6579,7 @@ void RendererStorageRD::reflection_probe_set_max_distance(RID p_probe, float p_d } void RendererStorageRD::reflection_probe_set_extents(RID p_probe, const Vector3 &p_extents) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); if (reflection_probe->extents == p_extents) { @@ -6174,7 +6590,7 @@ void RendererStorageRD::reflection_probe_set_extents(RID p_probe, const Vector3 } void RendererStorageRD::reflection_probe_set_origin_offset(RID p_probe, const Vector3 &p_offset) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->origin_offset = p_offset; @@ -6182,7 +6598,7 @@ void RendererStorageRD::reflection_probe_set_origin_offset(RID p_probe, const Ve } void RendererStorageRD::reflection_probe_set_as_interior(RID p_probe, bool p_enable) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->interior = p_enable; @@ -6190,14 +6606,14 @@ void RendererStorageRD::reflection_probe_set_as_interior(RID p_probe, bool p_ena } void RendererStorageRD::reflection_probe_set_enable_box_projection(RID p_probe, bool p_enable) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->box_projection = p_enable; } void RendererStorageRD::reflection_probe_set_enable_shadows(RID p_probe, bool p_enable) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->enable_shadows = p_enable; @@ -6205,7 +6621,7 @@ void RendererStorageRD::reflection_probe_set_enable_shadows(RID p_probe, bool p_ } void RendererStorageRD::reflection_probe_set_cull_mask(RID p_probe, uint32_t p_layers) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->cull_mask = p_layers; @@ -6213,7 +6629,7 @@ void RendererStorageRD::reflection_probe_set_cull_mask(RID p_probe, uint32_t p_l } void RendererStorageRD::reflection_probe_set_resolution(RID p_probe, int p_resolution) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); ERR_FAIL_COND(p_resolution < 32); @@ -6221,7 +6637,7 @@ void RendererStorageRD::reflection_probe_set_resolution(RID p_probe, int p_resol } void RendererStorageRD::reflection_probe_set_lod_threshold(RID p_probe, float p_ratio) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND(!reflection_probe); reflection_probe->lod_threshold = p_ratio; @@ -6230,7 +6646,7 @@ void RendererStorageRD::reflection_probe_set_lod_threshold(RID p_probe, float p_ } AABB RendererStorageRD::reflection_probe_get_aabb(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, AABB()); AABB aabb; @@ -6241,96 +6657,96 @@ AABB RendererStorageRD::reflection_probe_get_aabb(RID p_probe) const { } RS::ReflectionProbeUpdateMode RendererStorageRD::reflection_probe_get_update_mode(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, RS::REFLECTION_PROBE_UPDATE_ALWAYS); return reflection_probe->update_mode; } uint32_t RendererStorageRD::reflection_probe_get_cull_mask(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->cull_mask; } Vector3 RendererStorageRD::reflection_probe_get_extents(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, Vector3()); return reflection_probe->extents; } Vector3 RendererStorageRD::reflection_probe_get_origin_offset(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, Vector3()); return reflection_probe->origin_offset; } bool RendererStorageRD::reflection_probe_renders_shadows(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, false); return reflection_probe->enable_shadows; } float RendererStorageRD::reflection_probe_get_origin_max_distance(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->max_distance; } float RendererStorageRD::reflection_probe_get_lod_threshold(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->lod_threshold; } int RendererStorageRD::reflection_probe_get_resolution(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->resolution; } float RendererStorageRD::reflection_probe_get_intensity(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->intensity; } bool RendererStorageRD::reflection_probe_is_interior(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, false); return reflection_probe->interior; } bool RendererStorageRD::reflection_probe_is_box_projection(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, false); return reflection_probe->box_projection; } RS::ReflectionProbeAmbientMode RendererStorageRD::reflection_probe_get_ambient_mode(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, RS::REFLECTION_PROBE_AMBIENT_DISABLED); return reflection_probe->ambient_mode; } Color RendererStorageRD::reflection_probe_get_ambient_color(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, Color()); return reflection_probe->ambient_color; } float RendererStorageRD::reflection_probe_get_ambient_color_energy(RID p_probe) const { - const ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_probe); + const ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_probe); ERR_FAIL_COND_V(!reflection_probe, 0); return reflection_probe->ambient_color_energy; @@ -6344,14 +6760,14 @@ void RendererStorageRD::decal_initialize(RID p_decal) { } void RendererStorageRD::decal_set_extents(RID p_decal, const Vector3 &p_extents) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->extents = p_extents; decal->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } void RendererStorageRD::decal_set_texture(RID p_decal, RS::DecalTexture p_type, RID p_texture) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); ERR_FAIL_INDEX(p_type, RS::DECAL_TEXTURE_MAX); @@ -6375,32 +6791,32 @@ void RendererStorageRD::decal_set_texture(RID p_decal, RS::DecalTexture p_type, } void RendererStorageRD::decal_set_emission_energy(RID p_decal, float p_energy) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->emission_energy = p_energy; } void RendererStorageRD::decal_set_albedo_mix(RID p_decal, float p_mix) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->albedo_mix = p_mix; } void RendererStorageRD::decal_set_modulate(RID p_decal, const Color &p_modulate) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->modulate = p_modulate; } void RendererStorageRD::decal_set_cull_mask(RID p_decal, uint32_t p_layers) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->cull_mask = p_layers; decal->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } void RendererStorageRD::decal_set_distance_fade(RID p_decal, bool p_enabled, float p_begin, float p_length) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->distance_fade = p_enabled; decal->distance_fade_begin = p_begin; @@ -6408,20 +6824,20 @@ void RendererStorageRD::decal_set_distance_fade(RID p_decal, bool p_enabled, flo } void RendererStorageRD::decal_set_fade(RID p_decal, float p_above, float p_below) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->upper_fade = p_above; decal->lower_fade = p_below; } void RendererStorageRD::decal_set_normal_fade(RID p_decal, float p_fade) { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND(!decal); decal->normal_fade = p_fade; } AABB RendererStorageRD::decal_get_aabb(RID p_decal) const { - Decal *decal = decal_owner.getornull(p_decal); + Decal *decal = decal_owner.get_or_null(p_decal); ERR_FAIL_COND_V(!decal, AABB()); return AABB(-decal->extents, decal->extents * 2.0); @@ -6435,7 +6851,7 @@ void RendererStorageRD::voxel_gi_initialize(RID p_voxel_gi) { } void RendererStorageRD::voxel_gi_allocate_data(RID p_voxel_gi, const Transform3D &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); if (voxel_gi->octree_buffer.is_valid()) { @@ -6560,20 +6976,20 @@ void RendererStorageRD::voxel_gi_allocate_data(RID p_voxel_gi, const Transform3D } AABB RendererStorageRD::voxel_gi_get_bounds(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, AABB()); return voxel_gi->bounds; } Vector3i RendererStorageRD::voxel_gi_get_octree_size(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Vector3i()); return voxel_gi->octree_size; } Vector<uint8_t> RendererStorageRD::voxel_gi_get_octree_cells(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Vector<uint8_t>()); if (voxel_gi->octree_buffer.is_valid()) { @@ -6583,7 +6999,7 @@ Vector<uint8_t> RendererStorageRD::voxel_gi_get_octree_cells(RID p_voxel_gi) con } Vector<uint8_t> RendererStorageRD::voxel_gi_get_data_cells(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Vector<uint8_t>()); if (voxel_gi->data_buffer.is_valid()) { @@ -6593,7 +7009,7 @@ Vector<uint8_t> RendererStorageRD::voxel_gi_get_data_cells(RID p_voxel_gi) const } Vector<uint8_t> RendererStorageRD::voxel_gi_get_distance_field(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Vector<uint8_t>()); if (voxel_gi->data_buffer.is_valid()) { @@ -6603,21 +7019,21 @@ Vector<uint8_t> RendererStorageRD::voxel_gi_get_distance_field(RID p_voxel_gi) c } Vector<int> RendererStorageRD::voxel_gi_get_level_counts(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Vector<int>()); return voxel_gi->level_counts; } Transform3D RendererStorageRD::voxel_gi_get_to_cell_xform(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, Transform3D()); return voxel_gi->to_cell_xform; } void RendererStorageRD::voxel_gi_set_dynamic_range(RID p_voxel_gi, float p_range) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->dynamic_range = p_range; @@ -6625,14 +7041,14 @@ void RendererStorageRD::voxel_gi_set_dynamic_range(RID p_voxel_gi, float p_range } float RendererStorageRD::voxel_gi_get_dynamic_range(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->dynamic_range; } void RendererStorageRD::voxel_gi_set_propagation(RID p_voxel_gi, float p_range) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->propagation = p_range; @@ -6640,72 +7056,72 @@ void RendererStorageRD::voxel_gi_set_propagation(RID p_voxel_gi, float p_range) } float RendererStorageRD::voxel_gi_get_propagation(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->propagation; } void RendererStorageRD::voxel_gi_set_energy(RID p_voxel_gi, float p_energy) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->energy = p_energy; } float RendererStorageRD::voxel_gi_get_energy(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->energy; } void RendererStorageRD::voxel_gi_set_bias(RID p_voxel_gi, float p_bias) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->bias = p_bias; } float RendererStorageRD::voxel_gi_get_bias(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->bias; } void RendererStorageRD::voxel_gi_set_normal_bias(RID p_voxel_gi, float p_normal_bias) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->normal_bias = p_normal_bias; } float RendererStorageRD::voxel_gi_get_normal_bias(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->normal_bias; } void RendererStorageRD::voxel_gi_set_anisotropy_strength(RID p_voxel_gi, float p_strength) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->anisotropy_strength = p_strength; } float RendererStorageRD::voxel_gi_get_anisotropy_strength(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->anisotropy_strength; } void RendererStorageRD::voxel_gi_set_interior(RID p_voxel_gi, bool p_enable) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->interior = p_enable; } void RendererStorageRD::voxel_gi_set_use_two_bounces(RID p_voxel_gi, bool p_enable) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND(!voxel_gi); voxel_gi->use_two_bounces = p_enable; @@ -6713,43 +7129,43 @@ void RendererStorageRD::voxel_gi_set_use_two_bounces(RID p_voxel_gi, bool p_enab } bool RendererStorageRD::voxel_gi_is_using_two_bounces(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, false); return voxel_gi->use_two_bounces; } bool RendererStorageRD::voxel_gi_is_interior(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->interior; } uint32_t RendererStorageRD::voxel_gi_get_version(RID p_voxel_gi) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->version; } uint32_t RendererStorageRD::voxel_gi_get_data_version(RID p_voxel_gi) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, 0); return voxel_gi->data_version; } RID RendererStorageRD::voxel_gi_get_octree_buffer(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, RID()); return voxel_gi->octree_buffer; } RID RendererStorageRD::voxel_gi_get_data_buffer(RID p_voxel_gi) const { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, RID()); return voxel_gi->data_buffer; } RID RendererStorageRD::voxel_gi_get_sdf_texture(RID p_voxel_gi) { - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_voxel_gi); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_voxel_gi); ERR_FAIL_COND_V(!voxel_gi, RID()); return voxel_gi->sdf_texture; @@ -6766,20 +7182,20 @@ void RendererStorageRD::lightmap_initialize(RID p_lightmap) { } void RendererStorageRD::lightmap_set_textures(RID p_lightmap, RID p_light, bool p_uses_spherical_haromics) { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!lm); lightmap_array_version++; //erase lightmap users if (lm->light_texture.is_valid()) { - Texture *t = texture_owner.getornull(lm->light_texture); + Texture *t = texture_owner.get_or_null(lm->light_texture); if (t) { t->lightmap_users.erase(p_lightmap); } } - Texture *t = texture_owner.getornull(p_light); + Texture *t = texture_owner.get_or_null(p_light); lm->light_texture = p_light; lm->uses_spherical_harmonics = p_uses_spherical_haromics; @@ -6814,19 +7230,19 @@ void RendererStorageRD::lightmap_set_textures(RID p_lightmap, RID p_light, bool } void RendererStorageRD::lightmap_set_probe_bounds(RID p_lightmap, const AABB &p_bounds) { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!lm); lm->bounds = p_bounds; } void RendererStorageRD::lightmap_set_probe_interior(RID p_lightmap, bool p_interior) { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!lm); lm->interior = p_interior; } void RendererStorageRD::lightmap_set_probe_capture_data(RID p_lightmap, const PackedVector3Array &p_points, const PackedColorArray &p_point_sh, const PackedInt32Array &p_tetrahedra, const PackedInt32Array &p_bsp_tree) { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!lm); if (p_points.size()) { @@ -6842,26 +7258,26 @@ void RendererStorageRD::lightmap_set_probe_capture_data(RID p_lightmap, const Pa } PackedVector3Array RendererStorageRD::lightmap_get_probe_capture_points(RID p_lightmap) const { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, PackedVector3Array()); return lm->points; } PackedColorArray RendererStorageRD::lightmap_get_probe_capture_sh(RID p_lightmap) const { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, PackedColorArray()); return lm->point_sh; } PackedInt32Array RendererStorageRD::lightmap_get_probe_capture_tetrahedra(RID p_lightmap) const { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, PackedInt32Array()); return lm->tetrahedra; } PackedInt32Array RendererStorageRD::lightmap_get_probe_capture_bsp_tree(RID p_lightmap) const { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, PackedInt32Array()); return lm->bsp_tree; } @@ -6871,7 +7287,7 @@ void RendererStorageRD::lightmap_set_probe_capture_update_speed(float p_speed) { } void RendererStorageRD::lightmap_tap_sh_light(RID p_lightmap, const Vector3 &p_point, Color *r_sh) { - Lightmap *lm = lightmap_owner.getornull(p_lightmap); + Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND(!lm); for (int i = 0; i < 9; i++) { @@ -6921,13 +7337,13 @@ void RendererStorageRD::lightmap_tap_sh_light(RID p_lightmap, const Vector3 &p_p } bool RendererStorageRD::lightmap_is_interior(RID p_lightmap) const { - const Lightmap *lm = lightmap_owner.getornull(p_lightmap); + const Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, false); return lm->interior; } AABB RendererStorageRD::lightmap_get_aabb(RID p_lightmap) const { - const Lightmap *lm = lightmap_owner.getornull(p_lightmap); + const Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, AABB()); return lm->bounds; } @@ -6967,7 +7383,7 @@ void RendererStorageRD::_update_render_target(RenderTarget *rt) { //create a placeholder until updated rt->texture = texture_allocate(); texture_2d_placeholder_initialize(rt->texture); - Texture *tex = texture_owner.getornull(rt->texture); + Texture *tex = texture_owner.get_or_null(rt->texture); tex->is_render_target = true; } @@ -7014,7 +7430,7 @@ void RendererStorageRD::_update_render_target(RenderTarget *rt) { { //update texture - Texture *tex = texture_owner.getornull(rt->texture); + Texture *tex = texture_owner.get_or_null(rt->texture); //free existing textures if (RD::get_singleton()->texture_is_valid(tex->rd_texture)) { @@ -7121,7 +7537,7 @@ void RendererStorageRD::render_target_set_position(RID p_render_target, int p_x, } void RendererStorageRD::render_target_set_size(RID p_render_target, int p_width, int p_height, uint32_t p_view_count) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (rt->size.x != p_width || rt->size.y != p_height || rt->view_count != p_view_count) { rt->size.x = p_width; @@ -7132,7 +7548,7 @@ void RendererStorageRD::render_target_set_size(RID p_render_target, int p_width, } RID RendererStorageRD::render_target_get_texture(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->texture; @@ -7142,53 +7558,53 @@ void RendererStorageRD::render_target_set_external_texture(RID p_render_target, } void RendererStorageRD::render_target_set_flag(RID p_render_target, RenderTargetFlags p_flag, bool p_value) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->flags[p_flag] = p_value; _update_render_target(rt); } bool RendererStorageRD::render_target_was_used(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, false); return rt->was_used; } void RendererStorageRD::render_target_set_as_unused(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->was_used = false; } Size2 RendererStorageRD::render_target_get_size(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, Size2()); return rt->size; } RID RendererStorageRD::render_target_get_rd_framebuffer(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->framebuffer; } RID RendererStorageRD::render_target_get_rd_texture(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->color; } RID RendererStorageRD::render_target_get_rd_backbuffer(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->backbuffer; } RID RendererStorageRD::render_target_get_rd_backbuffer_framebuffer(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); if (!rt->backbuffer.is_valid()) { @@ -7199,32 +7615,32 @@ RID RendererStorageRD::render_target_get_rd_backbuffer_framebuffer(RID p_render_ } void RendererStorageRD::render_target_request_clear(RID p_render_target, const Color &p_clear_color) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->clear_requested = true; rt->clear_color = p_clear_color; } bool RendererStorageRD::render_target_is_clear_requested(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, false); return rt->clear_requested; } Color RendererStorageRD::render_target_get_clear_request_color(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, Color()); return rt->clear_color; } void RendererStorageRD::render_target_disable_clear_request(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->clear_requested = false; } void RendererStorageRD::render_target_do_clear_request(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (!rt->clear_requested) { return; @@ -7237,7 +7653,7 @@ void RendererStorageRD::render_target_do_clear_request(RID p_render_target) { } void RendererStorageRD::render_target_set_sdf_size_and_scale(RID p_render_target, RS::ViewportSDFOversize p_size, RS::ViewportSDFScale p_scale) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (rt->sdf_oversize == p_size && rt->sdf_scale == p_scale) { return; @@ -7279,28 +7695,28 @@ Rect2i RendererStorageRD::_render_target_get_sdf_rect(const RenderTarget *rt) co } Rect2i RendererStorageRD::render_target_get_sdf_rect(RID p_render_target) const { - const RenderTarget *rt = render_target_owner.getornull(p_render_target); + const RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, Rect2i()); return _render_target_get_sdf_rect(rt); } void RendererStorageRD::render_target_mark_sdf_enabled(RID p_render_target, bool p_enabled) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->sdf_enabled = p_enabled; } bool RendererStorageRD::render_target_is_sdf_enabled(RID p_render_target) const { - const RenderTarget *rt = render_target_owner.getornull(p_render_target); + const RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, false); return rt->sdf_enabled; } RID RendererStorageRD::render_target_get_sdf_texture(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); if (rt->sdf_buffer_read.is_null()) { // no texture, create a dummy one for the 2D uniform set @@ -7435,7 +7851,7 @@ void RendererStorageRD::_render_target_clear_sdf(RenderTarget *rt) { } RID RendererStorageRD::render_target_get_sdf_framebuffer(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); if (rt->sdf_buffer_write_fb.is_null()) { @@ -7445,7 +7861,7 @@ RID RendererStorageRD::render_target_get_sdf_framebuffer(RID p_render_target) { return rt->sdf_buffer_write_fb; } void RendererStorageRD::render_target_sdf_process(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); ERR_FAIL_COND(rt->sdf_buffer_write_fb.is_null()); @@ -7520,7 +7936,7 @@ void RendererStorageRD::render_target_sdf_process(RID p_render_target) { } void RendererStorageRD::render_target_copy_to_back_buffer(RID p_render_target, const Rect2i &p_region, bool p_gen_mipmaps) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (!rt->backbuffer.is_valid()) { _create_render_target_backbuffer(rt); @@ -7538,7 +7954,7 @@ void RendererStorageRD::render_target_copy_to_back_buffer(RID p_render_target, c //single texture copy for backbuffer //RD::get_singleton()->texture_copy(rt->color, rt->backbuffer_mipmap0, Vector3(region.position.x, region.position.y, 0), Vector3(region.position.x, region.position.y, 0), Vector3(region.size.x, region.size.y, 1), 0, 0, 0, 0, true); - effects.copy_to_rect(rt->color, rt->backbuffer_mipmap0, region, false, false, false, true, true); + effects->copy_to_rect(rt->color, rt->backbuffer_mipmap0, region, false, false, false, true, true); if (!p_gen_mipmaps) { return; @@ -7554,13 +7970,13 @@ void RendererStorageRD::render_target_copy_to_back_buffer(RID p_render_target, c region.size.y = MAX(1, region.size.y >> 1); const RenderTarget::BackbufferMipmap &mm = rt->backbuffer_mipmaps[i]; - effects.gaussian_blur(prev_texture, mm.mipmap, mm.mipmap_copy, region, true); + effects->gaussian_blur(prev_texture, mm.mipmap, mm.mipmap_copy, region, true); prev_texture = mm.mipmap; } } void RendererStorageRD::render_target_clear_back_buffer(RID p_render_target, const Rect2i &p_region, const Color &p_color) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (!rt->backbuffer.is_valid()) { _create_render_target_backbuffer(rt); @@ -7577,11 +7993,11 @@ void RendererStorageRD::render_target_clear_back_buffer(RID p_render_target, con } //single texture copy for backbuffer - effects.set_color(rt->backbuffer_mipmap0, p_color, region, true); + effects->set_color(rt->backbuffer_mipmap0, p_color, region, true); } void RendererStorageRD::render_target_gen_back_buffer_mipmaps(RID p_render_target, const Rect2i &p_region) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); if (!rt->backbuffer.is_valid()) { _create_render_target_backbuffer(rt); @@ -7607,72 +8023,72 @@ void RendererStorageRD::render_target_gen_back_buffer_mipmaps(RID p_render_targe region.size.y = MAX(1, region.size.y >> 1); const RenderTarget::BackbufferMipmap &mm = rt->backbuffer_mipmaps[i]; - effects.gaussian_blur(prev_texture, mm.mipmap, mm.mipmap_copy, region, true); + effects->gaussian_blur(prev_texture, mm.mipmap, mm.mipmap_copy, region, true); prev_texture = mm.mipmap; } } RID RendererStorageRD::render_target_get_framebuffer_uniform_set(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->framebuffer_uniform_set; } RID RendererStorageRD::render_target_get_backbuffer_uniform_set(RID p_render_target) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND_V(!rt, RID()); return rt->backbuffer_uniform_set; } void RendererStorageRD::render_target_set_framebuffer_uniform_set(RID p_render_target, RID p_uniform_set) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->framebuffer_uniform_set = p_uniform_set; } void RendererStorageRD::render_target_set_backbuffer_uniform_set(RID p_render_target, RID p_uniform_set) { - RenderTarget *rt = render_target_owner.getornull(p_render_target); + RenderTarget *rt = render_target_owner.get_or_null(p_render_target); ERR_FAIL_COND(!rt); rt->backbuffer_uniform_set = p_uniform_set; } void RendererStorageRD::base_update_dependency(RID p_base, DependencyTracker *p_instance) { if (mesh_owner.owns(p_base)) { - Mesh *mesh = mesh_owner.getornull(p_base); + Mesh *mesh = mesh_owner.get_or_null(p_base); p_instance->update_dependency(&mesh->dependency); } else if (multimesh_owner.owns(p_base)) { - MultiMesh *multimesh = multimesh_owner.getornull(p_base); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_base); p_instance->update_dependency(&multimesh->dependency); if (multimesh->mesh.is_valid()) { base_update_dependency(multimesh->mesh, p_instance); } } else if (reflection_probe_owner.owns(p_base)) { - ReflectionProbe *rp = reflection_probe_owner.getornull(p_base); + ReflectionProbe *rp = reflection_probe_owner.get_or_null(p_base); p_instance->update_dependency(&rp->dependency); } else if (decal_owner.owns(p_base)) { - Decal *decal = decal_owner.getornull(p_base); + Decal *decal = decal_owner.get_or_null(p_base); p_instance->update_dependency(&decal->dependency); } else if (voxel_gi_owner.owns(p_base)) { - VoxelGI *gip = voxel_gi_owner.getornull(p_base); + VoxelGI *gip = voxel_gi_owner.get_or_null(p_base); p_instance->update_dependency(&gip->dependency); } else if (lightmap_owner.owns(p_base)) { - Lightmap *lm = lightmap_owner.getornull(p_base); + Lightmap *lm = lightmap_owner.get_or_null(p_base); p_instance->update_dependency(&lm->dependency); } else if (light_owner.owns(p_base)) { - Light *l = light_owner.getornull(p_base); + Light *l = light_owner.get_or_null(p_base); p_instance->update_dependency(&l->dependency); } else if (particles_owner.owns(p_base)) { - Particles *p = particles_owner.getornull(p_base); + Particles *p = particles_owner.get_or_null(p_base); p_instance->update_dependency(&p->dependency); } else if (particles_collision_owner.owns(p_base)) { - ParticlesCollision *pc = particles_collision_owner.getornull(p_base); + ParticlesCollision *pc = particles_collision_owner.get_or_null(p_base); p_instance->update_dependency(&pc->dependency); } else if (visibility_notifier_owner.owns(p_base)) { - VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_base); + VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_base); p_instance->update_dependency(&vn->dependency); } } void RendererStorageRD::skeleton_update_dependency(RID p_skeleton, DependencyTracker *p_instance) { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND(!skeleton); p_instance->update_dependency(&skeleton->dependency); @@ -7778,7 +8194,7 @@ void RendererStorageRD::_update_decal_atlas() { while ((K = decal_atlas.textures.next(K))) { DecalAtlas::SortItem &si = itemsv.write[idx]; - Texture *src_tex = texture_owner.getornull(*K); + Texture *src_tex = texture_owner.get_or_null(*K); si.size.width = (src_tex->width / border) + 1; si.size.height = (src_tex->height / border) + 1; @@ -7927,15 +8343,15 @@ void RendererStorageRD::_update_decal_atlas() { const RID *K = nullptr; while ((K = decal_atlas.textures.next(K))) { DecalAtlas::Texture *t = decal_atlas.textures.getptr(*K); - Texture *src_tex = texture_owner.getornull(*K); - effects.copy_to_atlas_fb(src_tex->rd_texture, mm.fb, t->uv_rect, draw_list, false, t->panorama_to_dp_users > 0); + Texture *src_tex = texture_owner.get_or_null(*K); + effects->copy_to_atlas_fb(src_tex->rd_texture, mm.fb, t->uv_rect, draw_list, false, t->panorama_to_dp_users > 0); } RD::get_singleton()->draw_list_end(); prev_texture = mm.texture; } else { - effects.copy_to_fb_rect(prev_texture, mm.fb, Rect2i(Point2i(), mm.size)); + effects->copy_to_fb_rect(prev_texture, mm.fb, Rect2i(Point2i(), mm.size)); prev_texture = mm.texture; } } else { @@ -8343,7 +8759,7 @@ void RendererStorageRD::global_variable_set(const StringName &p_name, const Vari } else { //texture for (Set<RID>::Element *E = gv.texture_materials.front(); E; E = E->next()) { - Material *material = material_owner.getornull(E->get()); + Material *material = material_owner.get_or_null(E->get()); ERR_CONTINUE(!material); _material_queue_update(material, false, true); } @@ -8374,7 +8790,7 @@ void RendererStorageRD::global_variable_set_override(const StringName &p_name, c } else { //texture for (Set<RID>::Element *E = gv.texture_materials.front(); E; E = E->next()) { - Material *material = material_owner.getornull(E->get()); + Material *material = material_owner.get_or_null(E->get()); ERR_CONTINUE(!material); _material_queue_update(material, false, true); } @@ -8413,10 +8829,10 @@ void RendererStorageRD::global_variables_load_settings(bool p_load_textures) { List<PropertyInfo> settings; ProjectSettings::get_singleton()->get_property_list(&settings); - for (List<PropertyInfo>::Element *E = settings.front(); E; E = E->next()) { - if (E->get().name.begins_with("shader_globals/")) { - StringName name = E->get().name.get_slice("/", 1); - Dictionary d = ProjectSettings::get_singleton()->get(E->get().name); + for (const PropertyInfo &E : settings) { + if (E.name.begins_with("shader_globals/")) { + StringName name = E.name.get_slice("/", 1); + Dictionary d = ProjectSettings::get_singleton()->get(E.name); ERR_CONTINUE(!d.has("type")); ERR_CONTINUE(!d.has("value")); @@ -8555,7 +8971,7 @@ void RendererStorageRD::global_variables_instance_update(RID p_instance, int p_i pos += p_index; - _fill_std140_variant_ubo_value(datatype, p_value, (uint8_t *)&global_variables.buffer_values[pos], true); //instances always use linear color in this renderer + _fill_std140_variant_ubo_value(datatype, 0, p_value, (uint8_t *)&global_variables.buffer_values[pos], true); //instances always use linear color in this renderer _global_variable_mark_buffer_dirty(pos, 1); } @@ -8584,8 +9000,8 @@ void RendererStorageRD::_update_global_variables() { if (global_variables.must_update_buffer_materials) { // only happens in the case of a buffer variable added or removed, // so not often. - for (List<RID>::Element *E = global_variables.materials_using_buffer.front(); E; E = E->next()) { - Material *material = material_owner.getornull(E->get()); + for (const RID &E : global_variables.materials_using_buffer) { + Material *material = material_owner.get_or_null(E); ERR_CONTINUE(!material); //wtf _material_queue_update(material, true, false); @@ -8597,8 +9013,8 @@ void RendererStorageRD::_update_global_variables() { if (global_variables.must_update_texture_materials) { // only happens in the case of a buffer variable added or removed, // so not often. - for (List<RID>::Element *E = global_variables.materials_using_texture.front(); E; E = E->next()) { - Material *material = material_owner.getornull(E->get()); + for (const RID &E : global_variables.materials_using_texture) { + Material *material = material_owner.get_or_null(E); ERR_CONTINUE(!material); //wtf _material_queue_update(material, false, true); @@ -8643,7 +9059,7 @@ bool RendererStorageRD::has_os_feature(const String &p_feature) const { bool RendererStorageRD::free(RID p_rid) { if (texture_owner.owns(p_rid)) { - Texture *t = texture_owner.getornull(p_rid); + Texture *t = texture_owner.get_or_null(p_rid); ERR_FAIL_COND_V(!t, false); ERR_FAIL_COND_V(t->is_render_target, false); @@ -8657,7 +9073,7 @@ bool RendererStorageRD::free(RID p_rid) { } if (t->is_proxy && t->proxy_to.is_valid()) { - Texture *proxy_to = texture_owner.getornull(t->proxy_to); + Texture *proxy_to = texture_owner.get_or_null(t->proxy_to); if (proxy_to) { proxy_to->proxies.erase(p_rid); } @@ -8669,7 +9085,7 @@ bool RendererStorageRD::free(RID p_rid) { } for (int i = 0; i < t->proxies.size(); i++) { - Texture *p = texture_owner.getornull(t->proxies[i]); + Texture *p = texture_owner.get_or_null(t->proxies[i]); ERR_CONTINUE(!p); p->proxy_to = RID(); p->rd_texture = RID(); @@ -8684,7 +9100,7 @@ bool RendererStorageRD::free(RID p_rid) { } else if (canvas_texture_owner.owns(p_rid)) { canvas_texture_owner.free(p_rid); } else if (shader_owner.owns(p_rid)) { - Shader *shader = shader_owner.getornull(p_rid); + Shader *shader = shader_owner.get_or_null(p_rid); //make material unreference this while (shader->owners.size()) { material_set_shader(shader->owners.front()->get()->self, RID()); @@ -8696,7 +9112,7 @@ bool RendererStorageRD::free(RID p_rid) { shader_owner.free(p_rid); } else if (material_owner.owns(p_rid)) { - Material *material = material_owner.getornull(p_rid); + Material *material = material_owner.get_or_null(p_rid); material_set_shader(p_rid, RID()); //clean up shader material->dependency.deleted_notify(p_rid); @@ -8704,7 +9120,7 @@ bool RendererStorageRD::free(RID p_rid) { } else if (mesh_owner.owns(p_rid)) { mesh_clear(p_rid); mesh_set_shadow_mesh(p_rid, RID()); - Mesh *mesh = mesh_owner.getornull(p_rid); + Mesh *mesh = mesh_owner.get_or_null(p_rid); mesh->dependency.deleted_notify(p_rid); if (mesh->instances.size()) { ERR_PRINT("deleting mesh with active instances"); @@ -8718,7 +9134,7 @@ bool RendererStorageRD::free(RID p_rid) { } mesh_owner.free(p_rid); } else if (mesh_instance_owner.owns(p_rid)) { - MeshInstance *mi = mesh_instance_owner.getornull(p_rid); + MeshInstance *mi = mesh_instance_owner.get_or_null(p_rid); _mesh_instance_clear(mi); mi->mesh->instances.erase(mi->I); mi->I = nullptr; @@ -8728,21 +9144,21 @@ bool RendererStorageRD::free(RID p_rid) { } else if (multimesh_owner.owns(p_rid)) { _update_dirty_multimeshes(); multimesh_allocate_data(p_rid, 0, RS::MULTIMESH_TRANSFORM_2D); - MultiMesh *multimesh = multimesh_owner.getornull(p_rid); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_rid); multimesh->dependency.deleted_notify(p_rid); multimesh_owner.free(p_rid); } else if (skeleton_owner.owns(p_rid)) { _update_dirty_skeletons(); skeleton_allocate_data(p_rid, 0); - Skeleton *skeleton = skeleton_owner.getornull(p_rid); + Skeleton *skeleton = skeleton_owner.get_or_null(p_rid); skeleton->dependency.deleted_notify(p_rid); skeleton_owner.free(p_rid); } else if (reflection_probe_owner.owns(p_rid)) { - ReflectionProbe *reflection_probe = reflection_probe_owner.getornull(p_rid); + ReflectionProbe *reflection_probe = reflection_probe_owner.get_or_null(p_rid); reflection_probe->dependency.deleted_notify(p_rid); reflection_probe_owner.free(p_rid); } else if (decal_owner.owns(p_rid)) { - Decal *decal = decal_owner.getornull(p_rid); + Decal *decal = decal_owner.get_or_null(p_rid); for (int i = 0; i < RS::DECAL_TEXTURE_MAX; i++) { if (decal->textures[i].is_valid() && texture_owner.owns(decal->textures[i])) { texture_remove_from_decal_atlas(decal->textures[i]); @@ -8752,30 +9168,30 @@ bool RendererStorageRD::free(RID p_rid) { decal_owner.free(p_rid); } else if (voxel_gi_owner.owns(p_rid)) { voxel_gi_allocate_data(p_rid, Transform3D(), AABB(), Vector3i(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<int>()); //deallocate - VoxelGI *voxel_gi = voxel_gi_owner.getornull(p_rid); + VoxelGI *voxel_gi = voxel_gi_owner.get_or_null(p_rid); voxel_gi->dependency.deleted_notify(p_rid); voxel_gi_owner.free(p_rid); } else if (lightmap_owner.owns(p_rid)) { lightmap_set_textures(p_rid, RID(), false); - Lightmap *lightmap = lightmap_owner.getornull(p_rid); + Lightmap *lightmap = lightmap_owner.get_or_null(p_rid); lightmap->dependency.deleted_notify(p_rid); lightmap_owner.free(p_rid); } else if (light_owner.owns(p_rid)) { light_set_projector(p_rid, RID()); //clear projector // delete the texture - Light *light = light_owner.getornull(p_rid); + Light *light = light_owner.get_or_null(p_rid); light->dependency.deleted_notify(p_rid); light_owner.free(p_rid); } else if (particles_owner.owns(p_rid)) { update_particles(); - Particles *particles = particles_owner.getornull(p_rid); + Particles *particles = particles_owner.get_or_null(p_rid); particles->dependency.deleted_notify(p_rid); _particles_free_data(particles); particles_owner.free(p_rid); } else if (particles_collision_owner.owns(p_rid)) { - ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_rid); + ParticlesCollision *particles_collision = particles_collision_owner.get_or_null(p_rid); if (particles_collision->heightfield_texture.is_valid()) { RD::get_singleton()->free(particles_collision->heightfield_texture); @@ -8783,18 +9199,18 @@ bool RendererStorageRD::free(RID p_rid) { particles_collision->dependency.deleted_notify(p_rid); particles_collision_owner.free(p_rid); } else if (visibility_notifier_owner.owns(p_rid)) { - VisibilityNotifier *vn = visibility_notifier_owner.getornull(p_rid); + VisibilityNotifier *vn = visibility_notifier_owner.get_or_null(p_rid); vn->dependency.deleted_notify(p_rid); visibility_notifier_owner.free(p_rid); } else if (particles_collision_instance_owner.owns(p_rid)) { particles_collision_instance_owner.free(p_rid); } else if (render_target_owner.owns(p_rid)) { - RenderTarget *rt = render_target_owner.getornull(p_rid); + RenderTarget *rt = render_target_owner.get_or_null(p_rid); _clear_render_target(rt); if (rt->texture.is_valid()) { - Texture *tex = texture_owner.getornull(rt->texture); + Texture *tex = texture_owner.get_or_null(rt->texture); tex->is_render_target = false; free(rt->texture); } @@ -8807,8 +9223,13 @@ bool RendererStorageRD::free(RID p_rid) { return true; } +void RendererStorageRD::init_effects(bool p_prefer_raster_effects) { + effects = memnew(EffectsRD(p_prefer_raster_effects)); +} + EffectsRD *RendererStorageRD::get_effects() { - return &effects; + ERR_FAIL_NULL_V_MSG(effects, nullptr, "Effects haven't been initialised yet."); + return effects; } void RendererStorageRD::capture_timestamps_begin() { @@ -9126,26 +9547,42 @@ RendererStorageRD::RendererStorageRD() { } break; case RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS: { sampler_state.mag_filter = RD::SAMPLER_FILTER_NEAREST; - sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; - sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + sampler_state.min_filter = RD::SAMPLER_FILTER_NEAREST; + if (GLOBAL_GET("rendering/textures/default_filters/use_nearest_mipmap_filter")) { + sampler_state.mip_filter = RD::SAMPLER_FILTER_NEAREST; + } else { + sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + } } break; case RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS: { sampler_state.mag_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; - sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + if (GLOBAL_GET("rendering/textures/default_filters/use_nearest_mipmap_filter")) { + sampler_state.mip_filter = RD::SAMPLER_FILTER_NEAREST; + } else { + sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + } } break; case RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS_ANISOTROPIC: { sampler_state.mag_filter = RD::SAMPLER_FILTER_NEAREST; - sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; - sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + sampler_state.min_filter = RD::SAMPLER_FILTER_NEAREST; + if (GLOBAL_GET("rendering/textures/default_filters/use_nearest_mipmap_filter")) { + sampler_state.mip_filter = RD::SAMPLER_FILTER_NEAREST; + } else { + sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + } sampler_state.use_anisotropy = true; sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/textures/default_filters/anisotropic_filtering_level")); } break; case RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC: { sampler_state.mag_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; - sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + if (GLOBAL_GET("rendering/textures/default_filters/use_nearest_mipmap_filter")) { + sampler_state.mip_filter = RD::SAMPLER_FILTER_NEAREST; + } else { + sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; + } sampler_state.use_anisotropy = true; sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/textures/default_filters/anisotropic_filtering_level")); @@ -9292,15 +9729,6 @@ RendererStorageRD::RendererStorageRD() { } } - { - Vector<String> sdf_versions; - sdf_versions.push_back(""); //one only - voxel_gi_sdf_shader.initialize(sdf_versions); - voxel_gi_sdf_shader_version = voxel_gi_sdf_shader.version_create(); - voxel_gi_sdf_shader_version_shader = voxel_gi_sdf_shader.version_get_shader(voxel_gi_sdf_shader_version, 0); - voxel_gi_sdf_shader_pipeline = RD::get_singleton()->compute_pipeline_create(voxel_gi_sdf_shader_version_shader); - } - using_lightmap_array = true; // high end if (using_lightmap_array) { uint32_t textures_per_stage = RD::get_singleton()->limit_get(RD::LIMIT_MAX_TEXTURES_PER_SHADER_STAGE); @@ -9389,6 +9817,8 @@ RendererStorageRD::RendererStorageRD() { particles_shader.default_shader = shader_allocate(); shader_initialize(particles_shader.default_shader); shader_set_code(particles_shader.default_shader, R"( +// Default particles shader. + shader_type particles; void process() { @@ -9520,7 +9950,6 @@ RendererStorageRD::~RendererStorageRD() { RD::get_singleton()->free(mesh_default_rd_buffers[i]); } - voxel_gi_sdf_shader.version_free(voxel_gi_sdf_shader_version); particles_shader.copy_shader.version_free(particles_shader.copy_shader_version); rt_sdf.shader.version_free(rt_sdf.shader_version); @@ -9538,4 +9967,9 @@ RendererStorageRD::~RendererStorageRD() { if (decal_atlas.texture.is_valid()) { RD::get_singleton()->free(decal_atlas.texture); } + + if (effects) { + memdelete(effects); + effects = nullptr; + } } diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.h b/servers/rendering/renderer_rd/renderer_storage_rd.h index f471874c8e..d6129bb57b 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.h +++ b/servers/rendering/renderer_rd/renderer_storage_rd.h @@ -621,7 +621,6 @@ private: float color[4]; float custom[3]; float lifetime; - uint32_t pad[3]; }; struct ParticlesFrameParams { @@ -663,7 +662,7 @@ private: uint32_t type; uint32_t texture_index; //texture index for vector field - float scale; + real_t scale; uint32_t pad[2]; }; @@ -672,8 +671,8 @@ private: float prev_system_phase; uint32_t cycle; - float explosiveness; - float randomness; + real_t explosiveness; + real_t randomness; float time; float delta; @@ -715,14 +714,14 @@ private: struct Particles { RS::ParticlesMode mode = RS::PARTICLES_MODE_3D; bool inactive = true; - float inactive_time = 0.0; + double inactive_time = 0.0; bool emitting = false; bool one_shot = false; int amount = 0; - float lifetime = 1.0; - float pre_process_time = 0.0; - float explosiveness = 0.0; - float randomness = 0.0; + double lifetime = 1.0; + double pre_process_time = 0.0; + real_t explosiveness = 0.0; + real_t randomness = 0.0; bool restart_request = false; AABB custom_aabb = AABB(Vector3(-4, -4, -4), Vector3(8, 8, 8)); bool use_local_coords = true; @@ -766,20 +765,20 @@ private: RID sub_emitter; - float phase = 0.0; - float prev_phase = 0.0; + double phase = 0.0; + double prev_phase = 0.0; uint64_t prev_ticks = 0; uint32_t random_seed = 0; uint32_t cycle_number = 0; - float speed_scale = 1.0; + double speed_scale = 1.0; int fixed_fps = 30; bool interpolate = true; bool fractional_delta = false; - float frame_remainder = 0; - float collision_base_size = 0.01; + double frame_remainder = 0; + real_t collision_base_size = 0.01; bool clear = true; @@ -796,7 +795,7 @@ private: Dependency dependency; - float trail_length = 1.0; + double trail_length = 1.0; bool trails_enabled = false; LocalVector<ParticlesFrameParams> frame_history; LocalVector<ParticlesFrameParams> trail_params; @@ -805,7 +804,7 @@ private: } }; - void _particles_process(Particles *p_particles, float p_delta); + void _particles_process(Particles *p_particles, double p_delta); void _particles_allocate_emission_buffer(Particles *particles); void _particles_free_data(Particles *particles); void _particles_update_buffers(Particles *particles); @@ -1100,11 +1099,6 @@ private: Dependency dependency; }; - VoxelGiSdfShaderRD voxel_gi_sdf_shader; - RID voxel_gi_sdf_shader_version; - RID voxel_gi_sdf_shader_version_shader; - RID voxel_gi_sdf_shader_pipeline; - mutable RID_Owner<VoxelGI, true> voxel_gi_owner; /* REFLECTION PROBE */ @@ -1290,7 +1284,7 @@ private: void _update_global_variables(); /* EFFECTS */ - EffectsRD effects; + EffectsRD *effects = nullptr; public: virtual bool can_create_resources_async() const; @@ -1356,7 +1350,7 @@ public: if (p_texture.is_null()) { return RID(); } - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); if (!tex) { return RID(); @@ -1368,7 +1362,7 @@ public: if (p_texture.is_null()) { return Size2i(); } - Texture *tex = texture_owner.getornull(p_texture); + Texture *tex = texture_owner.get_or_null(p_texture); if (!tex) { return Size2i(); @@ -1436,12 +1430,12 @@ public: void material_set_data_request_function(ShaderType p_shader_type, MaterialDataRequestFunction p_function); _FORCE_INLINE_ uint32_t material_get_shader_id(RID p_material) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); return material->shader_id; } _FORCE_INLINE_ MaterialData *material_get_data(RID p_material, ShaderType p_shader_type) { - Material *material = material_owner.getornull(p_material); + Material *material = material_owner.get_or_null(p_material); if (!material || material->shader_type != p_shader_type) { return nullptr; } else { @@ -1494,7 +1488,7 @@ public: virtual void update_mesh_instances(); _FORCE_INLINE_ const RID *mesh_get_surface_count_and_materials(RID p_mesh, uint32_t &r_surface_count) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, nullptr); r_surface_count = mesh->surface_count; if (r_surface_count == 0) { @@ -1511,7 +1505,7 @@ public: } _FORCE_INLINE_ void *mesh_get_surface(RID p_mesh, uint32_t p_surface_index) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, nullptr); ERR_FAIL_UNSIGNED_INDEX_V(p_surface_index, mesh->surface_count, nullptr); @@ -1519,7 +1513,7 @@ public: } _FORCE_INLINE_ RID mesh_get_shadow_mesh(RID p_mesh) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); ERR_FAIL_COND_V(!mesh, RID()); return mesh->shadow_mesh; @@ -1605,7 +1599,7 @@ public: } _FORCE_INLINE_ void mesh_instance_surface_get_vertex_arrays_and_format(RID p_mesh_instance, uint32_t p_surface_index, uint32_t p_input_mask, RID &r_vertex_array_rd, RD::VertexFormatID &r_vertex_format) { - MeshInstance *mi = mesh_instance_owner.getornull(p_mesh_instance); + MeshInstance *mi = mesh_instance_owner.get_or_null(p_mesh_instance); ERR_FAIL_COND(!mi); Mesh *mesh = mi->mesh; ERR_FAIL_UNSIGNED_INDEX(p_surface_index, mesh->surface_count); @@ -1646,7 +1640,7 @@ public: } _FORCE_INLINE_ uint32_t mesh_surface_get_render_pass_index(RID p_mesh, uint32_t p_surface_index, uint64_t p_render_pass, uint32_t *r_index) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); Mesh::Surface *s = mesh->surfaces[p_surface_index]; if (s->render_pass != p_render_pass) { @@ -1659,7 +1653,7 @@ public: } _FORCE_INLINE_ uint32_t mesh_surface_get_multimesh_render_pass_index(RID p_mesh, uint32_t p_surface_index, uint64_t p_render_pass, uint32_t *r_index) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); Mesh::Surface *s = mesh->surfaces[p_surface_index]; if (s->multimesh_render_pass != p_render_pass) { @@ -1672,7 +1666,7 @@ public: } _FORCE_INLINE_ uint32_t mesh_surface_get_particles_render_pass_index(RID p_mesh, uint32_t p_surface_index, uint64_t p_render_pass, uint32_t *r_index) { - Mesh *mesh = mesh_owner.getornull(p_mesh); + Mesh *mesh = mesh_owner.get_or_null(p_mesh); Mesh::Surface *s = mesh->surfaces[p_surface_index]; if (s->particles_render_pass != p_render_pass) { @@ -1714,22 +1708,22 @@ public: AABB multimesh_get_aabb(RID p_multimesh) const; _FORCE_INLINE_ RS::MultimeshTransformFormat multimesh_get_transform_format(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); return multimesh->xform_format; } _FORCE_INLINE_ bool multimesh_uses_colors(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); return multimesh->uses_colors; } _FORCE_INLINE_ bool multimesh_uses_custom_data(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); return multimesh->uses_custom_data; } _FORCE_INLINE_ uint32_t multimesh_get_instances_to_draw(RID p_multimesh) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); if (multimesh->visible_instances >= 0) { return multimesh->visible_instances; } @@ -1737,7 +1731,7 @@ public: } _FORCE_INLINE_ RID multimesh_get_3d_uniform_set(RID p_multimesh, RID p_shader, uint32_t p_set) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); if (!multimesh->uniform_set_3d.is_valid()) { Vector<RD::Uniform> uniforms; RD::Uniform u; @@ -1752,7 +1746,7 @@ public: } _FORCE_INLINE_ RID multimesh_get_2d_uniform_set(RID p_multimesh, RID p_shader, uint32_t p_set) const { - MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); + MultiMesh *multimesh = multimesh_owner.get_or_null(p_multimesh); if (!multimesh->uniform_set_2d.is_valid()) { Vector<RD::Uniform> uniforms; RD::Uniform u; @@ -1781,11 +1775,11 @@ public: Transform2D skeleton_bone_get_transform_2d(RID p_skeleton, int p_bone) const; _FORCE_INLINE_ bool skeleton_is_valid(RID p_skeleton) { - return skeleton_owner.getornull(p_skeleton) != nullptr; + return skeleton_owner.get_or_null(p_skeleton) != nullptr; } _FORCE_INLINE_ RID skeleton_get_3d_uniform_set(RID p_skeleton, RID p_shader, uint32_t p_set) const { - Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); + Skeleton *skeleton = skeleton_owner.get_or_null(p_skeleton); ERR_FAIL_COND_V(!skeleton, RID()); ERR_FAIL_COND_V(skeleton->size == 0, RID()); if (skeleton->use_2d) { @@ -1839,7 +1833,7 @@ public: RS::LightOmniShadowMode light_omni_get_shadow_mode(RID p_light); _FORCE_INLINE_ RS::LightType light_get_type(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_DIRECTIONAL); return light->type; @@ -1847,70 +1841,70 @@ public: AABB light_get_aabb(RID p_light) const; _FORCE_INLINE_ float light_get_param(RID p_light, RS::LightParam p_param) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0); return light->param[p_param]; } _FORCE_INLINE_ RID light_get_projector(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RID()); return light->projector; } _FORCE_INLINE_ Color light_get_color(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, Color()); return light->color; } _FORCE_INLINE_ Color light_get_shadow_color(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, Color()); return light->shadow_color; } _FORCE_INLINE_ uint32_t light_get_cull_mask(RID p_light) { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0); return light->cull_mask; } _FORCE_INLINE_ bool light_has_shadow(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_DIRECTIONAL); return light->shadow; } _FORCE_INLINE_ bool light_has_projector(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_DIRECTIONAL); return texture_owner.owns(light->projector); } _FORCE_INLINE_ bool light_is_negative(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, RS::LIGHT_DIRECTIONAL); return light->negative; } _FORCE_INLINE_ float light_get_transmittance_bias(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0.0); return light->param[RS::LIGHT_PARAM_TRANSMITTANCE_BIAS]; } _FORCE_INLINE_ float light_get_shadow_volumetric_fog_fade(RID p_light) const { - const Light *light = light_owner.getornull(p_light); + const Light *light = light_owner.get_or_null(p_light); ERR_FAIL_COND_V(!light, 0.0); return light->param[RS::LIGHT_PARAM_SHADOW_VOLUMETRIC_FOG_FADE]; @@ -1977,62 +1971,62 @@ public: virtual void decal_set_normal_fade(RID p_decal, float p_fade); _FORCE_INLINE_ Vector3 decal_get_extents(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->extents; } _FORCE_INLINE_ RID decal_get_texture(RID p_decal, RS::DecalTexture p_texture) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->textures[p_texture]; } _FORCE_INLINE_ Color decal_get_modulate(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->modulate; } _FORCE_INLINE_ float decal_get_emission_energy(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->emission_energy; } _FORCE_INLINE_ float decal_get_albedo_mix(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->albedo_mix; } _FORCE_INLINE_ uint32_t decal_get_cull_mask(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->cull_mask; } _FORCE_INLINE_ float decal_get_upper_fade(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->upper_fade; } _FORCE_INLINE_ float decal_get_lower_fade(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->lower_fade; } _FORCE_INLINE_ float decal_get_normal_fade(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->normal_fade; } _FORCE_INLINE_ bool decal_is_distance_fade_enabled(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->distance_fade; } _FORCE_INLINE_ float decal_get_distance_fade_begin(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->distance_fade_begin; } _FORCE_INLINE_ float decal_get_distance_fade_length(RID p_decal) { - const Decal *decal = decal_owner.getornull(p_decal); + const Decal *decal = decal_owner.get_or_null(p_decal); return decal->distance_fade_length; } @@ -2107,18 +2101,18 @@ public: return lightmap_probe_capture_update_speed; } _FORCE_INLINE_ RID lightmap_get_texture(RID p_lightmap) const { - const Lightmap *lm = lightmap_owner.getornull(p_lightmap); + const Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); ERR_FAIL_COND_V(!lm, RID()); return lm->light_texture; } _FORCE_INLINE_ int32_t lightmap_get_array_index(RID p_lightmap) const { ERR_FAIL_COND_V(!using_lightmap_array, -1); //only for arrays - const Lightmap *lm = lightmap_owner.getornull(p_lightmap); + const Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); return lm->array_index; } _FORCE_INLINE_ bool lightmap_uses_spherical_harmonics(RID p_lightmap) const { ERR_FAIL_COND_V(!using_lightmap_array, false); //only for arrays - const Lightmap *lm = lightmap_owner.getornull(p_lightmap); + const Lightmap *lm = lightmap_owner.get_or_null(p_lightmap); return lm->uses_spherical_harmonics; } _FORCE_INLINE_ uint64_t lightmap_array_get_version() const { @@ -2144,22 +2138,22 @@ public: void particles_set_mode(RID p_particles, RS::ParticlesMode p_mode); void particles_set_emitting(RID p_particles, bool p_emitting); void particles_set_amount(RID p_particles, int p_amount); - void particles_set_lifetime(RID p_particles, float p_lifetime); + void particles_set_lifetime(RID p_particles, double p_lifetime); void particles_set_one_shot(RID p_particles, bool p_one_shot); - void particles_set_pre_process_time(RID p_particles, float p_time); - void particles_set_explosiveness_ratio(RID p_particles, float p_ratio); - void particles_set_randomness_ratio(RID p_particles, float p_ratio); + void particles_set_pre_process_time(RID p_particles, double p_time); + void particles_set_explosiveness_ratio(RID p_particles, real_t p_ratio); + void particles_set_randomness_ratio(RID p_particles, real_t p_ratio); void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb); - void particles_set_speed_scale(RID p_particles, float p_scale); + void particles_set_speed_scale(RID p_particles, double p_scale); void particles_set_use_local_coordinates(RID p_particles, bool p_enable); void particles_set_process_material(RID p_particles, RID p_material); void particles_set_fixed_fps(RID p_particles, int p_fps); void particles_set_interpolate(RID p_particles, bool p_enable); void particles_set_fractional_delta(RID p_particles, bool p_enable); - void particles_set_collision_base_size(RID p_particles, float p_size); + void particles_set_collision_base_size(RID p_particles, real_t p_size); void particles_set_transform_align(RID p_particles, RS::ParticlesTransformAlign p_transform_align); - void particles_set_trails(RID p_particles, bool p_enable, float p_length); + void particles_set_trails(RID p_particles, bool p_enable, double p_length); void particles_set_trail_bind_poses(RID p_particles, const Vector<Transform3D> &p_bind_poses); void particles_restart(RID p_particles); @@ -2187,13 +2181,13 @@ public: virtual bool particles_is_inactive(RID p_particles) const; _FORCE_INLINE_ RS::ParticlesMode particles_get_mode(RID p_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, RS::PARTICLES_MODE_2D); return particles->mode; } _FORCE_INLINE_ uint32_t particles_get_amount(RID p_particles, uint32_t &r_trail_divisor) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, 0); if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { @@ -2206,21 +2200,21 @@ public: } _FORCE_INLINE_ bool particles_has_collision(RID p_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, 0); return particles->has_collision_cache; } _FORCE_INLINE_ uint32_t particles_is_using_local_coords(RID p_particles) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, false); return particles->use_local_coords; } _FORCE_INLINE_ RID particles_get_instance_buffer_uniform_set(RID p_particles, RID p_shader, uint32_t p_set) { - Particles *particles = particles_owner.getornull(p_particles); + Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND_V(!particles, RID()); if (particles->particles_transforms_buffer_uniform_set.is_null()) { _particles_update_buffers(particles); @@ -2252,11 +2246,11 @@ public: virtual void particles_collision_set_collision_type(RID p_particles_collision, RS::ParticlesCollisionType p_type); virtual void particles_collision_set_cull_mask(RID p_particles_collision, uint32_t p_cull_mask); - virtual void particles_collision_set_sphere_radius(RID p_particles_collision, float p_radius); //for spheres + virtual void particles_collision_set_sphere_radius(RID p_particles_collision, real_t p_radius); //for spheres virtual void particles_collision_set_box_extents(RID p_particles_collision, const Vector3 &p_extents); //for non-spheres - virtual void particles_collision_set_attractor_strength(RID p_particles_collision, float p_strength); - virtual void particles_collision_set_attractor_directionality(RID p_particles_collision, float p_directionality); - virtual void particles_collision_set_attractor_attenuation(RID p_particles_collision, float p_curve); + virtual void particles_collision_set_attractor_strength(RID p_particles_collision, real_t p_strength); + virtual void particles_collision_set_attractor_directionality(RID p_particles_collision, real_t p_directionality); + virtual void particles_collision_set_attractor_attenuation(RID p_particles_collision, real_t p_curve); virtual void particles_collision_set_field_texture(RID p_particles_collision, RID p_texture); //for SDF and vector field, heightfield is dynamic virtual void particles_collision_height_field_update(RID p_particles_collision); //for SDF and vector field virtual void particles_collision_set_height_field_resolution(RID p_particles_collision, RS::ParticlesCollisionHeightfieldResolution p_resolution); //for SDF and vector field @@ -2374,6 +2368,7 @@ public: static RendererStorageRD *base_singleton; + void init_effects(bool p_prefer_raster_effects); EffectsRD *get_effects(); RendererStorageRD(); diff --git a/servers/rendering/renderer_rd/shader_compiler_rd.cpp b/servers/rendering/renderer_rd/shader_compiler_rd.cpp index 1dde92d8ff..215959bb6a 100644 --- a/servers/rendering/renderer_rd/shader_compiler_rd.cpp +++ b/servers/rendering/renderer_rd/shader_compiler_rd.cpp @@ -91,7 +91,7 @@ static int _get_datatype_size(SL::DataType p_type) { case SL::TYPE_VEC4: return 16; case SL::TYPE_MAT2: - return 32; //4 * 4 + 4 * 4 + return 32; // 4 * 4 + 4 * 4 case SL::TYPE_MAT3: return 48; // 4 * 4 + 4 * 4 + 4 * 4 case SL::TYPE_MAT4: @@ -213,7 +213,7 @@ static String _interpstr(SL::DataInterpolation p_interp) { return ""; } -static String _prestr(SL::DataPrecision p_pres) { +static String _prestr(SL::DataPrecision p_pres, bool p_force_highp = false) { switch (p_pres) { case SL::PRECISION_LOWP: return "lowp "; @@ -222,7 +222,7 @@ static String _prestr(SL::DataPrecision p_pres) { case SL::PRECISION_HIGHP: return "highp "; case SL::PRECISION_DEFAULT: - return ""; + return p_force_highp ? "highp " : ""; } return ""; } @@ -566,12 +566,12 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge int max_texture_uniforms = 0; int max_uniforms = 0; - for (Map<StringName, SL::ShaderNode::Uniform>::Element *E = pnode->uniforms.front(); E; E = E->next()) { - if (SL::is_sampler_type(E->get().type)) { + for (const KeyValue<StringName, SL::ShaderNode::Uniform> &E : pnode->uniforms) { + if (SL::is_sampler_type(E.value.type)) { max_texture_uniforms++; } else { - if (E->get().scope == SL::ShaderNode::Uniform::SCOPE_INSTANCE) { - continue; //instances are indexed directly, dont need index uniforms + if (E.value.scope == SL::ShaderNode::Uniform::SCOPE_INSTANCE) { + continue; // Instances are indexed directly, don't need index uniforms. } max_uniforms++; @@ -590,8 +590,8 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge Vector<StringName> uniform_names; - for (Map<StringName, SL::ShaderNode::Uniform>::Element *E = pnode->uniforms.front(); E; E = E->next()) { - uniform_names.push_back(E->key()); + for (const KeyValue<StringName, SL::ShaderNode::Uniform> &E : pnode->uniforms) { + uniform_names.push_back(E.key); } uniform_names.sort_custom<StringName::AlphCompare>(); //ensure order is deterministic so the same shader is always produced @@ -605,10 +605,10 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge if (uniform.scope == SL::ShaderNode::Uniform::SCOPE_INSTANCE) { //insert, but don't generate any code. p_actions.uniforms->insert(uniform_name, uniform); - continue; //instances are indexed directly, dont need index uniforms + continue; // Instances are indexed directly, don't need index uniforms. } if (SL::is_sampler_type(uniform.type)) { - ucode = "layout(set = " + itos(actions.texture_layout_set) + ", binding = " + itos(actions.base_texture_binding_index + uniform.texture_order) + ") uniform "; + ucode = "layout(set = " + itos(actions.texture_layout_set) + ", binding = " + itos(actions.base_texture_binding_index + uniform.texture_binding) + ") uniform "; } bool is_buffer_global = !SL::is_sampler_type(uniform.type) && uniform.scope == SL::ShaderNode::Uniform::SCOPE_GLOBAL; @@ -617,11 +617,16 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge //this is an integer to index the global table ucode += _typestr(ShaderLanguage::TYPE_UINT); } else { - ucode += _prestr(uniform.precision); + ucode += _prestr(uniform.precision, ShaderLanguage::is_float_type(uniform.type)); ucode += _typestr(uniform.type); } ucode += " " + _mkid(uniform_name); + if (uniform.array_size > 0) { + ucode += "["; + ucode += itos(uniform.array_size); + ucode += "]"; + } ucode += ";\n"; if (SL::is_sampler_type(uniform.type)) { for (int j = 0; j < STAGE_MAX; j++) { @@ -635,6 +640,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge texture.filter = uniform.filter; texture.repeat = uniform.repeat; texture.global = uniform.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL; + texture.array_size = uniform.array_size; if (texture.global) { r_gen_code.uses_global_textures = true; } @@ -650,7 +656,16 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge uniform_sizes.write[uniform.order] = _get_datatype_size(ShaderLanguage::TYPE_UINT); uniform_alignments.write[uniform.order] = _get_datatype_alignment(ShaderLanguage::TYPE_UINT); } else { - uniform_sizes.write[uniform.order] = _get_datatype_size(uniform.type); + if (uniform.array_size > 0) { + int size = _get_datatype_size(uniform.type) * uniform.array_size; + int m = (16 * uniform.array_size); + if ((size % m) != 0) { + size += m - (size % m); + } + uniform_sizes.write[uniform.order] = size; + } else { + uniform_sizes.write[uniform.order] = _get_datatype_size(uniform.type); + } uniform_alignments.write[uniform.order] = _get_datatype_alignment(uniform.type); } } @@ -724,8 +739,8 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge Vector<StringName> varying_names; - for (Map<StringName, SL::ShaderNode::Varying>::Element *E = pnode->varyings.front(); E; E = E->next()) { - varying_names.push_back(E->key()); + for (const KeyValue<StringName, SL::ShaderNode::Varying> &E : pnode->varyings) { + varying_names.push_back(E.key); } varying_names.sort_custom<StringName::AlphCompare>(); //ensure order is deterministic so the same shader is always produced @@ -742,7 +757,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge String vcode; String interp_mode = _interpstr(varying.interpolation); - vcode += _prestr(varying.precision); + vcode += _prestr(varying.precision, ShaderLanguage::is_float_type(varying.type)); vcode += _typestr(varying.type); vcode += " " + _mkid(varying_name); if (varying.array_size > 0) { @@ -760,11 +775,11 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge if (var_frag_to_light.size() > 0) { String gcode = "\n\nstruct {\n"; - for (List<Pair<StringName, SL::ShaderNode::Varying>>::Element *E = var_frag_to_light.front(); E; E = E->next()) { - gcode += "\t" + _prestr(E->get().second.precision) + _typestr(E->get().second.type) + " " + _mkid(E->get().first); - if (E->get().second.array_size > 0) { + for (const Pair<StringName, SL::ShaderNode::Varying> &E : var_frag_to_light) { + gcode += "\t" + _prestr(E.second.precision) + _typestr(E.second.type) + " " + _mkid(E.first); + if (E.second.array_size > 0) { gcode += "["; - gcode += itos(E->get().second.array_size); + gcode += itos(E.second.array_size); gcode += "]"; } gcode += ";\n"; @@ -777,7 +792,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge const SL::ShaderNode::Constant &cnode = pnode->vconstants[i]; String gcode; gcode += "const "; - gcode += _prestr(cnode.precision); + gcode += _prestr(cnode.precision, ShaderLanguage::is_float_type(cnode.type)); if (cnode.type == SL::TYPE_STRUCT) { gcode += _mkid(cnode.type_str); } else { @@ -887,7 +902,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge SL::VariableNode *vnode = (SL::VariableNode *)p_node; bool use_fragment_varying = false; - if (!(p_actions.entry_point_stages.has(current_func_name) && p_actions.entry_point_stages[current_func_name] == STAGE_VERTEX)) { + if (!vnode->is_local && !(p_actions.entry_point_stages.has(current_func_name) && p_actions.entry_point_stages[current_func_name] == STAGE_VERTEX)) { if (p_assigning) { if (shader->varyings.has(vnode->name)) { use_fragment_varying = true; @@ -1037,7 +1052,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge SL::ArrayNode *anode = (SL::ArrayNode *)p_node; bool use_fragment_varying = false; - if (!(p_actions.entry_point_stages.has(current_func_name) && p_actions.entry_point_stages[current_func_name] == STAGE_VERTEX)) { + if (!anode->is_local && !(p_actions.entry_point_stages.has(current_func_name) && p_actions.entry_point_stages[current_func_name] == STAGE_VERTEX)) { if (anode->assign_expression != nullptr && shader->varyings.has(anode->name)) { use_fragment_varying = true; } else { @@ -1074,10 +1089,32 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge if (p_default_actions.renames.has(anode->name)) { code = p_default_actions.renames[anode->name]; } else { - if (use_fragment_varying) { - code = "frag_to_light."; + if (shader->uniforms.has(anode->name)) { + //its a uniform! + const ShaderLanguage::ShaderNode::Uniform &u = shader->uniforms[anode->name]; + if (u.texture_order >= 0) { + code = _mkid(anode->name); //texture, use as is + } else { + //a scalar or vector + if (u.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL) { + code = actions.base_uniform_string + _mkid(anode->name); //texture, use as is + //global variable, this means the code points to an index to the global table + code = _get_global_variable_from_type_and_index(p_default_actions.global_buffer_array_variable, code, u.type); + } else if (u.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + //instance variable, index it as such + code = "(" + p_default_actions.instance_uniform_index_variable + "+" + itos(u.instance_index) + ")"; + code = _get_global_variable_from_type_and_index(p_default_actions.global_buffer_array_variable, code, u.type); + } else { + //regular uniform, index from UBO + code = actions.base_uniform_string + _mkid(anode->name); + } + } + } else { + if (use_fragment_varying) { + code = "frag_to_light."; + } + code += _mkid(anode->name); } - code += _mkid(anode->name); } if (anode->call_expression != nullptr) { @@ -1165,6 +1202,7 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge SL::VariableNode *vnode = (SL::VariableNode *)onode->arguments[0]; bool is_texture_func = false; + bool is_screen_texture = false; if (onode->op == SL::OP_STRUCT) { code += _mkid(vnode->name); } else if (onode->op == SL::OP_CONSTRUCT) { @@ -1192,50 +1230,73 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge code += ", "; } String node_code = _dump_node_code(onode->arguments[i], p_level, r_gen_code, p_actions, p_default_actions, p_assigning); - if (is_texture_func && i == 1 && onode->arguments[i]->type == SL::Node::TYPE_VARIABLE) { + if (is_texture_func && i == 1) { //need to map from texture to sampler in order to sample - const SL::VariableNode *varnode = static_cast<const SL::VariableNode *>(onode->arguments[i]); + StringName texture_uniform; + bool correct_texture_uniform = false; + + switch (onode->arguments[i]->type) { + case SL::Node::TYPE_VARIABLE: { + const SL::VariableNode *varnode = static_cast<const SL::VariableNode *>(onode->arguments[i]); + texture_uniform = varnode->name; + correct_texture_uniform = true; + } break; + case SL::Node::TYPE_ARRAY: { + const SL::ArrayNode *anode = static_cast<const SL::ArrayNode *>(onode->arguments[i]); + texture_uniform = anode->name; + correct_texture_uniform = true; + } break; + default: + break; + } - StringName texture_uniform = varnode->name; + if (correct_texture_uniform) { + is_screen_texture = (texture_uniform == "SCREEN_TEXTURE"); - String sampler_name; + String sampler_name; - if (actions.custom_samplers.has(texture_uniform)) { - sampler_name = actions.custom_samplers[texture_uniform]; - } else { - if (shader->uniforms.has(texture_uniform)) { - sampler_name = _get_sampler_name(shader->uniforms[texture_uniform].filter, shader->uniforms[texture_uniform].repeat); + if (actions.custom_samplers.has(texture_uniform)) { + sampler_name = actions.custom_samplers[texture_uniform]; } else { - bool found = false; - - for (int j = 0; j < function->arguments.size(); j++) { - if (function->arguments[j].name == texture_uniform) { - if (function->arguments[j].tex_builtin_check) { - ERR_CONTINUE(!actions.custom_samplers.has(function->arguments[j].tex_builtin)); - sampler_name = actions.custom_samplers[function->arguments[j].tex_builtin]; - found = true; - break; - } - if (function->arguments[j].tex_argument_check) { - sampler_name = _get_sampler_name(function->arguments[j].tex_argument_filter, function->arguments[j].tex_argument_repeat); - found = true; - break; + if (shader->uniforms.has(texture_uniform)) { + sampler_name = _get_sampler_name(shader->uniforms[texture_uniform].filter, shader->uniforms[texture_uniform].repeat); + } else { + bool found = false; + + for (int j = 0; j < function->arguments.size(); j++) { + if (function->arguments[j].name == texture_uniform) { + if (function->arguments[j].tex_builtin_check) { + ERR_CONTINUE(!actions.custom_samplers.has(function->arguments[j].tex_builtin)); + sampler_name = actions.custom_samplers[function->arguments[j].tex_builtin]; + found = true; + break; + } + if (function->arguments[j].tex_argument_check) { + sampler_name = _get_sampler_name(function->arguments[j].tex_argument_filter, function->arguments[j].tex_argument_repeat); + found = true; + break; + } } } - } - if (!found) { - //function was most likely unused, so use anything (compiler will remove it anyway) - sampler_name = _get_sampler_name(ShaderLanguage::FILTER_DEFAULT, ShaderLanguage::REPEAT_DEFAULT); + if (!found) { + //function was most likely unused, so use anything (compiler will remove it anyway) + sampler_name = _get_sampler_name(ShaderLanguage::FILTER_DEFAULT, ShaderLanguage::REPEAT_DEFAULT); + } } } - } - code += ShaderLanguage::get_datatype_name(onode->arguments[i]->get_datatype()) + "(" + node_code + ", " + sampler_name + ")"; + code += ShaderLanguage::get_datatype_name(onode->arguments[i]->get_datatype()) + "(" + node_code + ", " + sampler_name + ")"; + } else { + code += node_code; + } } else { code += node_code; } } code += ")"; + if (is_screen_texture && actions.apply_luminance_multiplier) { + code = "(" + code + " * vec4(vec3(sc_luminance_multiplier), 1.0))"; + } } break; case SL::OP_INDEX: { code += _dump_node_code(onode->arguments[0], p_level, r_gen_code, p_actions, p_default_actions, p_assigning); @@ -1360,7 +1421,7 @@ Error ShaderCompilerRD::compile(RS::ShaderMode p_mode, const String &p_code, Ide } } - _err_print_error(nullptr, p_path.utf8().get_data(), parser.get_error_line(), parser.get_error_text().utf8().get_data(), ERR_HANDLER_SHADER); + _err_print_error(nullptr, p_path.utf8().get_data(), parser.get_error_line(), parser.get_error_text().utf8().get_data(), false, ERR_HANDLER_SHADER); return err; } @@ -1394,14 +1455,15 @@ void ShaderCompilerRD::initialize(DefaultIdentifierActions p_actions) { ShaderLanguage::get_builtin_funcs(&func_list); - for (List<String>::Element *E = func_list.front(); E; E = E->next()) { - internal_functions.insert(E->get()); + for (const String &E : func_list) { + internal_functions.insert(E); } texture_functions.insert("texture"); texture_functions.insert("textureProj"); texture_functions.insert("textureLod"); texture_functions.insert("textureProjLod"); texture_functions.insert("textureGrad"); + texture_functions.insert("textureGather"); texture_functions.insert("textureSize"); texture_functions.insert("texelFetch"); } diff --git a/servers/rendering/renderer_rd/shader_compiler_rd.h b/servers/rendering/renderer_rd/shader_compiler_rd.h index 2da127ffa3..2ab689c27c 100644 --- a/servers/rendering/renderer_rd/shader_compiler_rd.h +++ b/servers/rendering/renderer_rd/shader_compiler_rd.h @@ -65,6 +65,7 @@ public: ShaderLanguage::TextureFilter filter; ShaderLanguage::TextureRepeat repeat; bool global; + int array_size; }; Vector<Texture> texture_uniforms; @@ -95,6 +96,7 @@ public: String global_buffer_array_variable; String instance_uniform_index_variable; uint32_t base_varying_index = 0; + bool apply_luminance_multiplier = false; }; private: diff --git a/servers/rendering/renderer_rd/shader_rd.cpp b/servers/rendering/renderer_rd/shader_rd.cpp index 27305cc938..b9a8947fa2 100644 --- a/servers/rendering/renderer_rd/shader_rd.cpp +++ b/servers/rendering/renderer_rd/shader_rd.cpp @@ -116,8 +116,10 @@ void ShaderRD::setup(const char *p_vertex_code, const char *p_fragment_code, con } StringBuilder tohash; - tohash.append("[VersionKey]"); - tohash.append(RenderingDevice::get_singleton()->shader_get_cache_key()); + tohash.append("[SpirvCacheKey]"); + tohash.append(RenderingDevice::get_singleton()->shader_get_spirv_cache_key()); + tohash.append("[BinaryCacheKey]"); + tohash.append(RenderingDevice::get_singleton()->shader_get_binary_cache_key()); tohash.append("[Vertex]"); tohash.append(p_vertex_code ? p_vertex_code : ""); tohash.append("[Fragment]"); @@ -144,12 +146,14 @@ void ShaderRD::_clear_version(Version *p_version) { //clear versions if they exist if (p_version->variants) { for (int i = 0; i < variant_defines.size(); i++) { - RD::get_singleton()->free(p_version->variants[i]); + if (variants_enabled[i]) { + RD::get_singleton()->free(p_version->variants[i]); + } } memdelete_arr(p_version->variants); - if (p_version->variant_stages) { - memdelete_arr(p_version->variant_stages); + if (p_version->variant_data) { + memdelete_arr(p_version->variant_data); } p_version->variants = nullptr; } @@ -170,8 +174,8 @@ void ShaderRD::_build_variant_code(StringBuilder &builder, uint32_t p_variant, c if (p_version->uniforms.size()) { builder.append("#define MATERIAL_UNIFORMS_USED\n"); } - for (Map<StringName, CharString>::Element *E = p_version->code_sections.front(); E; E = E->next()) { - builder.append(String("#define ") + String(E->key()) + "_CODE_USED\n"); + for (const KeyValue<StringName, CharString> &E : p_version->code_sections) { + builder.append(String("#define ") + String(E.key) + "_CODE_USED\n"); } } break; case StageTemplate::Chunk::TYPE_MATERIAL_UNIFORMS: { @@ -203,7 +207,7 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { return; //variant is disabled, return } - Vector<RD::ShaderStageData> &stages = p_version->variant_stages[p_variant]; + Vector<RD::ShaderStageSPIRVData> stages; String error; String current_source; @@ -217,8 +221,8 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { _build_variant_code(builder, p_variant, p_version, stage_templates[STAGE_TYPE_VERTEX]); current_source = builder.as_string(); - RD::ShaderStageData stage; - stage.spir_v = RD::get_singleton()->shader_compile_from_source(RD::SHADER_STAGE_VERTEX, current_source, RD::SHADER_LANGUAGE_GLSL, &error); + RD::ShaderStageSPIRVData stage; + stage.spir_v = RD::get_singleton()->shader_compile_spirv_from_source(RD::SHADER_STAGE_VERTEX, current_source, RD::SHADER_LANGUAGE_GLSL, &error); if (stage.spir_v.size() == 0) { build_ok = false; } else { @@ -235,8 +239,8 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { _build_variant_code(builder, p_variant, p_version, stage_templates[STAGE_TYPE_FRAGMENT]); current_source = builder.as_string(); - RD::ShaderStageData stage; - stage.spir_v = RD::get_singleton()->shader_compile_from_source(RD::SHADER_STAGE_FRAGMENT, current_source, RD::SHADER_LANGUAGE_GLSL, &error); + RD::ShaderStageSPIRVData stage; + stage.spir_v = RD::get_singleton()->shader_compile_spirv_from_source(RD::SHADER_STAGE_FRAGMENT, current_source, RD::SHADER_LANGUAGE_GLSL, &error); if (stage.spir_v.size() == 0) { build_ok = false; } else { @@ -254,8 +258,8 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { current_source = builder.as_string(); - RD::ShaderStageData stage; - stage.spir_v = RD::get_singleton()->shader_compile_from_source(RD::SHADER_STAGE_COMPUTE, current_source, RD::SHADER_LANGUAGE_GLSL, &error); + RD::ShaderStageSPIRVData stage; + stage.spir_v = RD::get_singleton()->shader_compile_spirv_from_source(RD::SHADER_STAGE_COMPUTE, current_source, RD::SHADER_LANGUAGE_GLSL, &error); if (stage.spir_v.size() == 0) { build_ok = false; } else { @@ -275,15 +279,20 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { return; } - RID shader = RD::get_singleton()->shader_create(stages); + Vector<uint8_t> shader_data = RD::get_singleton()->shader_compile_binary_from_spirv(stages, name + ":" + itos(p_variant)); + + ERR_FAIL_COND(shader_data.size() == 0); + + RID shader = RD::get_singleton()->shader_create_from_bytecode(shader_data); { MutexLock lock(variant_set_mutex); p_version->variants[p_variant] = shader; + p_version->variant_data[p_variant] = shader_data; } } RS::ShaderNativeSourceCode ShaderRD::version_get_native_source_code(RID p_version) { - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); RS::ShaderNativeSourceCode source_code; ERR_FAIL_COND_V(!version, source_code); @@ -346,8 +355,8 @@ String ShaderRD::_version_get_sha1(Version *p_version) const { hash_build.append(p_version->compute_globals.get_data()); Vector<StringName> code_sections; - for (Map<StringName, CharString>::Element *E = p_version->code_sections.front(); E; E = E->next()) { - code_sections.push_back(E->key()); + for (const KeyValue<StringName, CharString> &E : p_version->code_sections) { + code_sections.push_back(E.key); } code_sections.sort_custom<StringName::AlphCompare>(); @@ -364,14 +373,12 @@ String ShaderRD::_version_get_sha1(Version *p_version) const { } static const char *shader_file_header = "GDSC"; -static const uint32_t cache_file_version = 1; +static const uint32_t cache_file_version = 2; bool ShaderRD::_load_from_cache(Version *p_version) { String sha1 = _version_get_sha1(p_version); String path = shader_cache_dir.plus_file(name).plus_file(base_sha256).plus_file(sha1) + ".cache"; - uint64_t time_from = OS::get_singleton()->get_ticks_usec(); - FileAccessRef f = FileAccess::open(path, FileAccess::READ); if (!f) { return false; @@ -390,76 +397,43 @@ bool ShaderRD::_load_from_cache(Version *p_version) { ERR_FAIL_COND_V(variant_count != (uint32_t)variant_defines.size(), false); //should not happen but check - bool success = true; for (uint32_t i = 0; i < variant_count; i++) { - uint32_t stage_count = f->get_32(); - p_version->variant_stages[i].resize(stage_count); - for (uint32_t j = 0; j < stage_count; j++) { - p_version->variant_stages[i].write[j].shader_stage = RD::ShaderStage(f->get_32()); - - int compression = f->get_32(); - uint32_t length = f->get_32(); - - if (compression == 0) { - Vector<uint8_t> data; - data.resize(length); - - f->get_buffer(data.ptrw(), length); - - p_version->variant_stages[i].write[j].spir_v = data; - } else { - Vector<uint8_t> data; - - if (compression == 2) { - //zstd - int smol_length = f->get_32(); - Vector<uint8_t> zstd_data; - - zstd_data.resize(smol_length); - f->get_buffer(zstd_data.ptrw(), smol_length); - - data.resize(length); - Compression::decompress(data.ptrw(), data.size(), zstd_data.ptr(), zstd_data.size(), Compression::MODE_ZSTD); - - } else { - data.resize(length); - f->get_buffer(data.ptrw(), length); - } - - Vector<uint8_t> spirv; - uint32_t spirv_size = smolv::GetDecodedBufferSize(data.ptr(), data.size()); - spirv.resize(spirv_size); - if (!smolv::Decode(data.ptr(), data.size(), spirv.ptrw(), spirv_size)) { - ERR_PRINT("Malformed smolv input uncompressing shader " + name + ", variant #" + itos(i) + " stage :" + itos(j)); - success = false; - break; - } - p_version->variant_stages[i].write[j].spir_v = spirv; - } + uint32_t variant_size = f->get_32(); + ERR_FAIL_COND_V(variant_size == 0 && variants_enabled[i], false); + if (!variants_enabled[i]) { + continue; } - } + Vector<uint8_t> variant_bytes; + variant_bytes.resize(variant_size); - if (!success) { - for (uint32_t i = 0; i < variant_count; i++) { - p_version->variant_stages[i].resize(0); - } - return false; - } + uint32_t br = f->get_buffer(variant_bytes.ptrw(), variant_size); - float time_ms = double(OS::get_singleton()->get_ticks_usec() - time_from) / 1000.0; + ERR_FAIL_COND_V(br != variant_size, false); - print_verbose("Shader cache load success '" + path + "' " + rtos(time_ms) + "ms."); + p_version->variant_data[i] = variant_bytes; + } for (uint32_t i = 0; i < variant_count; i++) { - RID shader = RD::get_singleton()->shader_create(p_version->variant_stages[i]); + if (!variants_enabled[i]) { + MutexLock lock(variant_set_mutex); + p_version->variants[i] = RID(); + continue; + } + RID shader = RD::get_singleton()->shader_create_from_bytecode(p_version->variant_data[i]); + if (shader.is_null()) { + for (uint32_t j = 0; j < i; j++) { + RD::get_singleton()->free(p_version->variants[i]); + } + ERR_FAIL_COND_V(shader.is_null(), false); + } { MutexLock lock(variant_set_mutex); p_version->variants[i] = shader; } } - memdelete_arr(p_version->variant_stages); //clear stages - p_version->variant_stages = nullptr; + memdelete_arr(p_version->variant_data); //clear stages + p_version->variant_data = nullptr; p_version->valid = true; return true; } @@ -476,49 +450,8 @@ void ShaderRD::_save_to_cache(Version *p_version) { f->store_32(variant_count); //variant count for (uint32_t i = 0; i < variant_count; i++) { - f->store_32(p_version->variant_stages[i].size()); //stage count - for (int j = 0; j < p_version->variant_stages[i].size(); j++) { - f->store_32(p_version->variant_stages[i][j].shader_stage); //stage count - Vector<uint8_t> spirv = p_version->variant_stages[i][j].spir_v; - - bool save_uncompressed = true; - if (shader_cache_save_compressed) { - smolv::ByteArray smolv; - bool strip_debug = !shader_cache_save_debug; - if (!smolv::Encode(spirv.ptr(), spirv.size(), smolv, strip_debug ? smolv::kEncodeFlagStripDebugInfo : 0)) { - ERR_PRINT("Error compressing shader " + name + ", variant #" + itos(i) + " stage :" + itos(i)); - } else { - bool compress_zstd = shader_cache_save_compressed_zstd; - - if (compress_zstd) { - Vector<uint8_t> zstd; - zstd.resize(Compression::get_max_compressed_buffer_size(smolv.size(), Compression::MODE_ZSTD)); - int dst_size = Compression::compress(zstd.ptrw(), &smolv[0], smolv.size(), Compression::MODE_ZSTD); - if (dst_size >= 0 && (uint32_t)dst_size < smolv.size()) { - f->store_32(2); //compressed zstd - f->store_32(smolv.size()); //size of smolv buffer - f->store_32(dst_size); //size of smolv buffer - f->store_buffer(zstd.ptr(), dst_size); //smolv buffer - } else { - compress_zstd = false; - } - } - - if (!compress_zstd) { - f->store_32(1); //compressed - f->store_32(smolv.size()); //size of smolv buffer - f->store_buffer(&smolv[0], smolv.size()); //smolv buffer - } - save_uncompressed = false; - } - } - - if (save_uncompressed) { - f->store_32(0); //uncompressed - f->store_32(spirv.size()); //stage count - f->store_buffer(spirv.ptr(), spirv.size()); //stage count - } - } + f->store_32(p_version->variant_data[i].size()); //stage count + f->store_buffer(p_version->variant_data[i].ptr(), p_version->variant_data[i].size()); } f->close(); @@ -531,8 +464,8 @@ void ShaderRD::_compile_version(Version *p_version) { p_version->dirty = false; p_version->variants = memnew_arr(RID, variant_defines.size()); - typedef Vector<RD::ShaderStageData> ShaderStageArray; - p_version->variant_stages = memnew_arr(ShaderStageArray, variant_defines.size()); + typedef Vector<uint8_t> ShaderStageData; + p_version->variant_data = memnew_arr(ShaderStageData, variant_defines.size()); if (shader_cache_dir_valid) { if (_load_from_cache(p_version)) { @@ -571,19 +504,19 @@ void ShaderRD::_compile_version(Version *p_version) { } } memdelete_arr(p_version->variants); - if (p_version->variant_stages) { - memdelete_arr(p_version->variant_stages); + if (p_version->variant_data) { + memdelete_arr(p_version->variant_data); } p_version->variants = nullptr; - p_version->variant_stages = nullptr; + p_version->variant_data = nullptr; return; } else if (shader_cache_dir_valid) { //save shader cache _save_to_cache(p_version); } - memdelete_arr(p_version->variant_stages); //clear stages - p_version->variant_stages = nullptr; + memdelete_arr(p_version->variant_data); //clear stages + p_version->variant_data = nullptr; p_version->valid = true; } @@ -591,14 +524,14 @@ void ShaderRD::_compile_version(Version *p_version) { void ShaderRD::version_set_code(RID p_version, const Map<String, String> &p_code, const String &p_uniforms, const String &p_vertex_globals, const String &p_fragment_globals, const Vector<String> &p_custom_defines) { ERR_FAIL_COND(is_compute); - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); ERR_FAIL_COND(!version); version->vertex_globals = p_vertex_globals.utf8(); version->fragment_globals = p_fragment_globals.utf8(); version->uniforms = p_uniforms.utf8(); version->code_sections.clear(); - for (Map<String, String>::Element *E = p_code.front(); E; E = E->next()) { - version->code_sections[StringName(E->key().to_upper())] = E->get().utf8(); + for (const KeyValue<String, String> &E : p_code) { + version->code_sections[StringName(E.key.to_upper())] = E.value.utf8(); } version->custom_defines.clear(); @@ -616,15 +549,15 @@ void ShaderRD::version_set_code(RID p_version, const Map<String, String> &p_code void ShaderRD::version_set_compute_code(RID p_version, const Map<String, String> &p_code, const String &p_uniforms, const String &p_compute_globals, const Vector<String> &p_custom_defines) { ERR_FAIL_COND(!is_compute); - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); ERR_FAIL_COND(!version); version->compute_globals = p_compute_globals.utf8(); version->uniforms = p_uniforms.utf8(); version->code_sections.clear(); - for (Map<String, String>::Element *E = p_code.front(); E; E = E->next()) { - version->code_sections[StringName(E->key().to_upper())] = E->get().utf8(); + for (const KeyValue<String, String> &E : p_code) { + version->code_sections[StringName(E.key.to_upper())] = E.value.utf8(); } version->custom_defines.clear(); @@ -640,7 +573,7 @@ void ShaderRD::version_set_compute_code(RID p_version, const Map<String, String> } bool ShaderRD::version_is_valid(RID p_version) { - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); ERR_FAIL_COND_V(!version, false); if (version->dirty) { @@ -652,7 +585,7 @@ bool ShaderRD::version_is_valid(RID p_version) { bool ShaderRD::version_free(RID p_version) { if (version_owner.owns(p_version)) { - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); _clear_version(version); version_owner.free(p_version); } else { diff --git a/servers/rendering/renderer_rd/shader_rd.h b/servers/rendering/renderer_rd/shader_rd.h index 9a68e02007..984b168659 100644 --- a/servers/rendering/renderer_rd/shader_rd.h +++ b/servers/rendering/renderer_rd/shader_rd.h @@ -59,7 +59,7 @@ class ShaderRD { Map<StringName, CharString> code_sections; Vector<CharString> custom_defines; - Vector<RD::ShaderStageData> *variant_stages = nullptr; + Vector<uint8_t> *variant_data = nullptr; RID *variants = nullptr; //same size as version defines bool valid; @@ -141,7 +141,7 @@ public: ERR_FAIL_INDEX_V(p_variant, variant_defines.size(), RID()); ERR_FAIL_COND_V(!variants_enabled[p_variant], RID()); - Version *version = version_owner.getornull(p_version); + Version *version = version_owner.get_or_null(p_version); ERR_FAIL_COND_V(!version, RID()); if (version->dirty) { diff --git a/servers/rendering/renderer_rd/shaders/blit.glsl b/servers/rendering/renderer_rd/shaders/blit.glsl index 967da1e6e4..8051f96738 100644 --- a/servers/rendering/renderer_rd/shaders/blit.glsl +++ b/servers/rendering/renderer_rd/shaders/blit.glsl @@ -5,6 +5,7 @@ #VERSION_DEFINES layout(push_constant, binding = 0, std140) uniform Pos { + vec4 src_rect; vec4 dst_rect; vec2 eye_center; @@ -22,8 +23,8 @@ layout(location = 0) out vec2 uv; void main() { vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); - uv = base_arr[gl_VertexIndex]; - vec2 vtx = data.dst_rect.xy + uv * data.dst_rect.zw; + uv = data.src_rect.xy + base_arr[gl_VertexIndex] * data.src_rect.zw; + vec2 vtx = data.dst_rect.xy + base_arr[gl_VertexIndex] * data.dst_rect.zw; gl_Position = vec4(vtx * 2.0 - 1.0, 0.0, 1.0); } @@ -34,6 +35,7 @@ void main() { #VERSION_DEFINES layout(push_constant, binding = 0, std140) uniform Pos { + vec4 src_rect; vec4 dst_rect; vec2 eye_center; diff --git a/servers/rendering/renderer_rd/shaders/blur_raster.glsl b/servers/rendering/renderer_rd/shaders/blur_raster.glsl new file mode 100644 index 0000000000..f8b4e3f610 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/blur_raster.glsl @@ -0,0 +1,138 @@ +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "blur_raster_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + + gl_Position = vec4(uv_interp * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "blur_raster_inc.glsl" + +layout(location = 0) in vec2 uv_interp; +/* clang-format on */ + +layout(set = 0, binding = 0) uniform sampler2D source_color; + +#ifdef GLOW_USE_AUTO_EXPOSURE +layout(set = 1, binding = 0) uniform sampler2D source_auto_exposure; +#endif + +layout(location = 0) out vec4 frag_color; + +void main() { + // We do not apply our color scale for our mobile renderer here, we'll leave our colors at half brightness and apply scale in the tonemap raster. + +#ifdef MODE_MIPMAP + + vec2 pix_size = blur.pixel_size; + vec4 color = texture(source_color, uv_interp + vec2(-0.5, -0.5) * pix_size); + color += texture(source_color, uv_interp + vec2(0.5, -0.5) * pix_size); + color += texture(source_color, uv_interp + vec2(0.5, 0.5) * pix_size); + color += texture(source_color, uv_interp + vec2(-0.5, 0.5) * pix_size); + frag_color = color / 4.0; + +#endif + +#ifdef MODE_GAUSSIAN_BLUR + + //Simpler blur uses SIGMA2 for the gaussian kernel for a stronger effect + + if (bool(blur.flags & FLAG_HORIZONTAL)) { + vec2 pix_size = blur.pixel_size; + pix_size *= 0.5; //reading from larger buffer, so use more samples + vec4 color = texture(source_color, uv_interp + vec2(0.0, 0.0) * pix_size) * 0.214607; + color += texture(source_color, uv_interp + vec2(1.0, 0.0) * pix_size) * 0.189879; + color += texture(source_color, uv_interp + vec2(2.0, 0.0) * pix_size) * 0.131514; + color += texture(source_color, uv_interp + vec2(3.0, 0.0) * pix_size) * 0.071303; + color += texture(source_color, uv_interp + vec2(-1.0, 0.0) * pix_size) * 0.189879; + color += texture(source_color, uv_interp + vec2(-2.0, 0.0) * pix_size) * 0.131514; + color += texture(source_color, uv_interp + vec2(-3.0, 0.0) * pix_size) * 0.071303; + frag_color = color; + } else { + vec2 pix_size = blur.pixel_size; + vec4 color = texture(source_color, uv_interp + vec2(0.0, 0.0) * pix_size) * 0.38774; + color += texture(source_color, uv_interp + vec2(0.0, 1.0) * pix_size) * 0.24477; + color += texture(source_color, uv_interp + vec2(0.0, 2.0) * pix_size) * 0.06136; + color += texture(source_color, uv_interp + vec2(0.0, -1.0) * pix_size) * 0.24477; + color += texture(source_color, uv_interp + vec2(0.0, -2.0) * pix_size) * 0.06136; + frag_color = color; + } +#endif + +#ifdef MODE_GAUSSIAN_GLOW + + //Glow uses larger sigma 1 for a more rounded blur effect + +#define GLOW_ADD(m_ofs, m_mult) \ + { \ + vec2 ofs = uv_interp + m_ofs * pix_size; \ + vec4 c = texture(source_color, ofs) * m_mult; \ + if (any(lessThan(ofs, vec2(0.0))) || any(greaterThan(ofs, vec2(1.0)))) { \ + c *= 0.0; \ + } \ + color += c; \ + } + + if (bool(blur.flags & FLAG_HORIZONTAL)) { + vec2 pix_size = blur.pixel_size; + pix_size *= 0.5; //reading from larger buffer, so use more samples + vec4 color = texture(source_color, uv_interp + vec2(0.0, 0.0) * pix_size) * 0.174938; + GLOW_ADD(vec2(1.0, 0.0), 0.165569); + GLOW_ADD(vec2(2.0, 0.0), 0.140367); + GLOW_ADD(vec2(3.0, 0.0), 0.106595); + GLOW_ADD(vec2(-1.0, 0.0), 0.165569); + GLOW_ADD(vec2(-2.0, 0.0), 0.140367); + GLOW_ADD(vec2(-3.0, 0.0), 0.106595); + color *= blur.glow_strength; + frag_color = color; + } else { + vec2 pix_size = blur.pixel_size; + vec4 color = texture(source_color, uv_interp + vec2(0.0, 0.0) * pix_size) * 0.288713; + GLOW_ADD(vec2(0.0, 1.0), 0.233062); + GLOW_ADD(vec2(0.0, 2.0), 0.122581); + GLOW_ADD(vec2(0.0, -1.0), 0.233062); + GLOW_ADD(vec2(0.0, -2.0), 0.122581); + color *= blur.glow_strength; + frag_color = color; + } + +#undef GLOW_ADD + + if (bool(blur.flags & FLAG_GLOW_FIRST_PASS)) { +#ifdef GLOW_USE_AUTO_EXPOSURE + + frag_color /= texelFetch(source_auto_exposure, ivec2(0, 0), 0).r / blur.glow_auto_exposure_grey; +#endif + frag_color *= blur.glow_exposure; + + float luminance = max(frag_color.r, max(frag_color.g, frag_color.b)); + float feedback = max(smoothstep(blur.glow_hdr_threshold, blur.glow_hdr_threshold + blur.glow_hdr_scale, luminance), blur.glow_bloom); + + frag_color = min(frag_color * feedback, vec4(blur.glow_luminance_cap)); + } + +#endif + +#ifdef MODE_COPY + vec4 color = textureLod(source_color, uv_interp, 0.0); + frag_color = color; +#endif +} diff --git a/servers/rendering/renderer_rd/shaders/blur_raster_inc.glsl b/servers/rendering/renderer_rd/shaders/blur_raster_inc.glsl new file mode 100644 index 0000000000..52bf2886b5 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/blur_raster_inc.glsl @@ -0,0 +1,21 @@ +#define FLAG_HORIZONTAL (1 << 0) +#define FLAG_USE_ORTHOGONAL_PROJECTION (1 << 1) +#define FLAG_GLOW_FIRST_PASS (1 << 2) + +layout(push_constant, binding = 1, std430) uniform Blur { + vec2 pixel_size; + uint flags; + uint pad; + + // Glow. + float glow_strength; + float glow_bloom; + float glow_hdr_threshold; + float glow_hdr_scale; + + float glow_exposure; + float glow_white; + float glow_luminance_cap; + float glow_auto_exposure_grey; +} +blur; diff --git a/servers/rendering/renderer_rd/shaders/bokeh_dof.glsl b/servers/rendering/renderer_rd/shaders/bokeh_dof.glsl index b70e0b6bd5..0438671dd2 100644 --- a/servers/rendering/renderer_rd/shaders/bokeh_dof.glsl +++ b/servers/rendering/renderer_rd/shaders/bokeh_dof.glsl @@ -25,34 +25,7 @@ layout(set = 1, binding = 0) uniform sampler2D source_bokeh; // based on https://www.shadertoy.com/view/Xd3GDl -layout(push_constant, binding = 1, std430) uniform Params { - ivec2 size; - float z_far; - float z_near; - - bool orthogonal; - float blur_size; - float blur_scale; - int blur_steps; - - bool blur_near_active; - float blur_near_begin; - float blur_near_end; - bool blur_far_active; - - float blur_far_begin; - float blur_far_end; - bool second_pass; - bool half_size; - - bool use_jitter; - float jitter_seed; - uint pad[2]; -} -params; - -//used to work around downsampling filter -#define DEPTH_GAP 0.0 +#include "bokeh_dof_inc.glsl" #ifdef MODE_GEN_BLUR_SIZE @@ -80,15 +53,6 @@ float get_blur_size(float depth) { #endif -const float GOLDEN_ANGLE = 2.39996323; - -//note: uniform pdf rand [0;1[ -float hash12n(vec2 p) { - p = fract(p * vec2(5.3987, 5.4421)); - p += dot(p.yx, p.xy + vec2(21.5351, 14.3137)); - return fract(p.x * p.y * 95.4307); -} - #if defined(MODE_BOKEH_BOX) || defined(MODE_BOKEH_HEXAGONAL) vec4 weighted_filter_dir(vec2 dir, vec2 uv, vec2 pixel_size) { diff --git a/servers/rendering/renderer_rd/shaders/bokeh_dof_inc.glsl b/servers/rendering/renderer_rd/shaders/bokeh_dof_inc.glsl new file mode 100644 index 0000000000..fadea1631c --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/bokeh_dof_inc.glsl @@ -0,0 +1,37 @@ +layout(push_constant, binding = 1, std430) uniform Params { + ivec2 size; + float z_far; + float z_near; + + bool orthogonal; + float blur_size; + float blur_scale; + int blur_steps; + + bool blur_near_active; + float blur_near_begin; + float blur_near_end; + bool blur_far_active; + + float blur_far_begin; + float blur_far_end; + bool second_pass; + bool half_size; + + bool use_jitter; + float jitter_seed; + uint pad[2]; +} +params; + +//used to work around downsampling filter +#define DEPTH_GAP 0.0 + +const float GOLDEN_ANGLE = 2.39996323; + +//note: uniform pdf rand [0;1[ +float hash12n(vec2 p) { + p = fract(p * vec2(5.3987, 5.4421)); + p += dot(p.yx, p.xy + vec2(21.5351, 14.3137)); + return fract(p.x * p.y * 95.4307); +} diff --git a/servers/rendering/renderer_rd/shaders/bokeh_dof_raster.glsl b/servers/rendering/renderer_rd/shaders/bokeh_dof_raster.glsl new file mode 100644 index 0000000000..a3b3938ee9 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/bokeh_dof_raster.glsl @@ -0,0 +1,253 @@ +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "bokeh_dof_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + + gl_Position = vec4(uv_interp * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "bokeh_dof_inc.glsl" + +layout(location = 0) in vec2 uv_interp; +/* clang-format on */ + +#ifdef MODE_GEN_BLUR_SIZE +layout(location = 0) out float weight; + +layout(set = 0, binding = 0) uniform sampler2D source_depth; +#else +layout(location = 0) out vec4 frag_color; +#ifdef OUTPUT_WEIGHT +layout(location = 1) out float weight; +#endif + +layout(set = 0, binding = 0) uniform sampler2D source_color; +layout(set = 1, binding = 0) uniform sampler2D source_weight; +#ifdef MODE_COMPOSITE_BOKEH +layout(set = 2, binding = 0) uniform sampler2D original_weight; +#endif +#endif + +//DOF +// Bokeh single pass implementation based on https://tuxedolabs.blogspot.com/2018/05/bokeh-depth-of-field-in-single-pass.html + +#ifdef MODE_GEN_BLUR_SIZE + +float get_depth_at_pos(vec2 uv) { + float depth = textureLod(source_depth, uv, 0.0).x; + if (params.orthogonal) { + depth = ((depth + (params.z_far + params.z_near) / (params.z_far - params.z_near)) * (params.z_far - params.z_near)) / 2.0; + } else { + depth = 2.0 * params.z_near * params.z_far / (params.z_far + params.z_near - depth * (params.z_far - params.z_near)); + } + return depth; +} + +float get_blur_size(float depth) { + if (params.blur_near_active && depth < params.blur_near_begin) { + return -(1.0 - smoothstep(params.blur_near_end, params.blur_near_begin, depth)) * params.blur_size - DEPTH_GAP; //near blur is negative + } + + if (params.blur_far_active && depth > params.blur_far_begin) { + return smoothstep(params.blur_far_begin, params.blur_far_end, depth) * params.blur_size + DEPTH_GAP; + } + + return 0.0; +} + +#endif + +#if defined(MODE_BOKEH_BOX) || defined(MODE_BOKEH_HEXAGONAL) + +vec4 weighted_filter_dir(vec2 dir, vec2 uv, vec2 pixel_size) { + dir *= pixel_size; + vec4 color = texture(source_color, uv); + color.a = texture(source_weight, uv).r; + + vec4 accum = color; + float total = 1.0; + + float blur_scale = params.blur_size / float(params.blur_steps); + + if (params.use_jitter) { + uv += dir * (hash12n(uv + params.jitter_seed) - 0.5); + } + + for (int i = -params.blur_steps; i <= params.blur_steps; i++) { + if (i == 0) { + continue; + } + float radius = float(i) * blur_scale; + vec2 suv = uv + dir * radius; + radius = abs(radius); + + vec4 sample_color = texture(source_color, suv); + sample_color.a = texture(source_weight, suv).r; + float limit; + + if (sample_color.a < color.a) { + limit = abs(sample_color.a); + } else { + limit = abs(color.a); + } + + limit -= DEPTH_GAP; + + float m = smoothstep(radius - 0.5, radius + 0.5, limit); + + accum += mix(color, sample_color, m); + + total += 1.0; + } + + return accum / total; +} + +#endif + +void main() { + vec2 pixel_size = 1.0 / vec2(params.size); + vec2 uv = uv_interp; + +#ifdef MODE_GEN_BLUR_SIZE + uv += pixel_size * 0.5; + float center_depth = get_depth_at_pos(uv); + weight = get_blur_size(center_depth); +#endif + +#ifdef MODE_BOKEH_BOX + //pixel_size*=0.5; //resolution is doubled + if (params.second_pass || !params.half_size) { + uv += pixel_size * 0.5; //half pixel to read centers + } else { + uv += pixel_size * 0.25; //half pixel to read centers from full res + } + + float alpha = texture(source_color, uv).a; // retain this + vec2 dir = (params.second_pass ? vec2(0.0, 1.0) : vec2(1.0, 0.0)); + + vec4 color = weighted_filter_dir(dir, uv, pixel_size); + + frag_color = color; + frag_color.a = alpha; // attempt to retain this in case we have a transparent background, ignored if half_size +#ifdef OUTPUT_WEIGHT + weight = color.a; +#endif + +#endif + +#ifdef MODE_BOKEH_HEXAGONAL + + //pixel_size*=0.5; //resolution is doubled + if (params.second_pass || !params.half_size) { + uv += pixel_size * 0.5; //half pixel to read centers + } else { + uv += pixel_size * 0.25; //half pixel to read centers from full res + } + + float alpha = texture(source_color, uv).a; // retain this + + vec2 dir = (params.second_pass ? normalize(vec2(1.0, 0.577350269189626)) : vec2(0.0, 1.0)); + + vec4 color = weighted_filter_dir(dir, uv, pixel_size); + + if (params.second_pass) { + dir = normalize(vec2(-1.0, 0.577350269189626)); + + vec4 color2 = weighted_filter_dir(dir, uv, pixel_size); + + color.rgb = min(color.rgb, color2.rgb); + color.a = (color.a + color2.a) * 0.5; + } + + frag_color = color; + frag_color.a = alpha; // attempt to retain this in case we have a transparent background, ignored if half_size +#ifdef OUTPUT_WEIGHT + weight = color.a; +#endif + +#endif + +#ifdef MODE_BOKEH_CIRCULAR + if (params.half_size) { + pixel_size *= 0.5; //resolution is doubled + } + + uv += pixel_size * 0.5; //half pixel to read centers + + vec4 color = texture(source_color, uv); + float alpha = color.a; // retain this + color.a = texture(source_weight, uv).r; + + vec4 color_accum = color; + float accum = 1.0; + + float radius = params.blur_scale; + for (float ang = 0.0; radius < params.blur_size; ang += GOLDEN_ANGLE) { + vec2 uv_adj = uv + vec2(cos(ang), sin(ang)) * pixel_size * radius; + + vec4 sample_color = texture(source_color, uv_adj); + sample_color.a = texture(source_weight, uv_adj).r; + + float limit; + + if (sample_color.a < color.a) { + limit = abs(sample_color.a); + } else { + limit = abs(color.a); + } + + limit -= DEPTH_GAP; + + float m = smoothstep(radius - 0.5, radius + 0.5, limit); + color_accum += mix(color_accum / accum, sample_color, m); + accum += 1.0; + + radius += params.blur_scale / radius; + } + + color_accum = color_accum / accum; + + frag_color.rgb = color_accum.rgb; + frag_color.a = alpha; // attempt to retain this in case we have a transparent background, ignored if half_size +#ifdef OUTPUT_WEIGHT + weight = color_accum.a; +#endif + +#endif + +#ifdef MODE_COMPOSITE_BOKEH + frag_color.rgb = texture(source_color, uv).rgb; + + float center_weigth = texture(source_weight, uv).r; + float sample_weight = texture(original_weight, uv).r; + + float mix_amount; + if (sample_weight < center_weigth) { + mix_amount = min(1.0, max(0.0, max(abs(center_weigth), abs(sample_weight)) - DEPTH_GAP)); + } else { + mix_amount = min(1.0, max(0.0, abs(center_weigth) - DEPTH_GAP)); + } + + // let alpha blending take care of mixing + frag_color.a = mix_amount; +#endif +} diff --git a/servers/rendering/renderer_rd/shaders/canvas.glsl b/servers/rendering/renderer_rd/shaders/canvas.glsl index a443bcdcb8..2911e8b731 100644 --- a/servers/rendering/renderer_rd/shaders/canvas.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas.glsl @@ -458,6 +458,14 @@ void light_blend_compute(uint light_base, vec4 light_color, inout vec3 color) { #endif +float msdf_median(float r, float g, float b, float a) { + return min(max(min(r, g), min(max(r, g), b)), a); +} + +vec2 msdf_map(vec2 value, vec2 in_min, vec2 in_max, vec2 out_min, vec2 out_max) { + return out_min + (out_max - out_min) * (value - in_min) / (in_max - in_min); +} + void main() { vec4 color = color_interp; vec2 uv = uv_interp; @@ -485,7 +493,34 @@ void main() { #endif - color *= texture(sampler2D(color_texture, texture_sampler), uv); +#ifndef USE_PRIMITIVE + if (bool(draw_data.flags & FLAGS_USE_MSDF)) { + float px_range = draw_data.ninepatch_margins.x; + float outline_thickness = draw_data.ninepatch_margins.y; + //float reserved1 = draw_data.ninepatch_margins.z; + //float reserved2 = draw_data.ninepatch_margins.w; + + vec4 msdf_sample = texture(sampler2D(color_texture, texture_sampler), uv); + vec2 msdf_size = vec2(textureSize(sampler2D(color_texture, texture_sampler), 0)); + vec2 dest_size = vec2(1.0) / fwidth(uv); + float px_size = max(0.5 * dot((vec2(px_range) / msdf_size), dest_size), 1.0); + float d = msdf_median(msdf_sample.r, msdf_sample.g, msdf_sample.b, msdf_sample.a) - 0.5; + + if (outline_thickness > 0) { + float cr = clamp(outline_thickness, 0.0, px_range / 2) / px_range; + float a = clamp((d + cr) * px_size, 0.0, 1.0); + color.a = a * color.a; + } else { + float a = clamp(d * px_size + 0.5, 0.0, 1.0); + color.a = a * color.a; + } + + } else { +#else + { +#endif + color *= texture(sampler2D(color_texture, texture_sampler), uv); + } uint light_count = (draw_data.flags >> FLAGS_LIGHT_COUNT_SHIFT) & 0xF; //max 16 lights bool using_light = light_count > 0 || canvas_data.directional_light_count > 0; diff --git a/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl b/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl index 451f9b0089..0cff505cae 100644 --- a/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl @@ -24,6 +24,8 @@ #define FLAGS_DEFAULT_NORMAL_MAP_USED (1 << 26) #define FLAGS_DEFAULT_SPECULAR_MAP_USED (1 << 27) +#define FLAGS_USE_MSDF (1 << 28) + #define SAMPLER_NEAREST_CLAMP 0 #define SAMPLER_LINEAR_CLAMP 1 #define SAMPLER_NEAREST_WITH_MIPMAPS_CLAMP 2 diff --git a/servers/rendering/renderer_rd/shaders/cluster_render.glsl b/servers/rendering/renderer_rd/shaders/cluster_render.glsl index da7d189281..6d95722a57 100644 --- a/servers/rendering/renderer_rd/shaders/cluster_render.glsl +++ b/servers/rendering/renderer_rd/shaders/cluster_render.glsl @@ -117,7 +117,7 @@ void main() { uint cluster_thread_group_index; if (!gl_HelperInvocation) { - //http://advances.realtimerendering.com/s2017/2017_Sig_Improved_Culling_final.pdf + //https://advances.realtimerendering.com/s2017/2017_Sig_Improved_Culling_final.pdf uvec4 mask; diff --git a/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl b/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl index dfbce29119..69b895ed29 100644 --- a/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl +++ b/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl @@ -7,8 +7,7 @@ layout(push_constant, binding = 1, std430) uniform Params { float z_far; float z_near; - bool z_flip; - uint pad; + vec2 texel_size; vec4 screen_rect; } params; @@ -35,22 +34,23 @@ layout(set = 0, binding = 0) uniform samplerCube source_cube; layout(push_constant, binding = 1, std430) uniform Params { float z_far; float z_near; - bool z_flip; - uint pad; + vec2 texel_size; vec4 screen_rect; } params; void main() { vec2 uv = uv_interp; + vec2 texel_size = abs(params.texel_size); - vec3 normal = vec3(uv * 2.0 - 1.0, 0.0); + uv = clamp(uv * (1.0 + 2.0 * texel_size) - texel_size, vec2(0.0), vec2(1.0)); - normal.z = 0.5 - 0.5 * ((normal.x * normal.x) + (normal.y * normal.y)); + vec3 normal = vec3(uv * 2.0 - 1.0, 0.0); + normal.z = 0.5 * (1.0 - dot(normal.xy, normal.xy)); // z = 1/2 - 1/2 * (x^2 + y^2) normal = normalize(normal); normal.y = -normal.y; //needs to be flipped to match projection matrix - if (!params.z_flip) { + if (params.texel_size.x >= 0.0) { // Sign is used to encode Z flip normal.z = -normal.z; } diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl index 9fa84657d1..63f0ce690e 100644 --- a/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl @@ -32,53 +32,7 @@ layout(set = 0, binding = 0) uniform samplerCube source_cubemap; layout(rgba16f, set = 1, binding = 0) uniform restrict writeonly imageCube dest_cubemap; -layout(push_constant, binding = 1, std430) uniform Params { - uint face_size; -} -params; - -#define M_PI 3.14159265359 - -void get_dir_0(out vec3 dir, in float u, in float v) { - dir[0] = 1.0; - dir[1] = v; - dir[2] = -u; -} - -void get_dir_1(out vec3 dir, in float u, in float v) { - dir[0] = -1.0; - dir[1] = v; - dir[2] = u; -} - -void get_dir_2(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = 1.0; - dir[2] = -v; -} - -void get_dir_3(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = -1.0; - dir[2] = v; -} - -void get_dir_4(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = v; - dir[2] = 1.0; -} - -void get_dir_5(out vec3 dir, in float u, in float v) { - dir[0] = -u; - dir[1] = v; - dir[2] = -1.0; -} - -float calcWeight(float u, float v) { - float val = u * u + v * v + 1.0; - return val * sqrt(val); -} +#include "cubemap_downsampler_inc.glsl" void main() { uvec3 id = gl_GlobalInvocationID; diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl new file mode 100644 index 0000000000..b329e67293 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl @@ -0,0 +1,48 @@ +layout(push_constant, binding = 1, std430) uniform Params { + uint face_size; + uint face_id; // only used in raster shader +} +params; + +#define M_PI 3.14159265359 + +void get_dir_0(out vec3 dir, in float u, in float v) { + dir[0] = 1.0; + dir[1] = v; + dir[2] = -u; +} + +void get_dir_1(out vec3 dir, in float u, in float v) { + dir[0] = -1.0; + dir[1] = v; + dir[2] = u; +} + +void get_dir_2(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = 1.0; + dir[2] = -v; +} + +void get_dir_3(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = -1.0; + dir[2] = v; +} + +void get_dir_4(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = v; + dir[2] = 1.0; +} + +void get_dir_5(out vec3 dir, in float u, in float v) { + dir[0] = -u; + dir[1] = v; + dir[2] = -1.0; +} + +float calcWeight(float u, float v) { + float val = u * u + v * v + 1.0; + return val * sqrt(val); +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl new file mode 100644 index 0000000000..0828ffd921 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl @@ -0,0 +1,163 @@ +// Copyright 2016 Activision Publishing, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining +// a copy of this software and associated documentation files (the "Software"), +// to deal in the Software without restriction, including without limitation +// the rights to use, copy, modify, merge, publish, distribute, sublicense, +// and/or sell copies of the Software, and to permit persons to whom the Software +// is furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +// SOFTWARE. + +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_downsampler_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex] * float(params.face_size); + gl_Position = vec4(base_arr[gl_VertexIndex] * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_downsampler_inc.glsl" + +layout(set = 0, binding = 0) uniform samplerCube source_cubemap; + +layout(location = 0) in vec2 uv_interp; +layout(location = 0) out vec4 frag_color; +/* clang-format on */ + +void main() { + // Converted from compute shader which uses absolute coordinates. + // Could possibly simplify this + float face_size = float(params.face_size); + + if (uv_interp.x < face_size && uv_interp.y < face_size) { + float inv_face_size = 1.0 / face_size; + + float u0 = (uv_interp.x * 2.0 + 1.0 - 0.75) * inv_face_size - 1.0; + float u1 = (uv_interp.x * 2.0 + 1.0 + 0.75) * inv_face_size - 1.0; + + float v0 = (uv_interp.y * 2.0 + 1.0 - 0.75) * -inv_face_size + 1.0; + float v1 = (uv_interp.y * 2.0 + 1.0 + 0.75) * -inv_face_size + 1.0; + + float weights[4]; + weights[0] = calcWeight(u0, v0); + weights[1] = calcWeight(u1, v0); + weights[2] = calcWeight(u0, v1); + weights[3] = calcWeight(u1, v1); + + const float wsum = 0.5 / (weights[0] + weights[1] + weights[2] + weights[3]); + for (int i = 0; i < 4; i++) { + weights[i] = weights[i] * wsum + .125; + } + + vec3 dir; + vec4 color; + switch (params.face_id) { + case 0: + get_dir_0(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_0(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_0(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_0(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 1: + get_dir_1(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_1(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_1(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_1(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 2: + get_dir_2(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_2(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_2(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_2(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 3: + get_dir_3(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_3(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_3(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_3(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 4: + get_dir_4(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_4(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_4(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_4(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + default: + get_dir_5(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_5(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_5(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_5(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + } + frag_color = color; + } +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl new file mode 100644 index 0000000000..324d306218 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl @@ -0,0 +1,256 @@ +// Copyright 2016 Activision Publishing, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining +// a copy of this software and associated documentation files (the "Software"), +// to deal in the Software without restriction, including without limitation +// the rights to use, copy, modify, merge, publish, distribute, sublicense, +// and/or sell copies of the Software, and to permit persons to whom the Software +// is furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +// SOFTWARE. + +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +layout(push_constant, binding = 1, std430) uniform Params { + int mip_level; + uint face_id; +} +params; + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + gl_Position = vec4(base_arr[gl_VertexIndex] * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +layout(push_constant, binding = 1, std430) uniform Params { + int mip_level; + uint face_id; +} +params; + +layout(set = 0, binding = 0) uniform samplerCube source_cubemap; + +layout(location = 0) in vec2 uv_interp; +layout(location = 0) out vec4 frag_color; + +/* clang-format on */ + +#ifdef USE_HIGH_QUALITY +#define NUM_TAPS 32 +#else +#define NUM_TAPS 8 +#endif + +#define BASE_RESOLUTION 128 + +#ifdef USE_HIGH_QUALITY +layout(set = 1, binding = 0, std430) buffer restrict readonly Data { + vec4[7][5][3][24] coeffs; +} +data; +#else +layout(set = 1, binding = 0, std430) buffer restrict readonly Data { + vec4[7][5][6] coeffs; +} +data; +#endif + +void get_dir(out vec3 dir, in vec2 uv, in uint face) { + switch (face) { + case 0: + dir = vec3(1.0, uv[1], -uv[0]); + break; + case 1: + dir = vec3(-1.0, uv[1], uv[0]); + break; + case 2: + dir = vec3(uv[0], 1.0, -uv[1]); + break; + case 3: + dir = vec3(uv[0], -1.0, uv[1]); + break; + case 4: + dir = vec3(uv[0], uv[1], 1.0); + break; + default: + dir = vec3(-uv[0], uv[1], -1.0); + break; + } +} + +void main() { + // determine dir / pos for the texel + vec3 dir, adir, frameZ; + { + vec2 uv; + uv.x = uv_interp.x; + uv.y = 1.0 - uv_interp.y; + uv = uv * 2.0 - 1.0; + + get_dir(dir, uv, params.face_id); + frameZ = normalize(dir); + + adir = abs(dir); + } + + // determine which texel this is + // NOTE (macOS/MoltenVK): Do not rename, "level" variable name conflicts with the Metal "level(float lod)" mipmap sampling function name. + int mip_level = 0; + + if (params.mip_level < 0) { + // return as is + frag_color.rgb = textureLod(source_cubemap, frameZ, 0.0).rgb; + frag_color.a = 1.0; + return; + } else if (params.mip_level > 6) { + // maximum level + mip_level = 6; + } else { + mip_level = params.mip_level; + } + + // GGX gather colors + vec4 color = vec4(0.0); + for (int axis = 0; axis < 3; axis++) { + const int otherAxis0 = 1 - (axis & 1) - (axis >> 1); + const int otherAxis1 = 2 - (axis >> 1); + + float frameweight = (max(adir[otherAxis0], adir[otherAxis1]) - .75) / .25; + if (frameweight > 0.0) { + // determine frame + vec3 UpVector; + switch (axis) { + case 0: + UpVector = vec3(1, 0, 0); + break; + case 1: + UpVector = vec3(0, 1, 0); + break; + default: + UpVector = vec3(0, 0, 1); + break; + } + + vec3 frameX = normalize(cross(UpVector, frameZ)); + vec3 frameY = cross(frameZ, frameX); + + // calculate parametrization for polynomial + float Nx = dir[otherAxis0]; + float Ny = dir[otherAxis1]; + float Nz = adir[axis]; + + float NmaxXY = max(abs(Ny), abs(Nx)); + Nx /= NmaxXY; + Ny /= NmaxXY; + + float theta; + if (Ny < Nx) { + if (Ny <= -0.999) + theta = Nx; + else + theta = Ny; + } else { + if (Ny >= 0.999) + theta = -Nx; + else + theta = -Ny; + } + + float phi; + if (Nz <= -0.999) + phi = -NmaxXY; + else if (Nz >= 0.999) + phi = NmaxXY; + else + phi = Nz; + + float theta2 = theta * theta; + float phi2 = phi * phi; + + // sample + for (int iSuperTap = 0; iSuperTap < NUM_TAPS / 4; iSuperTap++) { + const int index = (NUM_TAPS / 4) * axis + iSuperTap; + +#ifdef USE_HIGH_QUALITY + vec4 coeffsDir0[3]; + vec4 coeffsDir1[3]; + vec4 coeffsDir2[3]; + vec4 coeffsLevel[3]; + vec4 coeffsWeight[3]; + + for (int iCoeff = 0; iCoeff < 3; iCoeff++) { + coeffsDir0[iCoeff] = data.coeffs[mip_level][0][iCoeff][index]; + coeffsDir1[iCoeff] = data.coeffs[mip_level][1][iCoeff][index]; + coeffsDir2[iCoeff] = data.coeffs[mip_level][2][iCoeff][index]; + coeffsLevel[iCoeff] = data.coeffs[mip_level][3][iCoeff][index]; + coeffsWeight[iCoeff] = data.coeffs[mip_level][4][iCoeff][index]; + } + + for (int iSubTap = 0; iSubTap < 4; iSubTap++) { + // determine sample attributes (dir, weight, mip_level) + vec3 sample_dir = frameX * (coeffsDir0[0][iSubTap] + coeffsDir0[1][iSubTap] * theta2 + coeffsDir0[2][iSubTap] * phi2) + frameY * (coeffsDir1[0][iSubTap] + coeffsDir1[1][iSubTap] * theta2 + coeffsDir1[2][iSubTap] * phi2) + frameZ * (coeffsDir2[0][iSubTap] + coeffsDir2[1][iSubTap] * theta2 + coeffsDir2[2][iSubTap] * phi2); + + float sample_level = coeffsLevel[0][iSubTap] + coeffsLevel[1][iSubTap] * theta2 + coeffsLevel[2][iSubTap] * phi2; + + float sample_weight = coeffsWeight[0][iSubTap] + coeffsWeight[1][iSubTap] * theta2 + coeffsWeight[2][iSubTap] * phi2; +#else + vec4 coeffsDir0 = data.coeffs[mip_level][0][index]; + vec4 coeffsDir1 = data.coeffs[mip_level][1][index]; + vec4 coeffsDir2 = data.coeffs[mip_level][2][index]; + vec4 coeffsLevel = data.coeffs[mip_level][3][index]; + vec4 coeffsWeight = data.coeffs[mip_level][4][index]; + + for (int iSubTap = 0; iSubTap < 4; iSubTap++) { + // determine sample attributes (dir, weight, mip_level) + vec3 sample_dir = frameX * coeffsDir0[iSubTap] + frameY * coeffsDir1[iSubTap] + frameZ * coeffsDir2[iSubTap]; + + float sample_level = coeffsLevel[iSubTap]; + + float sample_weight = coeffsWeight[iSubTap]; +#endif + + sample_weight *= frameweight; + + // adjust for jacobian + sample_dir /= max(abs(sample_dir[0]), max(abs(sample_dir[1]), abs(sample_dir[2]))); + sample_level += 0.75 * log2(dot(sample_dir, sample_dir)); + // sample cubemap + color.xyz += textureLod(source_cubemap, normalize(sample_dir), sample_level).xyz * sample_weight; + color.w += sample_weight; + } + } + } + } + color /= color.w; + + // write color + color.xyz = max(vec3(0.0), color.xyz); + color.w = 1.0; + + frag_color = color; +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl index ce7c03c1d4..28f4dc59ec 100644 --- a/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl @@ -12,100 +12,7 @@ layout(set = 0, binding = 0) uniform samplerCube source_cube; layout(rgba16f, set = 1, binding = 0) uniform restrict writeonly imageCube dest_cubemap; -layout(push_constant, binding = 1, std430) uniform Params { - uint face_id; - uint sample_count; - float roughness; - bool use_direct_write; - float face_size; -} -params; - -#define M_PI 3.14159265359 - -vec3 texelCoordToVec(vec2 uv, uint faceID) { - mat3 faceUvVectors[6]; - - // -x - faceUvVectors[1][0] = vec3(0.0, 0.0, 1.0); // u -> +z - faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[1][2] = vec3(-1.0, 0.0, 0.0); // -x face - - // +x - faceUvVectors[0][0] = vec3(0.0, 0.0, -1.0); // u -> -z - faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[0][2] = vec3(1.0, 0.0, 0.0); // +x face - - // -y - faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[3][1] = vec3(0.0, 0.0, -1.0); // v -> -z - faceUvVectors[3][2] = vec3(0.0, -1.0, 0.0); // -y face - - // +y - faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[2][1] = vec3(0.0, 0.0, 1.0); // v -> +z - faceUvVectors[2][2] = vec3(0.0, 1.0, 0.0); // +y face - - // -z - faceUvVectors[5][0] = vec3(-1.0, 0.0, 0.0); // u -> -x - faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[5][2] = vec3(0.0, 0.0, -1.0); // -z face - - // +z - faceUvVectors[4][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[4][2] = vec3(0.0, 0.0, 1.0); // +z face - - // out = u * s_faceUv[0] + v * s_faceUv[1] + s_faceUv[2]. - vec3 result = (faceUvVectors[faceID][0] * uv.x) + (faceUvVectors[faceID][1] * uv.y) + faceUvVectors[faceID][2]; - return normalize(result); -} - -vec3 ImportanceSampleGGX(vec2 Xi, float Roughness, vec3 N) { - float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph] - - // Compute distribution direction - float Phi = 2.0 * M_PI * Xi.x; - float CosTheta = sqrt((1.0 - Xi.y) / (1.0 + (a * a - 1.0) * Xi.y)); - float SinTheta = sqrt(1.0 - CosTheta * CosTheta); - - // Convert to spherical direction - vec3 H; - H.x = SinTheta * cos(Phi); - H.y = SinTheta * sin(Phi); - H.z = CosTheta; - - vec3 UpVector = abs(N.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(1.0, 0.0, 0.0); - vec3 TangentX = normalize(cross(UpVector, N)); - vec3 TangentY = cross(N, TangentX); - - // Tangent to world space - return TangentX * H.x + TangentY * H.y + N * H.z; -} - -// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html -float GGX(float NdotV, float a) { - float k = a / 2.0; - return NdotV / (NdotV * (1.0 - k) + k); -} - -// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html -float G_Smith(float a, float nDotV, float nDotL) { - return GGX(nDotL, a * a) * GGX(nDotV, a * a); -} - -float radicalInverse_VdC(uint bits) { - bits = (bits << 16u) | (bits >> 16u); - bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u); - bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u); - bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u); - bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u); - return float(bits) * 2.3283064365386963e-10; // / 0x100000000 -} - -vec2 Hammersley(uint i, uint N) { - return vec2(float(i) / float(N), radicalInverse_VdC(i)); -} +#include "cubemap_roughness_inc.glsl" void main() { uvec3 id = gl_GlobalInvocationID; diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl new file mode 100644 index 0000000000..be12be5dec --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl @@ -0,0 +1,94 @@ +#define M_PI 3.14159265359 + +layout(push_constant, binding = 1, std430) uniform Params { + uint face_id; + uint sample_count; + float roughness; + bool use_direct_write; + float face_size; +} +params; + +vec3 texelCoordToVec(vec2 uv, uint faceID) { + mat3 faceUvVectors[6]; + + // -x + faceUvVectors[1][0] = vec3(0.0, 0.0, 1.0); // u -> +z + faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[1][2] = vec3(-1.0, 0.0, 0.0); // -x face + + // +x + faceUvVectors[0][0] = vec3(0.0, 0.0, -1.0); // u -> -z + faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[0][2] = vec3(1.0, 0.0, 0.0); // +x face + + // -y + faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[3][1] = vec3(0.0, 0.0, -1.0); // v -> -z + faceUvVectors[3][2] = vec3(0.0, -1.0, 0.0); // -y face + + // +y + faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[2][1] = vec3(0.0, 0.0, 1.0); // v -> +z + faceUvVectors[2][2] = vec3(0.0, 1.0, 0.0); // +y face + + // -z + faceUvVectors[5][0] = vec3(-1.0, 0.0, 0.0); // u -> -x + faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[5][2] = vec3(0.0, 0.0, -1.0); // -z face + + // +z + faceUvVectors[4][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[4][2] = vec3(0.0, 0.0, 1.0); // +z face + + // out = u * s_faceUv[0] + v * s_faceUv[1] + s_faceUv[2]. + vec3 result = (faceUvVectors[faceID][0] * uv.x) + (faceUvVectors[faceID][1] * uv.y) + faceUvVectors[faceID][2]; + return normalize(result); +} + +vec3 ImportanceSampleGGX(vec2 Xi, float Roughness, vec3 N) { + float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph] + + // Compute distribution direction + float Phi = 2.0 * M_PI * Xi.x; + float CosTheta = sqrt((1.0 - Xi.y) / (1.0 + (a * a - 1.0) * Xi.y)); + float SinTheta = sqrt(1.0 - CosTheta * CosTheta); + + // Convert to spherical direction + vec3 H; + H.x = SinTheta * cos(Phi); + H.y = SinTheta * sin(Phi); + H.z = CosTheta; + + vec3 UpVector = abs(N.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(1.0, 0.0, 0.0); + vec3 TangentX = normalize(cross(UpVector, N)); + vec3 TangentY = cross(N, TangentX); + + // Tangent to world space + return TangentX * H.x + TangentY * H.y + N * H.z; +} + +// https://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float GGX(float NdotV, float a) { + float k = a / 2.0; + return NdotV / (NdotV * (1.0 - k) + k); +} + +// https://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float G_Smith(float a, float nDotV, float nDotL) { + return GGX(nDotL, a * a) * GGX(nDotV, a * a); +} + +float radicalInverse_VdC(uint bits) { + bits = (bits << 16u) | (bits >> 16u); + bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u); + bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u); + bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u); + bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u); + return float(bits) * 2.3283064365386963e-10; // / 0x100000000 +} + +vec2 Hammersley(uint i, uint N) { + return vec2(float(i) / float(N), radicalInverse_VdC(i)); +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl new file mode 100644 index 0000000000..2570308816 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl @@ -0,0 +1,63 @@ +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_roughness_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + gl_Position = vec4(uv_interp * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_roughness_inc.glsl" + +layout(location = 0) in vec2 uv_interp; + +layout(set = 0, binding = 0) uniform samplerCube source_cube; + +layout(location = 0) out vec4 frag_color; +/* clang-format on */ + +void main() { + vec3 N = texelCoordToVec(uv_interp * 2.0 - 1.0, params.face_id); + + //vec4 color = color_interp; + + if (params.use_direct_write) { + frag_color = vec4(texture(source_cube, N).rgb, 1.0); + } else { + vec4 sum = vec4(0.0, 0.0, 0.0, 0.0); + + for (uint sampleNum = 0u; sampleNum < params.sample_count; sampleNum++) { + vec2 xi = Hammersley(sampleNum, params.sample_count); + + vec3 H = ImportanceSampleGGX(xi, params.roughness, N); + vec3 V = N; + vec3 L = (2.0 * dot(V, H) * H - V); + + float ndotl = clamp(dot(N, L), 0.0, 1.0); + + if (ndotl > 0.0) { + sum.rgb += textureLod(source_cube, L, 0.0).rgb * ndotl; + sum.a += ndotl; + } + } + sum /= sum.a; + + frag_color = vec4(sum.rgb, 1.0); + } +} diff --git a/servers/rendering/renderer_rd/shaders/decal_data_inc.glsl b/servers/rendering/renderer_rd/shaders/decal_data_inc.glsl index ccaad13311..158096d3c7 100644 --- a/servers/rendering/renderer_rd/shaders/decal_data_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/decal_data_inc.glsl @@ -1,18 +1,18 @@ struct DecalData { - mat4 xform; //to decal transform - vec3 inv_extents; - float albedo_mix; - vec4 albedo_rect; - vec4 normal_rect; - vec4 orm_rect; - vec4 emission_rect; - vec4 modulate; - float emission_energy; + highp mat4 xform; //to decal transform + highp vec3 inv_extents; + mediump float albedo_mix; + highp vec4 albedo_rect; + highp vec4 normal_rect; + highp vec4 orm_rect; + highp vec4 emission_rect; + highp vec4 modulate; + mediump float emission_energy; uint mask; - float upper_fade; - float lower_fade; - mat3x4 normal_xform; - vec3 normal; - float normal_fade; + mediump float upper_fade; + mediump float lower_fade; + mediump mat3x4 normal_xform; + mediump vec3 normal; + mediump float normal_fade; }; diff --git a/servers/rendering/renderer_rd/shaders/gi.glsl b/servers/rendering/renderer_rd/shaders/gi.glsl index 60c881881d..5528ea3659 100644 --- a/servers/rendering/renderer_rd/shaders/gi.glsl +++ b/servers/rendering/renderer_rd/shaders/gi.glsl @@ -97,12 +97,10 @@ layout(push_constant, binding = 0, std430) uniform Params { vec4 proj_info; - vec3 ao_color; uint max_voxel_gi_instances; - bool high_quality_vct; bool orthogonal; - uint pad[2]; + uint pad; mat3x4 cam_rotation; } diff --git a/servers/rendering/renderer_rd/shaders/giprobe_write.glsl b/servers/rendering/renderer_rd/shaders/giprobe_write.glsl index 5dc2d08a3b..25d87ca45d 100644 --- a/servers/rendering/renderer_rd/shaders/giprobe_write.glsl +++ b/servers/rendering/renderer_rd/shaders/giprobe_write.glsl @@ -202,12 +202,7 @@ void main() { vec3 emission = vec3(ivec3(cell_data.data[cell_index].emission & 0x3FF, (cell_data.data[cell_index].emission >> 10) & 0x7FF, cell_data.data[cell_index].emission >> 21)) * params.emission_scale; vec4 normal = unpackSnorm4x8(cell_data.data[cell_index].normal); -#ifdef MODE_ANISOTROPIC - vec3 accum[6] = vec3[](vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0)); - const vec3 accum_dirs[6] = vec3[](vec3(1.0, 0.0, 0.0), vec3(-1.0, 0.0, 0.0), vec3(0.0, 1.0, 0.0), vec3(0.0, -1.0, 0.0), vec3(0.0, 0.0, 1.0), vec3(0.0, 0.0, -1.0)); -#else vec3 accum = vec3(0.0); -#endif for (uint i = 0; i < params.light_count; i++) { float attenuation; @@ -242,77 +237,35 @@ void main() { vec3 light = lights.data[i].color * albedo.rgb * attenuation * lights.data[i].energy; -#ifdef MODE_ANISOTROPIC - for (uint j = 0; j < 6; j++) { - accum[j] += max(0.0, dot(accum_dir, -light_dir)) * light + emission; - } -#else if (length(normal.xyz) > 0.2) { accum += max(0.0, dot(normal.xyz, -light_dir)) * light + emission; } else { //all directions accum += light + emission; } -#endif } -#ifdef MODE_ANISOTROPIC - - output.data[cell_index * 6 + 0] = vec4(accum[0], 0.0); - output.data[cell_index * 6 + 1] = vec4(accum[1], 0.0); - output.data[cell_index * 6 + 2] = vec4(accum[2], 0.0); - output.data[cell_index * 6 + 3] = vec4(accum[3], 0.0); - output.data[cell_index * 6 + 4] = vec4(accum[4], 0.0); - output.data[cell_index * 6 + 5] = vec4(accum[5], 0.0); -#else output.data[cell_index] = vec4(accum, 0.0); -#endif - #endif //MODE_COMPUTE_LIGHT #ifdef MODE_UPDATE_MIPMAPS { -#ifdef MODE_ANISOTROPIC - vec3 light_accum[6] = vec3[](vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0)); -#else vec3 light_accum = vec3(0.0); -#endif float count = 0.0; for (uint i = 0; i < 8; i++) { uint child_index = cell_children.data[cell_index].children[i]; if (child_index == NO_CHILDREN) { continue; } -#ifdef MODE_ANISOTROPIC - light_accum[1] += output.data[child_index * 6 + 0].rgb; - light_accum[2] += output.data[child_index * 6 + 1].rgb; - light_accum[3] += output.data[child_index * 6 + 2].rgb; - light_accum[4] += output.data[child_index * 6 + 3].rgb; - light_accum[5] += output.data[child_index * 6 + 4].rgb; - light_accum[6] += output.data[child_index * 6 + 5].rgb; - -#else light_accum += output.data[child_index].rgb; -#endif - count += 1.0; } float divisor = mix(8.0, count, params.propagation); -#ifdef MODE_ANISOTROPIC - output.data[cell_index * 6 + 0] = vec4(light_accum[0] / divisor, 0.0); - output.data[cell_index * 6 + 1] = vec4(light_accum[1] / divisor, 0.0); - output.data[cell_index * 6 + 2] = vec4(light_accum[2] / divisor, 0.0); - output.data[cell_index * 6 + 3] = vec4(light_accum[3] / divisor, 0.0); - output.data[cell_index * 6 + 4] = vec4(light_accum[4] / divisor, 0.0); - output.data[cell_index * 6 + 5] = vec4(light_accum[5] / divisor, 0.0); - -#else output.data[cell_index] = vec4(light_accum / divisor, 0.0); -#endif } #endif diff --git a/servers/rendering/renderer_rd/shaders/light_data_inc.glsl b/servers/rendering/renderer_rd/shaders/light_data_inc.glsl index 2fce258cff..fdc7729338 100644 --- a/servers/rendering/renderer_rd/shaders/light_data_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/light_data_inc.glsl @@ -3,31 +3,31 @@ #define LIGHT_BAKE_STATIC 2 struct LightData { //this structure needs to be as packed as possible - vec3 position; - float inv_radius; + highp vec3 position; + highp float inv_radius; - vec3 direction; - float size; + mediump vec3 direction; + highp float size; - vec3 color; - float attenuation; + mediump vec3 color; + mediump float attenuation; - float cone_attenuation; - float cone_angle; - float specular_amount; + mediump float cone_attenuation; + mediump float cone_angle; + mediump float specular_amount; bool shadow_enabled; - vec4 atlas_rect; // rect in the shadow atlas - mat4 shadow_matrix; - float shadow_bias; - float shadow_normal_bias; - float transmittance_bias; - float soft_shadow_size; // for spot, it's the size in uv coordinates of the light, for omni it's the span angle - float soft_shadow_scale; // scales the shadow kernel for blurrier shadows + highp vec4 atlas_rect; // rect in the shadow atlas + highp mat4 shadow_matrix; + highp float shadow_bias; + highp float shadow_normal_bias; + highp float transmittance_bias; + highp float soft_shadow_size; // for spot, it's the size in uv coordinates of the light, for omni it's the span angle + highp float soft_shadow_scale; // scales the shadow kernel for blurrier shadows uint mask; - float shadow_volumetric_fog_fade; + mediump float shadow_volumetric_fog_fade; uint bake_mode; - vec4 projector_rect; //projector rect in srgb decal atlas + highp vec4 projector_rect; //projector rect in srgb decal atlas }; #define REFLECTION_AMBIENT_DISABLED 0 @@ -35,53 +35,53 @@ struct LightData { //this structure needs to be as packed as possible #define REFLECTION_AMBIENT_COLOR 2 struct ReflectionData { - vec3 box_extents; - float index; - vec3 box_offset; + highp vec3 box_extents; + mediump float index; + highp vec3 box_offset; uint mask; - vec3 ambient; // ambient color - float intensity; + mediump vec3 ambient; // ambient color + mediump float intensity; bool exterior; bool box_project; uint ambient_mode; uint pad; //0-8 is intensity,8-9 is ambient, mode - mat4 local_matrix; // up to here for spot and omni, rest is for directional + highp mat4 local_matrix; // up to here for spot and omni, rest is for directional // notes: for ambientblend, use distance to edge to blend between already existing global environment }; struct DirectionalLightData { - vec3 direction; - float energy; - vec3 color; - float size; - float specular; + mediump vec3 direction; + mediump float energy; + mediump vec3 color; + mediump float size; + mediump float specular; uint mask; - float softshadow_angle; - float soft_shadow_scale; + highp float softshadow_angle; + highp float soft_shadow_scale; bool blend_splits; bool shadow_enabled; - float fade_from; - float fade_to; + highp float fade_from; + highp float fade_to; uvec2 pad; uint bake_mode; - float shadow_volumetric_fog_fade; - vec4 shadow_bias; - vec4 shadow_normal_bias; - vec4 shadow_transmittance_bias; - vec4 shadow_z_range; - vec4 shadow_range_begin; - vec4 shadow_split_offsets; - mat4 shadow_matrix1; - mat4 shadow_matrix2; - mat4 shadow_matrix3; - mat4 shadow_matrix4; - vec4 shadow_color1; - vec4 shadow_color2; - vec4 shadow_color3; - vec4 shadow_color4; - vec2 uv_scale1; - vec2 uv_scale2; - vec2 uv_scale3; - vec2 uv_scale4; + mediump float shadow_volumetric_fog_fade; + highp vec4 shadow_bias; + highp vec4 shadow_normal_bias; + highp vec4 shadow_transmittance_bias; + highp vec4 shadow_z_range; + highp vec4 shadow_range_begin; + highp vec4 shadow_split_offsets; + highp mat4 shadow_matrix1; + highp mat4 shadow_matrix2; + highp mat4 shadow_matrix3; + highp mat4 shadow_matrix4; + mediump vec4 shadow_color1; + mediump vec4 shadow_color2; + mediump vec4 shadow_color3; + mediump vec4 shadow_color4; + highp vec2 uv_scale1; + highp vec2 uv_scale2; + highp vec2 uv_scale3; + highp vec2 uv_scale4; }; diff --git a/servers/rendering/renderer_rd/shaders/luminance_reduce_raster.glsl b/servers/rendering/renderer_rd/shaders/luminance_reduce_raster.glsl new file mode 100644 index 0000000000..29ebd74a90 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/luminance_reduce_raster.glsl @@ -0,0 +1,74 @@ +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "luminance_reduce_raster_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + + gl_Position = vec4(uv_interp * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "luminance_reduce_raster_inc.glsl" + +layout(location = 0) in vec2 uv_interp; +/* clang-format on */ + +layout(set = 0, binding = 0) uniform sampler2D source_exposure; + +#ifdef FINAL_PASS +layout(set = 1, binding = 0) uniform sampler2D prev_luminance; +#endif + +layout(location = 0) out highp float luminance; + +void main() { + ivec2 dest_pos = ivec2(uv_interp * settings.dest_size); + ivec2 src_pos = ivec2(uv_interp * settings.source_size); + + ivec2 next_pos = (dest_pos + ivec2(1)) * settings.source_size / settings.dest_size; + next_pos = max(next_pos, src_pos + ivec2(1)); //so it at least reads one pixel + + highp vec3 source_color = vec3(0.0); + for (int i = src_pos.x; i < next_pos.x; i++) { + for (int j = src_pos.y; j < next_pos.y; j++) { + source_color += texelFetch(source_exposure, ivec2(i, j), 0).rgb; + } + } + + source_color /= float((next_pos.x - src_pos.x) * (next_pos.y - src_pos.y)); + +#ifdef FIRST_PASS + luminance = max(source_color.r, max(source_color.g, source_color.b)); + + // This formula should be more "accurate" but gave an overexposed result when testing. + // Leaving it here so we can revisit it if we want. + // luminance = source_color.r * 0.21 + source_color.g * 0.71 + source_color.b * 0.07; +#else + luminance = source_color.r; +#endif + +#ifdef FINAL_PASS + // Obtain our target luminance + luminance = clamp(luminance, settings.min_luminance, settings.max_luminance); + + // Now smooth to our transition + highp float prev_lum = texelFetch(prev_luminance, ivec2(0, 0), 0).r; //1 pixel previous luminance + luminance = prev_lum + (luminance - prev_lum) * clamp(settings.exposure_adjust, 0.0, 1.0); +#endif +} diff --git a/servers/rendering/renderer_rd/shaders/luminance_reduce_raster_inc.glsl b/servers/rendering/renderer_rd/shaders/luminance_reduce_raster_inc.glsl new file mode 100644 index 0000000000..3cde9923fa --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/luminance_reduce_raster_inc.glsl @@ -0,0 +1,11 @@ + +layout(push_constant, binding = 1, std430) uniform PushConstant { + ivec2 source_size; + ivec2 dest_size; + + float exposure_adjust; + float min_luminance; + float max_luminance; + uint pad1; +} +settings; diff --git a/servers/rendering/renderer_rd/shaders/particles_copy.glsl b/servers/rendering/renderer_rd/shaders/particles_copy.glsl index 4dceeea995..e88e68b511 100644 --- a/servers/rendering/renderer_rd/shaders/particles_copy.glsl +++ b/servers/rendering/renderer_rd/shaders/particles_copy.glsl @@ -138,7 +138,7 @@ void main() { if (bool(particles.data[particle].flags & PARTICLE_FLAG_ACTIVE) || bool(particles.data[particle].flags & PARTICLE_FLAG_TRAILED)) { txform = particles.data[particle].xform; if (params.trail_size > 1) { - // since the steps dont fit precisely in the history frames, must do a tiny bit of + // Since the steps don't fit precisely in the history frames, must do a tiny bit of // interpolation to get them close to their intended location. uint part_ofs = particle % params.trail_size; float natural_ofs = fract((float(part_ofs) / float(params.trail_size)) * float(params.trail_total)) * params.frame_delta; diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl index ef1c9bacfb..987960069b 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl @@ -118,7 +118,7 @@ void main() { mat3 world_normal_matrix; if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_NON_UNIFORM_SCALE)) { - world_normal_matrix = inverse(mat3(world_matrix)); + world_normal_matrix = transpose(inverse(mat3(world_matrix))); } else { world_normal_matrix = mat3(world_matrix); } @@ -374,6 +374,9 @@ layout(constant_id = 9) const uint sc_directional_penumbra_shadow_samples = 4; layout(constant_id = 10) const bool sc_decal_use_mipmaps = true; layout(constant_id = 11) const bool sc_projector_use_mipmaps = true; +// not used in clustered renderer but we share some code with the mobile renderer that requires this. +const float sc_luminance_multiplier = 1.0; + #include "scene_forward_clustered_inc.glsl" /* Varyings */ @@ -466,6 +469,11 @@ layout(location = 0) out vec4 frag_color; #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) +/* Make a default specular mode SPECULAR_SCHLICK_GGX. */ +#if !defined(SPECULAR_DISABLED) && !defined(SPECULAR_SCHLICK_GGX) && !defined(SPECULAR_BLINN) && !defined(SPECULAR_PHONG) && !defined(SPECULAR_TOON) +#define SPECULAR_SCHLICK_GGX +#endif + #include "scene_forward_lights_inc.glsl" #include "scene_forward_gi_inc.glsl" @@ -516,14 +524,14 @@ vec4 fog_process(vec3 vertex) { } } - float fog_amount = 1.0 - exp(min(0.0, vertex.z * scene_data.fog_density)); + float fog_amount = 1.0 - exp(min(0.0, -length(vertex) * scene_data.fog_density)); - if (abs(scene_data.fog_height_density) > 0.001) { + if (abs(scene_data.fog_height_density) >= 0.0001) { float y = (scene_data.camera_matrix * vec4(vertex, 1.0)).y; - float y_dist = scene_data.fog_height - y; + float y_dist = y - scene_data.fog_height; - float vfog_amount = clamp(exp(y_dist * scene_data.fog_height_density), 0.0, 1.0); + float vfog_amount = 1.0 - exp(min(0.0, y_dist * scene_data.fog_height_density)); fog_amount = max(vfog_amount, fog_amount); } @@ -876,7 +884,7 @@ void main() { #ifdef NORMAL_USED if (scene_data.roughness_limiter_enabled) { - //http://www.jp.square-enix.com/tech/library/pdf/ImprovedGeometricSpecularAA.pdf + //https://www.jp.square-enix.com/tech/library/pdf/ImprovedGeometricSpecularAA.pdf float roughness2 = roughness * roughness; vec3 dndu = dFdx(normal), dndv = dFdy(normal); float variance = scene_data.roughness_limiter_amount * (dot(dndu, dndu) + dot(dndv, dndv)); @@ -895,6 +903,7 @@ void main() { if (scene_data.use_reflection_cubemap) { vec3 ref_vec = reflect(-view, normal); + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); ref_vec = scene_data.radiance_inverse_xform * ref_vec; #ifdef USE_RADIANCE_CUBEMAP_ARRAY @@ -907,6 +916,7 @@ void main() { specular_light = textureLod(samplerCube(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), ref_vec, roughness * MAX_ROUGHNESS_LOD).rgb; #endif //USE_RADIANCE_CUBEMAP_ARRAY + specular_light *= horizon * horizon; specular_light *= scene_data.ambient_light_color_energy.a; } @@ -1214,7 +1224,7 @@ void main() { specular_light *= specular * metallic * albedo * 2.0; #else - // scales the specular reflections, needs to be be computed before lighting happens, + // scales the specular reflections, needs to be computed before lighting happens, // but after environment, GI, and reflection probes are added // Environment brdf approximation (Lazarov 2013) // see https://www.unrealengine.com/en-US/blog/physically-based-shading-on-mobile @@ -1260,21 +1270,21 @@ void main() { float shadow = 1.0; - //version with soft shadows, more expensive if (directional_lights.data[i].shadow_enabled) { - if (sc_use_directional_soft_shadows && directional_lights.data[i].softshadow_angle > 0) { - float depth_z = -vertex.z; - - vec3 shadow_color = vec3(0.0); - vec3 light_dir = directional_lights.data[i].direction; + float depth_z = -vertex.z; + vec3 light_dir = directional_lights.data[i].direction; + vec3 base_normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))); -#define BIAS_FUNC(m_var, m_idx) \ - m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ - vec3 normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))) * directional_lights.data[i].shadow_normal_bias[m_idx]; \ - normal_bias -= light_dir * dot(light_dir, normal_bias); \ +#define BIAS_FUNC(m_var, m_idx) \ + m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ + vec3 normal_bias = base_normal_bias * directional_lights.data[i].shadow_normal_bias[m_idx]; \ + normal_bias -= light_dir * dot(light_dir, normal_bias); \ m_var.xyz += normal_bias; - uint blend_index = 0; + //version with soft shadows, more expensive + if (sc_use_directional_soft_shadows && directional_lights.data[i].softshadow_angle > 0) { + uint blend_count = 0; + const uint blend_max = directional_lights.data[i].blend_splits ? 2 : 1; if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { vec4 v = vec4(vertex, 1.0); @@ -1289,10 +1299,10 @@ void main() { float test_radius = (range_pos - range_begin) * directional_lights.data[i].softshadow_angle; vec2 tex_scale = directional_lights.data[i].uv_scale1 * test_radius; shadow = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - blend_index++; + blend_count++; } - if (blend_index < 2 && depth_z < directional_lights.data[i].shadow_split_offsets.y) { + if (blend_count < blend_max && depth_z < directional_lights.data[i].shadow_split_offsets.y) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 1) @@ -1306,7 +1316,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale2 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1314,10 +1324,10 @@ void main() { shadow = mix(shadow, s, blend); } - blend_index++; + blend_count++; } - if (blend_index < 2 && depth_z < directional_lights.data[i].shadow_split_offsets.z) { + if (blend_count < blend_max && depth_z < directional_lights.data[i].shadow_split_offsets.z) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 2) @@ -1331,7 +1341,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale3 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1339,10 +1349,10 @@ void main() { shadow = mix(shadow, s, blend); } - blend_index++; + blend_count++; } - if (blend_index < 2) { + if (blend_count < blend_max) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 3) @@ -1356,7 +1366,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale4 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1365,21 +1375,9 @@ void main() { } } -#undef BIAS_FUNC } else { //no soft shadows - float depth_z = -vertex.z; - vec4 pssm_coord; - vec3 light_dir = directional_lights.data[i].direction; - vec3 base_normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))); - -#define BIAS_FUNC(m_var, m_idx) \ - m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ - vec3 normal_bias = base_normal_bias * directional_lights.data[i].shadow_normal_bias[m_idx]; \ - normal_bias -= light_dir * dot(light_dir, normal_bias); \ - m_var.xyz += normal_bias; - if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { vec4 v = vec4(vertex, 1.0); @@ -1438,11 +1436,11 @@ void main() { float shadow2 = sample_directional_pcf_shadow(directional_shadow_atlas, scene_data.directional_shadow_pixel_size * directional_lights.data[i].soft_shadow_scale, pssm_coord); shadow = mix(shadow, shadow2, pssm_blend); } + } - shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance + shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance #undef BIAS_FUNC - } } // shadows if (i < 4) { @@ -1594,7 +1592,7 @@ void main() { continue; // Statically baked light and object uses lightmap, skip } - float shadow = light_process_omni_shadow(light_index, vertex, view); + float shadow = light_process_omni_shadow(light_index, vertex, normal); shadow = blur_shadow(shadow); @@ -1670,7 +1668,7 @@ void main() { continue; // Statically baked light and object uses lightmap, skip } - float shadow = light_process_spot_shadow(light_index, vertex, view); + float shadow = light_process_spot_shadow(light_index, vertex, normal); shadow = blur_shadow(shadow); diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl index 3a05275652..b943d81784 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl @@ -177,7 +177,7 @@ layout(set = 1, binding = 0, std140) uniform SceneData { uint cluster_type_size; uint max_cluster_element_count_div_32; - //use vec4s because std140 doesnt play nice with vec2s, z and w are wasted + // Use vec4s because std140 doesn't play nice with vec2s, z and w are wasted. vec4 directional_penumbra_shadow_kernel[32]; vec4 directional_soft_shadow_kernel[32]; vec4 penumbra_shadow_kernel[32]; @@ -209,8 +209,6 @@ layout(set = 1, binding = 0, std140) uniform SceneData { float roughness_limiter_limit; uvec2 roughness_limiter_pad; - vec4 ao_color; - mat4 sdf_to_bounds; ivec3 sdf_offset; diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl index 7039ea2942..4d466342f8 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl @@ -97,6 +97,11 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #endif inout vec3 diffuse_light, inout vec3 specular_light) { + vec4 orms_unpacked = unpackUnorm4x8(orms); + + float roughness = orms_unpacked.y; + float metallic = orms_unpacked.z; + #if defined(LIGHT_CODE_USED) // light is written by the light shader @@ -125,9 +130,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float cLdotH = clamp(A + dot(L, H), 0.0, 1.0); #endif - float metallic = unpackUnorm4x8(orms).z; if (metallic < 1.0) { - float roughness = unpackUnorm4x8(orms).y; float diffuse_brdf_NL; // BRDF times N.L for calculating diffuse radiance #if defined(DIFFUSE_LAMBERT_WRAP) @@ -199,7 +202,6 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #endif //LIGHT_TRANSMITTANCE_USED } - float roughness = unpackUnorm4x8(orms).y; if (roughness > 0.0) { // FIXME: roughness == 0 should not disable specular light entirely // D @@ -208,11 +210,10 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte //normalized blinn float shininess = exp2(15.0 * (1.0 - roughness) + 1.0) * 0.25; - float blinn = pow(cNdotH, shininess) * cNdotL; - blinn *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); - float intensity = blinn; + float blinn = pow(cNdotH, shininess); + blinn *= (shininess + 2.0) * (1.0 / (8.0 * M_PI)); - specular_light += light_color * intensity * attenuation * specular_amount; + specular_light += light_color * attenuation * specular_amount * blinn * f0 * orms_unpacked.w; #elif defined(SPECULAR_PHONG) @@ -220,10 +221,9 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float cRdotV = clamp(A + dot(R, V), 0.0, 1.0); float shininess = exp2(15.0 * (1.0 - roughness) + 1.0) * 0.25; float phong = pow(cRdotV, shininess); - phong *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); - float intensity = (phong) / max(4.0 * cNdotV * cNdotL, 0.75); + phong *= (shininess + 1.0) * (1.0 / (8.0 * M_PI)); - specular_light += light_color * intensity * attenuation * specular_amount; + specular_light += light_color * attenuation * specular_amount * phong * f0 * orms_unpacked.w; #elif defined(SPECULAR_TOON) @@ -281,7 +281,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte } #ifdef USE_SHADOW_TO_OPACITY - alpha = min(alpha, clamp(1.0 - attenuation), 0.0, 1.0)); + alpha = min(alpha, clamp(1.0 - attenuation, 0.0, 1.0)); #endif #endif //defined(LIGHT_CODE_USED) @@ -290,7 +290,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #ifndef USE_NO_SHADOWS // Interleaved Gradient Noise -// http://www.iryoku.com/next-generation-post-processing-in-call-of-duty-advanced-warfare +// https://www.iryoku.com/next-generation-post-processing-in-call-of-duty-advanced-warfare float quick_hash(vec2 pos) { const vec3 magic = vec3(0.06711056f, 0.00583715f, 52.9829189f); return fract(magic.z * fract(dot(pos, magic.xy))); @@ -322,7 +322,7 @@ float sample_directional_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, ve return avg * (1.0 / float(sc_directional_soft_shadow_samples)); } -float sample_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec4 coord) { +float sample_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec3 coord) { vec2 pos = coord.xy; float depth = coord.z; @@ -348,6 +348,49 @@ float sample_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec4 coord) { return avg * (1.0 / float(sc_soft_shadow_samples)); } +float sample_omni_pcf_shadow(texture2D shadow, float blur_scale, vec2 coord, vec4 uv_rect, vec2 flip_offset, float depth) { + //if only one sample is taken, take it from the center + if (sc_soft_shadow_samples == 1) { + vec2 pos = coord * 0.5 + 0.5; + pos = uv_rect.xy + pos * uv_rect.zw; + return textureProj(sampler2DShadow(shadow, shadow_sampler), vec4(pos, depth, 1.0)); + } + + mat2 disk_rotation; + { + float r = quick_hash(gl_FragCoord.xy) * 2.0 * M_PI; + float sr = sin(r); + float cr = cos(r); + disk_rotation = mat2(vec2(cr, -sr), vec2(sr, cr)); + } + + float avg = 0.0; + vec2 offset_scale = blur_scale * 2.0 * scene_data.shadow_atlas_pixel_size / uv_rect.zw; + + for (uint i = 0; i < sc_soft_shadow_samples; i++) { + vec2 offset = offset_scale * (disk_rotation * scene_data.soft_shadow_kernel[i].xy); + vec2 sample_coord = coord + offset; + + float sample_coord_length_sqaured = dot(sample_coord, sample_coord); + bool do_flip = sample_coord_length_sqaured > 1.0; + + if (do_flip) { + float len = sqrt(sample_coord_length_sqaured); + sample_coord = sample_coord * (2.0 / len - 1.0); + } + + sample_coord = sample_coord * 0.5 + 0.5; + sample_coord = uv_rect.xy + sample_coord * uv_rect.zw; + + if (do_flip) { + sample_coord += flip_offset; + } + avg += textureProj(sampler2DShadow(shadow, shadow_sampler), vec4(sample_coord, depth, 1.0)); + } + + return avg * (1.0 / float(sc_soft_shadow_samples)); +} + float sample_directional_soft_shadow(texture2D shadow, vec3 pssm_coord, vec2 tex_scale) { //find blocker float blocker_count = 0.0; @@ -405,21 +448,21 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { #ifndef USE_NO_SHADOWS if (omni_lights.data[idx].shadow_enabled) { // there is a shadowmap + vec2 texel_size = scene_data.shadow_atlas_pixel_size; + vec4 base_uv_rect = omni_lights.data[idx].atlas_rect; + base_uv_rect.xy += texel_size; + base_uv_rect.zw -= texel_size * 2.0; - vec3 light_rel_vec = omni_lights.data[idx].position - vertex; - float light_length = length(light_rel_vec); + // Omni lights use direction.xy to store to store the offset between the two paraboloid regions + vec2 flip_offset = omni_lights.data[idx].direction.xy; - vec4 v = vec4(vertex, 1.0); + vec3 local_vert = (omni_lights.data[idx].shadow_matrix * vec4(vertex, 1.0)).xyz; - vec4 splane = (omni_lights.data[idx].shadow_matrix * v); - float shadow_len = length(splane.xyz); //need to remember shadow len from here + float shadow_len = length(local_vert); //need to remember shadow len from here + vec3 shadow_dir = normalize(local_vert); - { - vec3 nofs = normal_interp * omni_lights.data[idx].shadow_normal_bias / omni_lights.data[idx].inv_radius; - nofs *= (1.0 - max(0.0, dot(normalize(light_rel_vec), normalize(normal_interp)))); - v.xyz += nofs; - splane = (omni_lights.data[idx].shadow_matrix * v); - } + vec3 local_normal = normalize(mat3(omni_lights.data[idx].shadow_matrix) * normal); + vec3 normal_bias = local_normal * omni_lights.data[idx].shadow_normal_bias * (1.0 - abs(dot(local_normal, shadow_dir))); float shadow; @@ -439,10 +482,10 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { disk_rotation = mat2(vec2(cr, -sr), vec2(sr, cr)); } - vec3 normal = normalize(splane.xyz); - vec3 v0 = abs(normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); - vec3 tangent = normalize(cross(v0, normal)); - vec3 bitangent = normalize(cross(tangent, normal)); + vec3 basis_normal = shadow_dir; + vec3 v0 = abs(basis_normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); + vec3 tangent = normalize(cross(v0, basis_normal)); + vec3 bitangent = normalize(cross(tangent, basis_normal)); float z_norm = shadow_len * omni_lights.data[idx].inv_radius; tangent *= omni_lights.data[idx].soft_shadow_size * omni_lights.data[idx].soft_shadow_scale; @@ -451,18 +494,17 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { for (uint i = 0; i < sc_penumbra_shadow_samples; i++) { vec2 disk = disk_rotation * scene_data.penumbra_shadow_kernel[i].xy; - vec3 pos = splane.xyz + tangent * disk.x + bitangent * disk.y; + vec3 pos = local_vert + tangent * disk.x + bitangent * disk.y; pos = normalize(pos); - vec4 uv_rect = omni_lights.data[idx].atlas_rect; + + vec4 uv_rect = base_uv_rect; if (pos.z >= 0.0) { - pos.z += 1.0; - uv_rect.y += uv_rect.w; - } else { - pos.z = 1.0 - pos.z; + uv_rect.xy += flip_offset; } + pos.z = 1.0 + abs(pos.z); pos.xy /= pos.z; pos.xy = pos.xy * 0.5 + 0.5; @@ -487,18 +529,18 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { shadow = 0.0; for (uint i = 0; i < sc_penumbra_shadow_samples; i++) { vec2 disk = disk_rotation * scene_data.penumbra_shadow_kernel[i].xy; - vec3 pos = splane.xyz + tangent * disk.x + bitangent * disk.y; + vec3 pos = local_vert + tangent * disk.x + bitangent * disk.y; pos = normalize(pos); - vec4 uv_rect = omni_lights.data[idx].atlas_rect; + pos = normalize(pos + normal_bias); + + vec4 uv_rect = base_uv_rect; if (pos.z >= 0.0) { - pos.z += 1.0; - uv_rect.y += uv_rect.w; - } else { - pos.z = 1.0 - pos.z; + uv_rect.xy += flip_offset; } + pos.z = 1.0 + abs(pos.z); pos.xy /= pos.z; pos.xy = pos.xy * 0.5 + 0.5; @@ -513,25 +555,19 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { shadow = 1.0; } } else { - splane.xyz = normalize(splane.xyz); - vec4 clamp_rect = omni_lights.data[idx].atlas_rect; - - if (splane.z >= 0.0) { - splane.z += 1.0; + vec4 uv_rect = base_uv_rect; - clamp_rect.y += clamp_rect.w; - - } else { - splane.z = 1.0 - splane.z; + vec3 shadow_sample = normalize(shadow_dir + normal_bias); + if (shadow_sample.z >= 0.0) { + uv_rect.xy += flip_offset; + flip_offset *= -1.0; } - splane.xy /= splane.z; - - splane.xy = splane.xy * 0.5 + 0.5; - splane.z = (shadow_len - omni_lights.data[idx].shadow_bias) * omni_lights.data[idx].inv_radius; - splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; - splane.w = 1.0; //needed? i think it should be 1 already - shadow = sample_pcf_shadow(shadow_atlas, omni_lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, splane); + shadow_sample.z = 1.0 + abs(shadow_sample.z); + vec2 pos = shadow_sample.xy / shadow_sample.z; + float depth = shadow_len - omni_lights.data[idx].shadow_bias; + depth *= omni_lights.data[idx].inv_radius; + shadow = sample_omni_pcf_shadow(shadow_atlas, omni_lights.data[idx].soft_shadow_scale / shadow_sample.z, pos, uv_rect, flip_offset, depth); } return shadow; @@ -615,13 +651,11 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v vec4 atlas_rect = omni_lights.data[idx].projector_rect; if (local_v.z >= 0.0) { - local_v.z += 1.0; atlas_rect.y += atlas_rect.w; - - } else { - local_v.z = 1.0 - local_v.z; } + local_v.z = 1.0 + abs(local_v.z); + local_v.xy /= local_v.z; local_v.xy = local_v.xy * 0.5 + 0.5; vec2 proj_uv = local_v.xy * atlas_rect.zw; @@ -701,30 +735,23 @@ float light_process_spot_shadow(uint idx, vec3 vertex, vec3 normal) { vec3 light_rel_vec = spot_lights.data[idx].position - vertex; float light_length = length(light_rel_vec); vec3 spot_dir = spot_lights.data[idx].direction; - //there is a shadowmap - vec4 v = vec4(vertex, 1.0); - - v.xyz -= spot_dir * spot_lights.data[idx].shadow_bias; - - float z_norm = dot(spot_dir, -light_rel_vec) * spot_lights.data[idx].inv_radius; - float depth_bias_scale = 1.0 / (max(0.0001, z_norm)); //the closer to the light origin, the more you have to offset to reach 1px in the map - vec3 normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(spot_dir, -normalize(normal_interp)))) * spot_lights.data[idx].shadow_normal_bias * depth_bias_scale; - normal_bias -= spot_dir * dot(spot_dir, normal_bias); //only XY, no Z - v.xyz += normal_bias; + vec3 shadow_dir = light_rel_vec / light_length; + vec3 normal_bias = normal * light_length * spot_lights.data[idx].shadow_normal_bias * (1.0 - abs(dot(normal, shadow_dir))); - //adjust with bias - z_norm = dot(spot_dir, v.xyz - spot_lights.data[idx].position) * spot_lights.data[idx].inv_radius; - - float shadow; + //there is a shadowmap + vec4 v = vec4(vertex + normal_bias, 1.0); vec4 splane = (spot_lights.data[idx].shadow_matrix * v); + splane.z -= spot_lights.data[idx].shadow_bias / (light_length * spot_lights.data[idx].inv_radius); splane /= splane.w; + float shadow; if (sc_use_light_soft_shadows && spot_lights.data[idx].soft_shadow_size > 0.0) { //soft shadow //find blocker + float z_norm = dot(spot_dir, -light_rel_vec) * spot_lights.data[idx].inv_radius; vec2 shadow_uv = splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy; @@ -770,11 +797,9 @@ float light_process_spot_shadow(uint idx, vec3 vertex, vec3 normal) { //no blockers found, so no shadow shadow = 1.0; } - } else { //hard shadow - vec4 shadow_uv = vec4(splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy, splane.z, 1.0); - + vec3 shadow_uv = vec3(splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy, splane.z); shadow = sample_pcf_shadow(shadow_atlas, spot_lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, shadow_uv); } @@ -946,7 +971,7 @@ void reflection_process(uint ref_index, vec3 vertex, vec3 normal, float roughnes vec4 reflection; - reflection.rgb = textureLod(samplerCubeArray(reflection_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(local_ref_vec, reflections.data[ref_index].index), roughness * MAX_ROUGHNESS_LOD).rgb; + reflection.rgb = textureLod(samplerCubeArray(reflection_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(local_ref_vec, reflections.data[ref_index].index), roughness * MAX_ROUGHNESS_LOD).rgb * sc_luminance_multiplier; if (reflections.data[ref_index].exterior) { reflection.rgb = mix(specular_light, reflection.rgb, blend); diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl index 7e72ece5dc..0ee68d5e10 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl @@ -59,27 +59,27 @@ layout(location = 11) in vec4 weight_attrib; /* Varyings */ -layout(location = 0) out vec3 vertex_interp; +layout(location = 0) highp out vec3 vertex_interp; #ifdef NORMAL_USED -layout(location = 1) out vec3 normal_interp; +layout(location = 1) mediump out vec3 normal_interp; #endif #if defined(COLOR_USED) -layout(location = 2) out vec4 color_interp; +layout(location = 2) mediump out vec4 color_interp; #endif #ifdef UV_USED -layout(location = 3) out vec2 uv_interp; +layout(location = 3) mediump out vec2 uv_interp; #endif #if defined(UV2_USED) || defined(USE_LIGHTMAP) -layout(location = 4) out vec2 uv2_interp; +layout(location = 4) mediump out vec2 uv2_interp; #endif #if defined(TANGENT_USED) || defined(NORMAL_MAP_USED) || defined(LIGHT_ANISOTROPY_USED) -layout(location = 5) out vec3 tangent_interp; -layout(location = 6) out vec3 binormal_interp; +layout(location = 5) mediump out vec3 tangent_interp; +layout(location = 6) mediump out vec3 binormal_interp; #endif #ifdef MATERIAL_UNIFORMS_USED @@ -92,7 +92,7 @@ layout(set = MATERIAL_UNIFORM_SET, binding = 0, std140) uniform MaterialUniforms #ifdef MODE_DUAL_PARABOLOID -layout(location = 8) out float dp_clip; +layout(location = 8) out highp float dp_clip; #endif @@ -124,7 +124,7 @@ void main() { mat3 world_normal_matrix; if (bool(draw_call.flags & INSTANCE_FLAGS_NON_UNIFORM_SCALE)) { - world_normal_matrix = inverse(mat3(world_matrix)); + world_normal_matrix = transpose(inverse(mat3(world_matrix))); } else { world_normal_matrix = mat3(world_matrix); } @@ -372,55 +372,68 @@ void main() { /* Specialization Constants */ -/* Specialization Constants (Toggles) */ +#if !defined(MODE_RENDER_DEPTH) + +#if !defined(MODE_UNSHADED) + +layout(constant_id = 0) const bool sc_use_light_projector = false; +layout(constant_id = 1) const bool sc_use_light_soft_shadows = false; +layout(constant_id = 2) const bool sc_use_directional_soft_shadows = false; + +layout(constant_id = 3) const uint sc_soft_shadow_samples = 4; +layout(constant_id = 4) const uint sc_penumbra_shadow_samples = 4; -layout(constant_id = 0) const bool sc_use_forward_gi = false; -layout(constant_id = 1) const bool sc_use_light_projector = false; -layout(constant_id = 2) const bool sc_use_light_soft_shadows = false; -layout(constant_id = 3) const bool sc_use_directional_soft_shadows = false; +layout(constant_id = 5) const uint sc_directional_soft_shadow_samples = 4; +layout(constant_id = 6) const uint sc_directional_penumbra_shadow_samples = 4; -/* Specialization Constants (Values) */ +layout(constant_id = 8) const bool sc_projector_use_mipmaps = true; -layout(constant_id = 6) const uint sc_soft_shadow_samples = 4; -layout(constant_id = 7) const uint sc_penumbra_shadow_samples = 4; +layout(constant_id = 9) const bool sc_disable_omni_lights = false; +layout(constant_id = 10) const bool sc_disable_spot_lights = false; +layout(constant_id = 11) const bool sc_disable_reflection_probes = false; +layout(constant_id = 12) const bool sc_disable_directional_lights = false; -layout(constant_id = 8) const uint sc_directional_soft_shadow_samples = 4; -layout(constant_id = 9) const uint sc_directional_penumbra_shadow_samples = 4; +#endif //!MODE_UNSHADED + +layout(constant_id = 7) const bool sc_decal_use_mipmaps = true; +layout(constant_id = 13) const bool sc_disable_decals = false; +layout(constant_id = 14) const bool sc_disable_fog = false; + +#endif //!MODE_RENDER_DEPTH -layout(constant_id = 10) const bool sc_decal_use_mipmaps = true; -layout(constant_id = 11) const bool sc_projector_use_mipmaps = true; +layout(constant_id = 15) const float sc_luminance_multiplier = 2.0; /* Include our forward mobile UBOs definitions etc. */ #include "scene_forward_mobile_inc.glsl" /* Varyings */ -layout(location = 0) in vec3 vertex_interp; +layout(location = 0) highp in vec3 vertex_interp; #ifdef NORMAL_USED -layout(location = 1) in vec3 normal_interp; +layout(location = 1) mediump in vec3 normal_interp; #endif #if defined(COLOR_USED) -layout(location = 2) in vec4 color_interp; +layout(location = 2) mediump in vec4 color_interp; #endif #ifdef UV_USED -layout(location = 3) in vec2 uv_interp; +layout(location = 3) mediump in vec2 uv_interp; #endif #if defined(UV2_USED) || defined(USE_LIGHTMAP) -layout(location = 4) in vec2 uv2_interp; +layout(location = 4) mediump in vec2 uv2_interp; #endif #if defined(TANGENT_USED) || defined(NORMAL_MAP_USED) || defined(LIGHT_ANISOTROPY_USED) -layout(location = 5) in vec3 tangent_interp; -layout(location = 6) in vec3 binormal_interp; +layout(location = 5) mediump in vec3 tangent_interp; +layout(location = 6) mediump in vec3 binormal_interp; #endif #ifdef MODE_DUAL_PARABOLOID -layout(location = 8) in float dp_clip; +layout(location = 8) highp in float dp_clip; #endif @@ -482,7 +495,7 @@ layout(location = 0) out vec4 diffuse_buffer; //diffuse (rgb) and roughness layout(location = 1) out vec4 specular_buffer; //specular and SSS (subsurface scatter) #else -layout(location = 0) out vec4 frag_color; +layout(location = 0) out mediump vec4 frag_color; #endif // MODE_MULTIPLE_RENDER_TARGETS #endif // RENDER DEPTH @@ -491,6 +504,11 @@ layout(location = 0) out vec4 frag_color; #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) +/* Make a default specular mode SPECULAR_SCHLICK_GGX. */ +#if !defined(SPECULAR_DISABLED) && !defined(SPECULAR_SCHLICK_GGX) && !defined(SPECULAR_BLINN) && !defined(SPECULAR_PHONG) && !defined(SPECULAR_TOON) +#define SPECULAR_SCHLICK_GGX +#endif + #include "scene_forward_lights_inc.glsl" #endif //!defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) @@ -532,14 +550,14 @@ vec4 fog_process(vec3 vertex) { } } - float fog_amount = 1.0 - exp(min(0.0, vertex.z * scene_data.fog_density)); + float fog_amount = 1.0 - exp(min(0.0, -length(vertex) * scene_data.fog_density)); - if (abs(scene_data.fog_height_density) > 0.001) { + if (abs(scene_data.fog_height_density) >= 0.0001) { float y = (scene_data.camera_matrix * vec4(vertex, 1.0)).y; - float y_dist = scene_data.fog_height - y; + float y_dist = y - scene_data.fog_height; - float vfog_amount = clamp(exp(y_dist * scene_data.fog_height_density), 0.0, 1.0); + float vfog_amount = 1.0 - exp(min(0.0, y_dist * scene_data.fog_height_density)); fog_amount = max(vfog_amount, fog_amount); } @@ -726,7 +744,7 @@ void main() { // to maximize VGPR usage // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. - if (scene_data.fog_enabled) { + if (!sc_disable_fog && scene_data.fog_enabled) { fog = fog_process(vertex); } @@ -744,7 +762,7 @@ void main() { vec3 vertex_ddx = dFdx(vertex); vec3 vertex_ddy = dFdy(vertex); - { //Decals + if (!sc_disable_decals) { //Decals // must implement uint decal_indices = draw_call.decals.x; @@ -765,25 +783,35 @@ void main() { continue; //out of decal } - //we need ddx/ddy for mipmaps, so simulate them - vec2 ddx = (decals.data[decal_index].xform * vec4(vertex_ddx, 0.0)).xz; - vec2 ddy = (decals.data[decal_index].xform * vec4(vertex_ddy, 0.0)).xz; - float fade = pow(1.0 - (uv_local.y > 0.0 ? uv_local.y : -uv_local.y), uv_local.y > 0.0 ? decals.data[decal_index].upper_fade : decals.data[decal_index].lower_fade); if (decals.data[decal_index].normal_fade > 0.0) { fade *= smoothstep(decals.data[decal_index].normal_fade, 1.0, dot(normal_interp, decals.data[decal_index].normal) * 0.5 + 0.5); } + //we need ddx/ddy for mipmaps, so simulate them + vec2 ddx = (decals.data[decal_index].xform * vec4(vertex_ddx, 0.0)).xz; + vec2 ddy = (decals.data[decal_index].xform * vec4(vertex_ddy, 0.0)).xz; + if (decals.data[decal_index].albedo_rect != vec4(0.0)) { //has albedo - vec4 decal_albedo = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].albedo_rect.zw + decals.data[decal_index].albedo_rect.xy, ddx * decals.data[decal_index].albedo_rect.zw, ddy * decals.data[decal_index].albedo_rect.zw); + vec4 decal_albedo; + if (sc_decal_use_mipmaps) { + decal_albedo = textureGrad(sampler2D(decal_atlas_srgb, decal_sampler), uv_local.xz * decals.data[decal_index].albedo_rect.zw + decals.data[decal_index].albedo_rect.xy, ddx * decals.data[decal_index].albedo_rect.zw, ddy * decals.data[decal_index].albedo_rect.zw); + } else { + decal_albedo = textureLod(sampler2D(decal_atlas_srgb, decal_sampler), uv_local.xz * decals.data[decal_index].albedo_rect.zw + decals.data[decal_index].albedo_rect.xy, 0.0); + } decal_albedo *= decals.data[decal_index].modulate; decal_albedo.a *= fade; albedo = mix(albedo, decal_albedo.rgb, decal_albedo.a * decals.data[decal_index].albedo_mix); if (decals.data[decal_index].normal_rect != vec4(0.0)) { - vec3 decal_normal = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].normal_rect.zw + decals.data[decal_index].normal_rect.xy, ddx * decals.data[decal_index].normal_rect.zw, ddy * decals.data[decal_index].normal_rect.zw).xyz; + vec3 decal_normal; + if (sc_decal_use_mipmaps) { + decal_normal = textureGrad(sampler2D(decal_atlas, decal_sampler), uv_local.xz * decals.data[decal_index].normal_rect.zw + decals.data[decal_index].normal_rect.xy, ddx * decals.data[decal_index].normal_rect.zw, ddy * decals.data[decal_index].normal_rect.zw).xyz; + } else { + decal_normal = textureLod(sampler2D(decal_atlas, decal_sampler), uv_local.xz * decals.data[decal_index].normal_rect.zw + decals.data[decal_index].normal_rect.xy, 0.0).xyz; + } decal_normal.xy = decal_normal.xy * vec2(2.0, -2.0) - vec2(1.0, -1.0); //users prefer flipped y normal maps in most authoring software decal_normal.z = sqrt(max(0.0, 1.0 - dot(decal_normal.xy, decal_normal.xy))); //convert to view space, use xzy because y is up @@ -793,7 +821,12 @@ void main() { } if (decals.data[decal_index].orm_rect != vec4(0.0)) { - vec3 decal_orm = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].orm_rect.zw + decals.data[decal_index].orm_rect.xy, ddx * decals.data[decal_index].orm_rect.zw, ddy * decals.data[decal_index].orm_rect.zw).xyz; + vec3 decal_orm; + if (sc_decal_use_mipmaps) { + decal_orm = textureGrad(sampler2D(decal_atlas, decal_sampler), uv_local.xz * decals.data[decal_index].orm_rect.zw + decals.data[decal_index].orm_rect.xy, ddx * decals.data[decal_index].orm_rect.zw, ddy * decals.data[decal_index].orm_rect.zw).xyz; + } else { + decal_orm = textureLod(sampler2D(decal_atlas, decal_sampler), uv_local.xz * decals.data[decal_index].orm_rect.zw + decals.data[decal_index].orm_rect.xy, 0.0).xyz; + } ao = mix(ao, decal_orm.r, decal_albedo.a); roughness = mix(roughness, decal_orm.g, decal_albedo.a); metallic = mix(metallic, decal_orm.b, decal_albedo.a); @@ -802,7 +835,11 @@ void main() { if (decals.data[decal_index].emission_rect != vec4(0.0)) { //emission is additive, so its independent from albedo - emission += textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].emission_rect.zw + decals.data[decal_index].emission_rect.xy, ddx * decals.data[decal_index].emission_rect.zw, ddy * decals.data[decal_index].emission_rect.zw).xyz * decals.data[decal_index].emission_energy * fade; + if (sc_decal_use_mipmaps) { + emission += textureGrad(sampler2D(decal_atlas_srgb, decal_sampler), uv_local.xz * decals.data[decal_index].emission_rect.zw + decals.data[decal_index].emission_rect.xy, ddx * decals.data[decal_index].emission_rect.zw, ddy * decals.data[decal_index].emission_rect.zw).xyz * decals.data[decal_index].emission_energy * fade; + } else { + emission += textureLod(sampler2D(decal_atlas_srgb, decal_sampler), uv_local.xz * decals.data[decal_index].emission_rect.zw + decals.data[decal_index].emission_rect.xy, 0.0).xyz * decals.data[decal_index].emission_energy * fade; + } } } } //Decals @@ -812,7 +849,7 @@ void main() { #ifdef NORMAL_USED if (scene_data.roughness_limiter_enabled) { - //http://www.jp.square-enix.com/tech/library/pdf/ImprovedGeometricSpecularAA.pdf + //https://www.jp.square-enix.com/tech/library/pdf/ImprovedGeometricSpecularAA.pdf float roughness2 = roughness * roughness; vec3 dndu = dFdx(normal), dndv = dFdy(normal); float variance = scene_data.roughness_limiter_amount * (dot(dndu, dndu) + dot(dndv, dndv)); @@ -831,6 +868,7 @@ void main() { if (scene_data.use_reflection_cubemap) { vec3 ref_vec = reflect(-view, normal); + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); ref_vec = scene_data.radiance_inverse_xform * ref_vec; #ifdef USE_RADIANCE_CUBEMAP_ARRAY @@ -843,6 +881,7 @@ void main() { specular_light = textureLod(samplerCube(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), ref_vec, roughness * MAX_ROUGHNESS_LOD).rgb; #endif //USE_RADIANCE_CUBEMAP_ARRAY + specular_light *= horizon * horizon; specular_light *= scene_data.ambient_light_color_energy.a; } @@ -940,7 +979,7 @@ void main() { // skipping ssao, do we remove ssao totally? - { //Reflection probes + if (!sc_disable_reflection_probes) { //Reflection probes vec4 reflection_accum = vec4(0.0, 0.0, 0.0, 0.0); vec4 ambient_accum = vec4(0.0, 0.0, 0.0, 0.0); @@ -981,7 +1020,7 @@ void main() { specular_light *= specular * metallic * albedo * 2.0; #else - // scales the specular reflections, needs to be be computed before lighting happens, + // scales the specular reflections, needs to be computed before lighting happens, // but after environment, GI, and reflection probes are added // Environment brdf approximation (Lazarov 2013) // see https://www.unrealengine.com/en-US/blog/physically-based-shading-on-mobile @@ -1006,7 +1045,7 @@ void main() { // LIGHTING #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) - { //directional light + if (!sc_disable_directional_lights) { //directional light // Do shadow and lighting in two passes to reduce register pressure uint shadow0 = 0; @@ -1336,7 +1375,7 @@ void main() { } } //directional light - { //omni lights + if (!sc_disable_omni_lights) { //omni lights uint light_indices = draw_call.omni_lights.x; for (uint i = 0; i < 8; i++) { uint light_index = light_indices & 0xFF; @@ -1350,7 +1389,7 @@ void main() { break; } - float shadow = light_process_omni_shadow(light_index, vertex, view); + float shadow = light_process_omni_shadow(light_index, vertex, normal); shadow = blur_shadow(shadow); @@ -1383,7 +1422,7 @@ void main() { } } //omni lights - { //spot lights + if (!sc_disable_spot_lights) { //spot lights uint light_indices = draw_call.spot_lights.x; for (uint i = 0; i < 8; i++) { @@ -1398,7 +1437,7 @@ void main() { break; } - float shadow = light_process_spot_shadow(light_index, vertex, view); + float shadow = light_process_spot_shadow(light_index, vertex, normal); shadow = blur_shadow(shadow); @@ -1514,12 +1553,15 @@ void main() { frag_color = vec4(albedo, alpha); #else // MODE_UNSHADED frag_color = vec4(emission + ambient_light + diffuse_light + specular_light, alpha); - //frag_color = vec4(1.0); #endif // MODE_UNSHADED // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); + // On mobile we use a UNORM buffer with 10bpp which results in a range from 0.0 - 1.0 resulting in HDR breaking + // We divide by sc_luminance_multiplier to support a range from 0.0 - 2.0 both increasing precision on bright and darker images + frag_color.rgb = frag_color.rgb / sc_luminance_multiplier; + #endif //MODE_MULTIPLE_RENDER_TARGETS #endif //MODE_RENDER_DEPTH diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl index 1b8e21551c..eb8fb49598 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl @@ -16,12 +16,12 @@ /* don't exceed 128 bytes!! */ /* put instance data into our push content, not a array */ layout(push_constant, binding = 0, std430) uniform DrawCall { - mat4 transform; // 64 - 64 + highp mat4 transform; // 64 - 64 uint flags; // 04 - 68 uint instance_uniforms_ofs; //base offset in global buffer for instance variables // 04 - 72 uint gi_offset; //GI information when using lightmapping (VCT or lightmap index) // 04 - 76 uint layer_mask; // 04 - 80 - vec4 lightmap_uv_scale; // 16 - 96 doubles as uv_offset when needed + highp vec4 lightmap_uv_scale; // 16 - 96 doubles as uv_offset when needed uvec2 reflection_probes; // 08 - 104 uvec2 omni_lights; // 08 - 112 @@ -93,7 +93,7 @@ directional_lights; #define LIGHTMAP_FLAG_USE_SPECULAR_DIRECTION 2 struct Lightmap { - mat3 normal_xform; + mediump mat3 normal_xform; }; layout(set = 0, binding = 9, std140) restrict readonly buffer Lightmaps { @@ -102,7 +102,7 @@ layout(set = 0, binding = 9, std140) restrict readonly buffer Lightmaps { lightmaps; struct LightmapCapture { - vec4 sh[9]; + mediump vec4 sh[9]; }; layout(set = 0, binding = 10, std140) restrict readonly buffer LightmapCaptures { @@ -110,8 +110,8 @@ layout(set = 0, binding = 10, std140) restrict readonly buffer LightmapCaptures } lightmap_captures; -layout(set = 0, binding = 11) uniform texture2D decal_atlas; -layout(set = 0, binding = 12) uniform texture2D decal_atlas_srgb; +layout(set = 0, binding = 11) uniform mediump texture2D decal_atlas; +layout(set = 0, binding = 12) uniform mediump texture2D decal_atlas_srgb; layout(set = 0, binding = 13, std430) restrict readonly buffer Decals { DecalData data[]; @@ -119,72 +119,70 @@ layout(set = 0, binding = 13, std430) restrict readonly buffer Decals { decals; layout(set = 0, binding = 14, std430) restrict readonly buffer GlobalVariableData { - vec4 data[]; + highp vec4 data[]; } global_variables; /* Set 1: Render Pass (changes per render pass) */ layout(set = 1, binding = 0, std140) uniform SceneData { - mat4 projection_matrix; - mat4 inv_projection_matrix; - mat4 camera_matrix; - mat4 inv_camera_matrix; + highp mat4 projection_matrix; + highp mat4 inv_projection_matrix; + highp mat4 camera_matrix; + highp mat4 inv_camera_matrix; // only used for multiview - mat4 projection_matrix_view[MAX_VIEWS]; - mat4 inv_projection_matrix_view[MAX_VIEWS]; + highp mat4 projection_matrix_view[MAX_VIEWS]; + highp mat4 inv_projection_matrix_view[MAX_VIEWS]; - vec2 viewport_size; - vec2 screen_pixel_size; + highp vec2 viewport_size; + highp vec2 screen_pixel_size; - //use vec4s because std140 doesnt play nice with vec2s, z and w are wasted - vec4 directional_penumbra_shadow_kernel[32]; - vec4 directional_soft_shadow_kernel[32]; - vec4 penumbra_shadow_kernel[32]; - vec4 soft_shadow_kernel[32]; + // Use vec4s because std140 doesn't play nice with vec2s, z and w are wasted. + highp vec4 directional_penumbra_shadow_kernel[32]; + highp vec4 directional_soft_shadow_kernel[32]; + highp vec4 penumbra_shadow_kernel[32]; + highp vec4 soft_shadow_kernel[32]; - vec4 ambient_light_color_energy; + mediump vec4 ambient_light_color_energy; - float ambient_color_sky_mix; + mediump float ambient_color_sky_mix; bool use_ambient_light; bool use_ambient_cubemap; bool use_reflection_cubemap; - mat3 radiance_inverse_xform; + mediump mat3 radiance_inverse_xform; - vec2 shadow_atlas_pixel_size; - vec2 directional_shadow_pixel_size; + highp vec2 shadow_atlas_pixel_size; + highp vec2 directional_shadow_pixel_size; uint directional_light_count; - float dual_paraboloid_side; - float z_far; - float z_near; + mediump float dual_paraboloid_side; + highp float z_far; + highp float z_near; bool ssao_enabled; - float ssao_light_affect; - float ssao_ao_affect; + mediump float ssao_light_affect; + mediump float ssao_ao_affect; bool roughness_limiter_enabled; - float roughness_limiter_amount; - float roughness_limiter_limit; + mediump float roughness_limiter_amount; + mediump float roughness_limiter_limit; uvec2 roughness_limiter_pad; - vec4 ao_color; - bool fog_enabled; - float fog_density; - float fog_height; - float fog_height_density; + highp float fog_density; + highp float fog_height; + highp float fog_height_density; - vec3 fog_light_color; - float fog_sun_scatter; + mediump vec3 fog_light_color; + mediump float fog_sun_scatter; - float fog_aerial_perspective; + mediump float fog_aerial_perspective; bool material_uv2_mode; - float time; - float reflection_multiplier; // one normally, zero when rendering reflections + highp float time; + mediump float reflection_multiplier; // one normally, zero when rendering reflections bool pancake_shadows; uint pad1; @@ -195,30 +193,30 @@ scene_data; #ifdef USE_RADIANCE_CUBEMAP_ARRAY -layout(set = 1, binding = 2) uniform textureCubeArray radiance_cubemap; +layout(set = 1, binding = 2) uniform mediump textureCubeArray radiance_cubemap; #else -layout(set = 1, binding = 2) uniform textureCube radiance_cubemap; +layout(set = 1, binding = 2) uniform mediump textureCube radiance_cubemap; #endif -layout(set = 1, binding = 3) uniform textureCubeArray reflection_atlas; +layout(set = 1, binding = 3) uniform mediump textureCubeArray reflection_atlas; -layout(set = 1, binding = 4) uniform texture2D shadow_atlas; +layout(set = 1, binding = 4) uniform highp texture2D shadow_atlas; -layout(set = 1, binding = 5) uniform texture2D directional_shadow_atlas; +layout(set = 1, binding = 5) uniform highp texture2D directional_shadow_atlas; // this needs to change to providing just the lightmap we're using.. layout(set = 1, binding = 6) uniform texture2DArray lightmap_textures[MAX_LIGHTMAP_TEXTURES]; -layout(set = 1, binding = 9) uniform texture2D depth_buffer; -layout(set = 1, binding = 10) uniform texture2D color_buffer; +layout(set = 1, binding = 9) uniform highp texture2D depth_buffer; +layout(set = 1, binding = 10) uniform mediump texture2D color_buffer; /* Set 2 Skeleton & Instancing (can change per item) */ layout(set = 2, binding = 0, std430) restrict readonly buffer Transforms { - vec4 data[]; + highp vec4 data[]; } transforms; diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl index 0eacbc5363..4290d5b869 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl @@ -24,7 +24,7 @@ layout(push_constant, binding = 0, std430) uniform Params { } params; -// http://in4k.untergrund.net/html_articles/hugi_27_-_coding_corner_polaris_sphere_tessellation_101.htm +// https://in4k.untergrund.net/html_articles/hugi_27_-_coding_corner_polaris_sphere_tessellation_101.htm vec3 get_sphere_vertex(uint p_vertex_id) { float x_angle = float(p_vertex_id & 1u) + (p_vertex_id >> params.band_power); diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl index 99db35bb34..d6e5c6a92e 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl @@ -20,10 +20,10 @@ layout(set = 0, binding = 3, std430) restrict readonly buffer DispatchData { dispatch_data; struct ProcessVoxel { - uint position; //xyz 7 bit packed, extra 11 bits for neigbours - uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours - uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours + uint position; // xyz 7 bit packed, extra 11 bits for neighbors. + uint albedo; // rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbors. + uint light; // rgbe8985 encoded total saved light, extra 2 bits for neighbors. + uint light_aniso; // 55555 light anisotropy, extra 2 bits for neighbors. //total neighbours: 26 }; diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl index bc376e9522..eedd28959c 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl @@ -266,9 +266,9 @@ void main() { } else if (params.sky_mode == SKY_MODE_SKY) { #ifdef USE_CUBEMAP_ARRAY - light.rgb = textureLod(samplerCubeArray(sky_irradiance, linear_sampler_mipmaps), vec4(ray_dir, 0.0), 2.0).rgb; //use second mipmap because we dont usually throw a lot of rays, so this compensates + light.rgb = textureLod(samplerCubeArray(sky_irradiance, linear_sampler_mipmaps), vec4(ray_dir, 0.0), 2.0).rgb; // Use second mipmap because we don't usually throw a lot of rays, so this compensates. #else - light.rgb = textureLod(samplerCube(sky_irradiance, linear_sampler_mipmaps), ray_dir, 2.0).rgb; //use second mipmap because we dont usually throw a lot of rays, so this compensates + light.rgb = textureLod(samplerCube(sky_irradiance, linear_sampler_mipmaps), ray_dir, 2.0).rgb; // Use second mipmap because we don't usually throw a lot of rays, so this compensates. #endif light.rgb *= params.sky_energy; light.a = 0.0; diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl index aa4ded146f..4d9fa85a74 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl @@ -101,7 +101,7 @@ layout(set = 0, binding = 10, std430) restrict buffer DispatchData { dispatch_data; struct ProcessVoxel { - uint position; //xyz 7 bit packed, extra 11 bits for neigbours + uint position; // xyz 7 bit packed, extra 11 bits for neighbors. uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours @@ -134,7 +134,7 @@ layout(set = 0, binding = 5, std430) restrict buffer readonly DispatchData { dispatch_data; struct ProcessVoxel { - uint position; //xyz 7 bit packed, extra 11 bits for neigbours + uint position; // xyz 7 bit packed, extra 11 bits for neighbors. uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours @@ -183,7 +183,7 @@ void main() { ivec3 write_pos = read_pos + params.scroll; if (any(lessThan(write_pos, ivec3(0))) || any(greaterThanEqual(write_pos, ivec3(params.grid_size)))) { - return; //fits outside the 3D texture, dont do anything + return; // Fits outside the 3D texture, don't do anything. } uint albedo = ((src_process_voxels.data[index].albedo & 0x7FFF) << 1) | 1; //add solid bit diff --git a/servers/rendering/renderer_rd/shaders/sky.glsl b/servers/rendering/renderer_rd/shaders/sky.glsl index 41c6325bc5..d07a454ade 100644 --- a/servers/rendering/renderer_rd/shaders/sky.glsl +++ b/servers/rendering/renderer_rd/shaders/sky.glsl @@ -17,6 +17,8 @@ layout(push_constant, binding = 1, std430) uniform Params { vec4 projections[MAX_VIEWS]; vec4 position_multiplier; float time; + float luminance_multiplier; + float pad[2]; } params; @@ -55,6 +57,8 @@ layout(push_constant, binding = 1, std430) uniform Params { vec4 projections[MAX_VIEWS]; vec4 position_multiplier; float time; + float luminance_multiplier; + float pad[2]; } params; @@ -199,17 +203,17 @@ void main() { vec3 inverted_cube_normal = cube_normal; inverted_cube_normal.z *= -1.0; #ifdef USES_HALF_RES_COLOR - half_res_color = texture(samplerCube(half_res, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), inverted_cube_normal); + half_res_color = texture(samplerCube(half_res, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), inverted_cube_normal) * params.luminance_multiplier; #endif #ifdef USES_QUARTER_RES_COLOR - quarter_res_color = texture(samplerCube(quarter_res, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), inverted_cube_normal); + quarter_res_color = texture(samplerCube(quarter_res, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), inverted_cube_normal) * params.luminance_multiplier; #endif #else #ifdef USES_HALF_RES_COLOR - half_res_color = textureLod(sampler2D(half_res, material_samplers[SAMPLER_LINEAR_CLAMP]), uv, 0.0); + half_res_color = textureLod(sampler2D(half_res, material_samplers[SAMPLER_LINEAR_CLAMP]), uv, 0.0) * params.luminance_multiplier; #endif #ifdef USES_QUARTER_RES_COLOR - quarter_res_color = textureLod(sampler2D(quarter_res, material_samplers[SAMPLER_LINEAR_CLAMP]), uv, 0.0); + quarter_res_color = textureLod(sampler2D(quarter_res, material_samplers[SAMPLER_LINEAR_CLAMP]), uv, 0.0) * params.luminance_multiplier; #endif #endif @@ -246,4 +250,7 @@ void main() { if (!AT_CUBEMAP_PASS && !AT_HALF_RES_PASS && !AT_QUARTER_RES_PASS) { frag_color.a = 0.0; } + + // For mobile renderer we're dividing by 2.0 as we're using a UNORM buffer + frag_color.rgb = frag_color.rgb / params.luminance_multiplier; } diff --git a/servers/rendering/renderer_rd/shaders/tonemap.glsl b/servers/rendering/renderer_rd/shaders/tonemap.glsl index 23f83b3b9c..1ce3e04421 100644 --- a/servers/rendering/renderer_rd/shaders/tonemap.glsl +++ b/servers/rendering/renderer_rd/shaders/tonemap.glsl @@ -35,13 +35,17 @@ void main() { layout(location = 0) in vec2 uv_interp; -#ifdef MULTIVIEW +#ifdef SUBPASS +layout(input_attachment_index = 0, set = 0, binding = 0) uniform subpassInput input_color; +#elif defined(MULTIVIEW) layout(set = 0, binding = 0) uniform sampler2DArray source_color; #else layout(set = 0, binding = 0) uniform sampler2D source_color; #endif + layout(set = 1, binding = 0) uniform sampler2D source_auto_exposure; layout(set = 2, binding = 0) uniform sampler2D source_glow; + #ifdef USE_1D_LUT layout(set = 3, binding = 0) uniform sampler2D source_color_correction; #else @@ -67,7 +71,7 @@ layout(push_constant, binding = 1, std430) uniform Params { float exposure; float white; float auto_exposure_grey; - uint pad2; + float luminance_multiplier; vec2 pixel_size; bool use_fxaa; @@ -165,25 +169,38 @@ vec3 tonemap_filmic(vec3 color, float white) { return color_tonemapped / white_tonemapped; } +// Adapted from https://github.com/TheRealMJP/BakingLab/blob/master/BakingLab/ACES.hlsl +// (MIT License). vec3 tonemap_aces(vec3 color, float white) { - const float exposure_bias = 0.85f; - const float A = 2.51f * exposure_bias * exposure_bias; - const float B = 0.03f * exposure_bias; - const float C = 2.43f * exposure_bias * exposure_bias; - const float D = 0.59f * exposure_bias; - const float E = 0.14f; - - vec3 color_tonemapped = (color * (A * color + B)) / (color * (C * color + D) + E); - float white_tonemapped = (white * (A * white + B)) / (white * (C * white + D) + E); + const float exposure_bias = 1.8f; + const float A = 0.0245786f; + const float B = 0.000090537f; + const float C = 0.983729f; + const float D = 0.432951f; + const float E = 0.238081f; + + // Exposure bias baked into transform to save shader instructions. Equivalent to `color *= exposure_bias` + const mat3 rgb_to_rrt = mat3( + vec3(0.59719f * exposure_bias, 0.35458f * exposure_bias, 0.04823f * exposure_bias), + vec3(0.07600f * exposure_bias, 0.90834f * exposure_bias, 0.01566f * exposure_bias), + vec3(0.02840f * exposure_bias, 0.13383f * exposure_bias, 0.83777f * exposure_bias)); + + const mat3 odt_to_rgb = mat3( + vec3(1.60475f, -0.53108f, -0.07367f), + vec3(-0.10208f, 1.10813f, -0.00605f), + vec3(-0.00327f, -0.07276f, 1.07602f)); + + color *= rgb_to_rrt; + vec3 color_tonemapped = (color * (color + A) - B) / (color * (C * color + D) + E); + color_tonemapped *= odt_to_rgb; + + white *= exposure_bias; + float white_tonemapped = (white * (white + A) - B) / (white * (C * white + D) + E); return color_tonemapped / white_tonemapped; } vec3 tonemap_reinhard(vec3 color, float white) { - // Ensure color values are positive. - // They can be negative in the case of negative lights, which leads to undesired behavior. - color = max(vec3(0.0), color); - return (white * color + color) / (color * white + white); } @@ -200,15 +217,16 @@ vec3 linear_to_srgb(vec3 color) { #define TONEMAPPER_ACES 3 vec3 apply_tonemapping(vec3 color, float white) { // inputs are LINEAR, always outputs clamped [0;1] color - + // Ensure color values passed to tonemappers are positive. + // They can be negative in the case of negative lights, which leads to undesired behavior. if (params.tonemapper == TONEMAPPER_LINEAR) { return color; } else if (params.tonemapper == TONEMAPPER_REINHARD) { - return tonemap_reinhard(color, white); + return tonemap_reinhard(max(vec3(0.0f), color), white); } else if (params.tonemapper == TONEMAPPER_FILMIC) { - return tonemap_filmic(color, white); - } else { //aces - return tonemap_aces(color, white); + return tonemap_filmic(max(vec3(0.0f), color), white); + } else { // TONEMAPPER_ACES + return tonemap_aces(max(vec3(0.0f), color), white); } } @@ -291,21 +309,22 @@ vec3 apply_color_correction(vec3 color) { } #endif +#ifndef SUBPASS vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { const float FXAA_REDUCE_MIN = (1.0 / 128.0); const float FXAA_REDUCE_MUL = (1.0 / 8.0); const float FXAA_SPAN_MAX = 8.0; #ifdef MULTIVIEW - vec3 rgbNW = textureLod(source_color, vec3(uv_interp + vec2(-1.0, -1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure; - vec3 rgbNE = textureLod(source_color, vec3(uv_interp + vec2(1.0, -1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure; - vec3 rgbSW = textureLod(source_color, vec3(uv_interp + vec2(-1.0, 1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure; - vec3 rgbSE = textureLod(source_color, vec3(uv_interp + vec2(1.0, 1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure; + vec3 rgbNW = textureLod(source_color, vec3(uv_interp + vec2(-1.0, -1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbNE = textureLod(source_color, vec3(uv_interp + vec2(1.0, -1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbSW = textureLod(source_color, vec3(uv_interp + vec2(-1.0, 1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbSE = textureLod(source_color, vec3(uv_interp + vec2(1.0, 1.0) * params.pixel_size, ViewIndex), 0.0).xyz * exposure * params.luminance_multiplier; #else - vec3 rgbNW = textureLod(source_color, uv_interp + vec2(-1.0, -1.0) * params.pixel_size, 0.0).xyz * exposure; - vec3 rgbNE = textureLod(source_color, uv_interp + vec2(1.0, -1.0) * params.pixel_size, 0.0).xyz * exposure; - vec3 rgbSW = textureLod(source_color, uv_interp + vec2(-1.0, 1.0) * params.pixel_size, 0.0).xyz * exposure; - vec3 rgbSE = textureLod(source_color, uv_interp + vec2(1.0, 1.0) * params.pixel_size, 0.0).xyz * exposure; + vec3 rgbNW = textureLod(source_color, uv_interp + vec2(-1.0, -1.0) * params.pixel_size, 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbNE = textureLod(source_color, uv_interp + vec2(1.0, -1.0) * params.pixel_size, 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbSW = textureLod(source_color, uv_interp + vec2(-1.0, 1.0) * params.pixel_size, 0.0).xyz * exposure * params.luminance_multiplier; + vec3 rgbSE = textureLod(source_color, uv_interp + vec2(1.0, 1.0) * params.pixel_size, 0.0).xyz * exposure * params.luminance_multiplier; #endif vec3 rgbM = color; vec3 luma = vec3(0.299, 0.587, 0.114); @@ -332,11 +351,11 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { params.pixel_size; #ifdef MULTIVIEW - vec3 rgbA = 0.5 * exposure * (textureLod(source_color, vec3(uv_interp + dir * (1.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz + textureLod(source_color, vec3(uv_interp + dir * (2.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz); - vec3 rgbB = rgbA * 0.5 + 0.25 * exposure * (textureLod(source_color, vec3(uv_interp + dir * -0.5, ViewIndex), 0.0).xyz + textureLod(source_color, vec3(uv_interp + dir * 0.5, ViewIndex), 0.0).xyz); + vec3 rgbA = 0.5 * exposure * (textureLod(source_color, vec3(uv_interp + dir * (1.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz + textureLod(source_color, vec3(uv_interp + dir * (2.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz) * params.luminance_multiplier; + vec3 rgbB = rgbA * 0.5 + 0.25 * exposure * (textureLod(source_color, vec3(uv_interp + dir * -0.5, ViewIndex), 0.0).xyz + textureLod(source_color, vec3(uv_interp + dir * 0.5, ViewIndex), 0.0).xyz) * params.luminance_multiplier; #else - vec3 rgbA = 0.5 * exposure * (textureLod(source_color, uv_interp + dir * (1.0 / 3.0 - 0.5), 0.0).xyz + textureLod(source_color, uv_interp + dir * (2.0 / 3.0 - 0.5), 0.0).xyz); - vec3 rgbB = rgbA * 0.5 + 0.25 * exposure * (textureLod(source_color, uv_interp + dir * -0.5, 0.0).xyz + textureLod(source_color, uv_interp + dir * 0.5, 0.0).xyz); + vec3 rgbA = 0.5 * exposure * (textureLod(source_color, uv_interp + dir * (1.0 / 3.0 - 0.5), 0.0).xyz + textureLod(source_color, uv_interp + dir * (2.0 / 3.0 - 0.5), 0.0).xyz) * params.luminance_multiplier; + vec3 rgbB = rgbA * 0.5 + 0.25 * exposure * (textureLod(source_color, uv_interp + dir * -0.5, 0.0).xyz + textureLod(source_color, uv_interp + dir * 0.5, 0.0).xyz) * params.luminance_multiplier; #endif float lumaB = dot(rgbB, luma); @@ -346,8 +365,9 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { return rgbB; } } +#endif // !SUBPASS -// From http://alex.vlachos.com/graphics/Alex_Vlachos_Advanced_VR_Rendering_GDC2015.pdf +// From https://alex.vlachos.com/graphics/Alex_Vlachos_Advanced_VR_Rendering_GDC2015.pdf // and https://www.shadertoy.com/view/MslGR8 (5th one starting from the bottom) // NOTE: `frag_coord` is in pixels (i.e. not normalized UV). vec3 screen_space_dither(vec2 frag_coord) { @@ -360,45 +380,54 @@ vec3 screen_space_dither(vec2 frag_coord) { } void main() { -#ifdef MULTIVIEW - vec3 color = textureLod(source_color, vec3(uv_interp, ViewIndex), 0.0f).rgb; +#ifdef SUBPASS + // SUBPASS and MULTIVIEW can be combined but in that case we're already reading from the correct layer + vec3 color = subpassLoad(input_color).rgb * params.luminance_multiplier; +#elif defined(MULTIVIEW) + vec3 color = textureLod(source_color, vec3(uv_interp, ViewIndex), 0.0f).rgb * params.luminance_multiplier; #else - vec3 color = textureLod(source_color, uv_interp, 0.0f).rgb; + vec3 color = textureLod(source_color, uv_interp, 0.0f).rgb * params.luminance_multiplier; #endif // Exposure float exposure = params.exposure; +#ifndef SUBPASS if (params.use_auto_exposure) { - exposure *= 1.0 / (texelFetch(source_auto_exposure, ivec2(0, 0), 0).r / params.auto_exposure_grey); + exposure *= 1.0 / (texelFetch(source_auto_exposure, ivec2(0, 0), 0).r * params.luminance_multiplier / params.auto_exposure_grey); } +#endif color *= exposure; // Early Tonemap & SRGB Conversion - +#ifndef SUBPASS if (params.use_glow && params.glow_mode == GLOW_MODE_MIX) { - vec3 glow = gather_glow(source_glow, uv_interp); + vec3 glow = gather_glow(source_glow, uv_interp) * params.luminance_multiplier; color.rgb = mix(color.rgb, glow, params.glow_intensity); } if (params.use_fxaa) { color = do_fxaa(color, exposure, uv_interp); } +#endif + if (params.use_debanding) { // For best results, debanding should be done before tonemapping. // Otherwise, we're adding noise to an already-quantized image. color += screen_space_dither(gl_FragCoord.xy); } + color = apply_tonemapping(color, params.white); color = linear_to_srgb(color); // regular linear -> SRGB conversion +#ifndef SUBPASS // Glow if (params.use_glow && params.glow_mode != GLOW_MODE_MIX) { - vec3 glow = gather_glow(source_glow, uv_interp) * params.glow_intensity; + vec3 glow = gather_glow(source_glow, uv_interp) * params.glow_intensity * params.luminance_multiplier; // high dynamic range -> SRGB glow = apply_tonemapping(glow, params.white); @@ -406,6 +435,7 @@ void main() { color = apply_glow(color, glow); } +#endif // Additional effects diff --git a/servers/rendering/renderer_rd/shaders/voxel_gi.glsl b/servers/rendering/renderer_rd/shaders/voxel_gi.glsl index 49a493cdc7..779f04ed35 100644 --- a/servers/rendering/renderer_rd/shaders/voxel_gi.glsl +++ b/servers/rendering/renderer_rd/shaders/voxel_gi.glsl @@ -71,11 +71,6 @@ lights; layout(set = 0, binding = 5) uniform texture3D color_texture; -#ifdef MODE_ANISOTROPIC -layout(set = 0, binding = 7) uniform texture3D aniso_pos_texture; -layout(set = 0, binding = 8) uniform texture3D aniso_neg_texture; -#endif // MODE ANISOTROPIC - #endif // MODE_SECOND_BOUNCE #ifndef MODE_DYNAMIC @@ -110,13 +105,6 @@ layout(set = 0, binding = 10) uniform sampler texture_sampler; layout(rgba8, set = 0, binding = 5) uniform restrict writeonly image3D color_tex; -#ifdef MODE_ANISOTROPIC - -layout(r16ui, set = 0, binding = 6) uniform restrict writeonly uimage3D aniso_pos_tex; -layout(r16ui, set = 0, binding = 7) uniform restrict writeonly uimage3D aniso_neg_tex; - -#endif - #endif #ifdef MODE_DYNAMIC @@ -170,13 +158,6 @@ layout(r32f, set = 0, binding = 8) uniform restrict writeonly image2D depth; layout(rgba8, set = 0, binding = 11) uniform restrict image3D color_texture; -#ifdef MODE_ANISOTROPIC - -layout(r16ui, set = 0, binding = 12) uniform restrict writeonly uimage3D aniso_pos_texture; -layout(r16ui, set = 0, binding = 13) uniform restrict writeonly uimage3D aniso_neg_texture; - -#endif // MODE ANISOTROPIC - #endif //MODE_DYNAMIC_SHRINK_PLOT #endif // MODE_DYNAMIC_SHRINK @@ -374,12 +355,7 @@ void main() { vec3 emission = vec3(uvec3(cell_data.data[cell_index].emission & 0x1ff, (cell_data.data[cell_index].emission >> 9) & 0x1ff, (cell_data.data[cell_index].emission >> 18) & 0x1ff)) * pow(2.0, float(cell_data.data[cell_index].emission >> 27) - 15.0 - 9.0); vec3 normal = unpackSnorm4x8(cell_data.data[cell_index].normal).xyz; -#ifdef MODE_ANISOTROPIC - vec3 accum[6] = vec3[](vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0)); - const vec3 accum_dirs[6] = vec3[](vec3(1.0, 0.0, 0.0), vec3(-1.0, 0.0, 0.0), vec3(0.0, 1.0, 0.0), vec3(0.0, -1.0, 0.0), vec3(0.0, 0.0, 1.0), vec3(0.0, 0.0, -1.0)); -#else vec3 accum = vec3(0.0); -#endif for (uint i = 0; i < params.light_count; i++) { vec3 light; @@ -390,38 +366,16 @@ void main() { light *= albedo.rgb; -#ifdef MODE_ANISOTROPIC - for (uint j = 0; j < 6; j++) { - accum[j] += max(0.0, dot(accum_dirs[j], -light_dir)) * light; - } -#else if (length(normal) > 0.2) { accum += max(0.0, dot(normal, -light_dir)) * light; } else { //all directions accum += light; } -#endif } -#ifdef MODE_ANISOTROPIC - - for (uint i = 0; i < 6; i++) { - vec3 light = accum[i]; - if (length(normal) > 0.2) { - light += max(0.0, dot(accum_dirs[i], -normal)) * emission; - } else { - light += emission; - } - - outputs.data[cell_index * 6 + i] = vec4(light, 0.0); - } - -#else outputs.data[cell_index] = vec4(accum + emission, 0.0); -#endif - #endif //MODE_COMPUTE_LIGHT /////////////////SECOND BOUNCE/////////////////////////////// @@ -431,32 +385,8 @@ void main() { ivec3 ipos = ivec3(posu); vec4 normal = unpackSnorm4x8(cell_data.data[cell_index].normal); -#ifdef MODE_ANISOTROPIC - vec3 accum[6]; - const vec3 accum_dirs[6] = vec3[](vec3(1.0, 0.0, 0.0), vec3(-1.0, 0.0, 0.0), vec3(0.0, 1.0, 0.0), vec3(0.0, -1.0, 0.0), vec3(0.0, 0.0, 1.0), vec3(0.0, 0.0, -1.0)); - - /*vec3 src_color = texelFetch(sampler3D(color_texture,texture_sampler),ipos,0).rgb * params.dynamic_range; - vec3 src_aniso_pos = texelFetch(sampler3D(aniso_pos_texture,texture_sampler),ipos,0).rgb; - vec3 src_anisp_neg = texelFetch(sampler3D(anisp_neg_texture,texture_sampler),ipos,0).rgb; - accum[0]=src_col * src_aniso_pos.x; - accum[1]=src_col * src_aniso_neg.x; - accum[2]=src_col * src_aniso_pos.y; - accum[3]=src_col * src_aniso_neg.y; - accum[4]=src_col * src_aniso_pos.z; - accum[5]=src_col * src_aniso_neg.z;*/ - - accum[0] = outputs.data[cell_index * 6 + 0].rgb; - accum[1] = outputs.data[cell_index * 6 + 1].rgb; - accum[2] = outputs.data[cell_index * 6 + 2].rgb; - accum[3] = outputs.data[cell_index * 6 + 3].rgb; - accum[4] = outputs.data[cell_index * 6 + 4].rgb; - accum[5] = outputs.data[cell_index * 6 + 5].rgb; - -#else vec3 accum = outputs.data[cell_index].rgb; -#endif - if (length(normal.xyz) > 0.2) { vec3 v0 = abs(normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); vec3 tangent = normalize(cross(v0, normal.xyz)); @@ -484,9 +414,6 @@ void main() { float max_distance = length(vec3(params.limits)); vec3 cell_size = 1.0 / vec3(params.limits); -#ifdef MODE_ANISOTROPIC - vec3 aniso_normal = mix(direction, normal.xyz, params.aniso_strength); -#endif while (dist < max_distance && color.a < 0.95) { float diameter = max(1.0, 2.0 * tan_half_angle * dist); vec3 uvw_pos = (pos + dist * direction) * cell_size; @@ -498,42 +425,18 @@ void main() { float log2_diameter = log2(diameter); vec4 scolor = textureLod(sampler3D(color_texture, texture_sampler), uvw_pos, log2_diameter); -#ifdef MODE_ANISOTROPIC - - vec3 aniso_neg = textureLod(sampler3D(aniso_neg_texture, texture_sampler), uvw_pos, log2_diameter).rgb; - vec3 aniso_pos = textureLod(sampler3D(aniso_pos_texture, texture_sampler), uvw_pos, log2_diameter).rgb; - - scolor.rgb *= dot(max(vec3(0.0), (aniso_normal * aniso_pos)), vec3(1.0)) + dot(max(vec3(0.0), (-aniso_normal * aniso_neg)), vec3(1.0)); -#endif float a = (1.0 - color.a); color += a * scolor; dist += half_diameter; } } color *= cone_weights[i] * vec4(albedo.rgb, 1.0) * params.dynamic_range; //restore range -#ifdef MODE_ANISOTROPIC - for (uint j = 0; j < 6; j++) { - accum[j] += max(0.0, dot(accum_dirs[j], direction)) * color.rgb; - } -#else accum += color.rgb; -#endif } } -#ifdef MODE_ANISOTROPIC - - outputs.data[cell_index * 6 + 0] = vec4(accum[0], 0.0); - outputs.data[cell_index * 6 + 1] = vec4(accum[1], 0.0); - outputs.data[cell_index * 6 + 2] = vec4(accum[2], 0.0); - outputs.data[cell_index * 6 + 3] = vec4(accum[3], 0.0); - outputs.data[cell_index * 6 + 4] = vec4(accum[4], 0.0); - outputs.data[cell_index * 6 + 5] = vec4(accum[5], 0.0); -#else outputs.data[cell_index] = vec4(accum, 0.0); -#endif - #endif // MODE_SECOND_BOUNCE /////////////////UPDATE MIPMAPS/////////////////////////////// @@ -541,45 +444,20 @@ void main() { #ifdef MODE_UPDATE_MIPMAPS { -#ifdef MODE_ANISOTROPIC - vec3 light_accum[6] = vec3[](vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0)); -#else vec3 light_accum = vec3(0.0); -#endif float count = 0.0; for (uint i = 0; i < 8; i++) { uint child_index = cell_children.data[cell_index].children[i]; if (child_index == NO_CHILDREN) { continue; } -#ifdef MODE_ANISOTROPIC - light_accum[0] += outputs.data[child_index * 6 + 0].rgb; - light_accum[1] += outputs.data[child_index * 6 + 1].rgb; - light_accum[2] += outputs.data[child_index * 6 + 2].rgb; - light_accum[3] += outputs.data[child_index * 6 + 3].rgb; - light_accum[4] += outputs.data[child_index * 6 + 4].rgb; - light_accum[5] += outputs.data[child_index * 6 + 5].rgb; - -#else light_accum += outputs.data[child_index].rgb; -#endif - count += 1.0; } float divisor = mix(8.0, count, params.propagation); -#ifdef MODE_ANISOTROPIC - outputs.data[cell_index * 6 + 0] = vec4(light_accum[0] / divisor, 0.0); - outputs.data[cell_index * 6 + 1] = vec4(light_accum[1] / divisor, 0.0); - outputs.data[cell_index * 6 + 2] = vec4(light_accum[2] / divisor, 0.0); - outputs.data[cell_index * 6 + 3] = vec4(light_accum[3] / divisor, 0.0); - outputs.data[cell_index * 6 + 4] = vec4(light_accum[4] / divisor, 0.0); - outputs.data[cell_index * 6 + 5] = vec4(light_accum[5] / divisor, 0.0); - -#else outputs.data[cell_index] = vec4(light_accum / divisor, 0.0); -#endif } #endif @@ -587,40 +465,7 @@ void main() { #ifdef MODE_WRITE_TEXTURE { -#ifdef MODE_ANISOTROPIC - vec3 accum_total = vec3(0.0); - accum_total += outputs.data[cell_index * 6 + 0].rgb; - accum_total += outputs.data[cell_index * 6 + 1].rgb; - accum_total += outputs.data[cell_index * 6 + 2].rgb; - accum_total += outputs.data[cell_index * 6 + 3].rgb; - accum_total += outputs.data[cell_index * 6 + 4].rgb; - accum_total += outputs.data[cell_index * 6 + 5].rgb; - - float accum_total_energy = max(dot(accum_total, GREY_VEC), 0.00001); - vec3 iso_positive = vec3(dot(outputs.data[cell_index * 6 + 0].rgb, GREY_VEC), dot(outputs.data[cell_index * 6 + 2].rgb, GREY_VEC), dot(outputs.data[cell_index * 6 + 4].rgb, GREY_VEC)) / vec3(accum_total_energy); - vec3 iso_negative = vec3(dot(outputs.data[cell_index * 6 + 1].rgb, GREY_VEC), dot(outputs.data[cell_index * 6 + 3].rgb, GREY_VEC), dot(outputs.data[cell_index * 6 + 5].rgb, GREY_VEC)) / vec3(accum_total_energy); - - { - uint aniso_pos = uint(clamp(iso_positive.b * 31.0, 0.0, 31.0)); - aniso_pos |= uint(clamp(iso_positive.g * 63.0, 0.0, 63.0)) << 5; - aniso_pos |= uint(clamp(iso_positive.r * 31.0, 0.0, 31.0)) << 11; - imageStore(aniso_pos_tex, ivec3(posu), uvec4(aniso_pos)); - } - - { - uint aniso_neg = uint(clamp(iso_negative.b * 31.0, 0.0, 31.0)); - aniso_neg |= uint(clamp(iso_negative.g * 63.0, 0.0, 63.0)) << 5; - aniso_neg |= uint(clamp(iso_negative.r * 31.0, 0.0, 31.0)) << 11; - imageStore(aniso_neg_tex, ivec3(posu), uvec4(aniso_neg)); - } - - imageStore(color_tex, ivec3(posu), vec4(accum_total / params.dynamic_range, albedo.a)); - -#else - imageStore(color_tex, ivec3(posu), vec4(outputs.data[cell_index].rgb / params.dynamic_range, albedo.a)); - -#endif } #endif @@ -763,13 +608,6 @@ void main() { color.rgb /= params.dynamic_range; imageStore(color_texture, pos3d, color); //imageStore(color_texture,pos3d,vec4(1,1,1,1)); - -#ifdef MODE_ANISOTROPIC - //do not care about anisotropy for dynamic objects, just store full lit in all directions - imageStore(aniso_pos_texture, pos3d, uvec4(0xFFFF)); - imageStore(aniso_neg_texture, pos3d, uvec4(0xFFFF)); - -#endif // ANISOTROPIC } #endif // MODE_DYNAMIC_SHRINK_PLOT } diff --git a/servers/rendering/renderer_rd/shaders/voxel_gi_debug.glsl b/servers/rendering/renderer_rd/shaders/voxel_gi_debug.glsl index 7d4d72967a..281c496df3 100644 --- a/servers/rendering/renderer_rd/shaders/voxel_gi_debug.glsl +++ b/servers/rendering/renderer_rd/shaders/voxel_gi_debug.glsl @@ -20,11 +20,6 @@ layout(set = 0, binding = 2) uniform texture3D color_tex; layout(set = 0, binding = 3) uniform sampler tex_sampler; -#ifdef USE_ANISOTROPY -layout(set = 0, binding = 4) uniform texture3D aniso_pos_tex; -layout(set = 0, binding = 5) uniform texture3D aniso_neg_tex; -#endif - layout(push_constant, binding = 0, std430) uniform Params { mat4 projection; uint cell_offset; |