diff options
Diffstat (limited to 'servers/rendering')
33 files changed, 235 insertions, 187 deletions
diff --git a/servers/rendering/dummy/storage/texture_storage.h b/servers/rendering/dummy/storage/texture_storage.h index 62e1eb326d..fd36e7ac10 100644 --- a/servers/rendering/dummy/storage/texture_storage.h +++ b/servers/rendering/dummy/storage/texture_storage.h @@ -154,6 +154,7 @@ public: virtual RID decal_instance_create(RID p_decal) override { return RID(); } virtual void decal_instance_free(RID p_decal_instance) override {} virtual void decal_instance_set_transform(RID p_decal, const Transform3D &p_transform) override {} + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) override {} /* RENDER TARGET */ diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.cpp b/servers/rendering/renderer_rd/cluster_builder_rd.cpp index 73a0c652a4..959a752fba 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.cpp +++ b/servers/rendering/renderer_rd/cluster_builder_rd.cpp @@ -74,7 +74,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { cluster_debug.shader_pipeline = RD::get_singleton()->compute_pipeline_create(cluster_debug.shader); } - { // SPHERE + { // Sphere mesh data. static const uint32_t icosphere_vertex_count = 42; static const float icosphere_vertices[icosphere_vertex_count * 3] = { 0, 0, -1, 0.7236073, -0.5257253, -0.4472195, -0.276388, -0.8506492, -0.4472199, -0.8944262, 0, -0.4472156, -0.276388, 0.8506492, -0.4472199, 0.7236073, 0.5257253, -0.4472195, 0.276388, -0.8506492, 0.4472199, -0.7236073, -0.5257253, 0.4472195, -0.7236073, 0.5257253, 0.4472195, 0.276388, 0.8506492, 0.4472199, 0.8944262, 0, 0.4472156, 0, 0, 1, -0.1624555, -0.4999952, -0.8506544, 0.4253227, -0.3090114, -0.8506542, 0.2628688, -0.8090116, -0.5257377, 0.8506479, 0, -0.5257359, 0.4253227, 0.3090114, -0.8506542, -0.5257298, 0, -0.8506517, -0.6881894, -0.4999969, -0.5257362, -0.1624555, 0.4999952, -0.8506544, -0.6881894, 0.4999969, -0.5257362, 0.2628688, 0.8090116, -0.5257377, 0.9510579, -0.3090126, 0, 0.9510579, 0.3090126, 0, 0, -1, 0, 0.5877856, -0.8090167, 0, -0.9510579, -0.3090126, 0, -0.5877856, -0.8090167, 0, -0.5877856, 0.8090167, 0, -0.9510579, 0.3090126, 0, 0.5877856, 0.8090167, 0, 0, 1, 0, 0.6881894, -0.4999969, 0.5257362, -0.2628688, -0.8090116, 0.5257377, -0.8506479, 0, 0.5257359, -0.2628688, 0.8090116, 0.5257377, 0.6881894, 0.4999969, 0.5257362, 0.1624555, -0.4999952, 0.8506544, 0.5257298, 0, 0.8506517, -0.4253227, -0.3090114, 0.8506542, -0.4253227, 0.3090114, 0.8506542, 0.1624555, 0.4999952, 0.8506544 @@ -118,7 +118,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { sphere_overfit = 1.0 / min_d; } - { // CONE + { // Cone mesh data. static const uint32_t cone_vertex_count = 99; static const float cone_vertices[cone_vertex_count * 3] = { 0, 1, -1, 0.1950903, 0.9807853, -1, 0.3826835, 0.9238795, -1, 0.5555703, 0.8314696, -1, 0.7071068, 0.7071068, -1, 0.8314697, 0.5555702, -1, 0.9238795, 0.3826834, -1, 0.9807853, 0.1950903, -1, 1, 0, -1, 0.9807853, -0.1950902, -1, 0.9238796, -0.3826833, -1, 0.8314697, -0.5555702, -1, 0.7071068, -0.7071068, -1, 0.5555702, -0.8314697, -1, 0.3826833, -0.9238796, -1, 0.1950901, -0.9807853, -1, -3.25841e-7, -1, -1, -0.1950907, -0.9807852, -1, -0.3826839, -0.9238793, -1, -0.5555707, -0.8314693, -1, -0.7071073, -0.7071063, -1, -0.83147, -0.5555697, -1, -0.9238799, -0.3826827, -1, 0, 0, 0, -0.9807854, -0.1950894, -1, -1, 9.65599e-7, -1, -0.9807851, 0.1950913, -1, -0.9238791, 0.3826845, -1, -0.8314689, 0.5555713, -1, -0.7071059, 0.7071077, -1, -0.5555691, 0.8314704, -1, -0.3826821, 0.9238801, -1, -0.1950888, 0.9807856, -1 @@ -172,7 +172,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { cone_overfit = 1.0 / min_d; } - { // BOX + { // Box mesh data. static const uint32_t box_vertex_count = 8; static const float box_vertices[box_vertex_count * 3] = { -1, -1, -1, -1, -1, 1, -1, 1, -1, -1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, -1, 1, 1, 1 @@ -219,8 +219,9 @@ ClusterBuilderSharedDataRD::~ClusterBuilderSharedDataRD() { void ClusterBuilderRD::_clear() { if (cluster_buffer.is_null()) { - return; //nothing to clear + return; } + RD::get_singleton()->free(cluster_buffer); RD::get_singleton()->free(cluster_render_buffer); RD::get_singleton()->free(element_buffer); @@ -254,7 +255,7 @@ void ClusterBuilderRD::setup(Size2i p_screen_size, uint32_t p_max_elements, RID cluster_screen_size.height = (p_screen_size.height - 1) / cluster_size + 1; max_elements_by_type = p_max_elements; - if (max_elements_by_type % 32) { //need to be 32 aligned + if (max_elements_by_type % 32) { // Needs to be aligned to 32. max_elements_by_type += 32 - (max_elements_by_type % 32); } @@ -264,7 +265,8 @@ void ClusterBuilderRD::setup(Size2i p_screen_size, uint32_t p_max_elements, RID uint32_t element_tag_bits_size = render_element_max / 32; uint32_t element_tag_depth_bits_size = render_element_max; - cluster_render_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (element_tag_bits_size + element_tag_depth_bits_size) * 4; // tag bits (element was used) and tag depth (depth range in which it was used) + + cluster_render_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (element_tag_bits_size + element_tag_depth_bits_size) * 4; // Tag bits (element was used) and tag depth (depth range in which it was used). cluster_render_buffer = RD::get_singleton()->storage_buffer_create(cluster_render_buffer_size); cluster_buffer = RD::get_singleton()->storage_buffer_create(cluster_buffer_size); @@ -379,9 +381,9 @@ void ClusterBuilderRD::begin(const Transform3D &p_view_transform, const Projecti projection = p_cam_projection; z_near = projection.get_z_near(); z_far = projection.get_z_far(); - orthogonal = p_cam_projection.is_orthogonal(); + camera_orthogonal = p_cam_projection.is_orthogonal(); adjusted_projection = projection; - if (!orthogonal) { + if (!camera_orthogonal) { adjusted_projection.adjust_perspective_znear(0.0001); } @@ -390,7 +392,7 @@ void ClusterBuilderRD::begin(const Transform3D &p_view_transform, const Projecti projection = correction * projection; adjusted_projection = correction * adjusted_projection; - //reset counts + // Reset counts. render_element_count = 0; for (uint32_t i = 0; i < ELEMENT_TYPE_MAX; i++) { cluster_count_by_type[i] = 0; @@ -402,14 +404,14 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->draw_command_begin_label("Bake Light Cluster"); - //clear cluster buffer + // Clear cluster buffer. RD::get_singleton()->buffer_clear(cluster_buffer, 0, cluster_buffer_size, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); if (render_element_count > 0) { - //clear render buffer + // Clear render buffer. RD::get_singleton()->buffer_clear(cluster_render_buffer, 0, cluster_render_buffer_size, RD::BARRIER_MASK_RASTER); - { //fill state uniform + { // Fill state uniform. StateUniform state; @@ -425,13 +427,13 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->buffer_update(state_uniform, 0, sizeof(StateUniform), &state, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } - //update instances + // Update instances. RD::get_singleton()->buffer_update(element_buffer, 0, sizeof(RenderElementData) * render_element_count, render_elements, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); RENDER_TIMESTAMP("Render 3D Cluster Elements"); - //render elements + // Render elements. { RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); ClusterBuilderSharedDataRD::ClusterRender::PushConstant push_constant = {}; @@ -447,8 +449,16 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->sphere_index_array); } break; case ELEMENT_TYPE_SPOT_LIGHT: { - RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->cone_vertex_array); - RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->cone_index_array); + // If the spot angle is above a certain threshold, use a sphere instead of a cone for building the clusters + // since the cone gets too flat/large (spot angle close to 90 degrees) or + // can't even cover the affected area of the light (spot angle above 90 degrees). + if (render_elements[i].has_wide_spot_angle) { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->sphere_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->sphere_index_array); + } else { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->cone_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->cone_index_array); + } } break; case ELEMENT_TYPE_DECAL: case ELEMENT_TYPE_REFLECTION_PROBE: { @@ -465,7 +475,7 @@ void ClusterBuilderRD::bake_cluster() { } RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_COMPUTE); } - //store elements + // Store elements. RENDER_TIMESTAMP("Pack 3D Cluster Elements"); { @@ -509,7 +519,7 @@ void ClusterBuilderRD::debug(ElementType p_element) { push_constant.cluster_screen_size[1] = cluster_screen_size.y; push_constant.cluster_shift = get_shift_from_power_of_2(cluster_size); push_constant.cluster_type = p_element; - push_constant.orthogonal = orthogonal; + push_constant.orthogonal = camera_orthogonal; push_constant.z_far = z_far; push_constant.z_near = z_near; push_constant.max_cluster_element_count_div_32 = max_elements_by_type / 32; diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.h b/servers/rendering/renderer_rd/cluster_builder_rd.h index 0b20a5d7ee..a13e6c8172 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.h +++ b/servers/rendering/renderer_rd/cluster_builder_rd.h @@ -43,13 +43,13 @@ class ClusterBuilderSharedDataRD { RID sphere_vertex_array; RID sphere_index_buffer; RID sphere_index_array; - float sphere_overfit = 0.0; //because an icosphere is not a perfect sphere, we need to enlarge it to cover the sphere area + float sphere_overfit = 0.0; // Because an icosphere is not a perfect sphere, we need to enlarge it to cover the sphere area. RID cone_vertex_buffer; RID cone_vertex_array; RID cone_index_buffer; RID cone_index_array; - float cone_overfit = 0.0; //because an cone mesh is not a perfect sphere, we need to enlarge it to cover the actual cone area + float cone_overfit = 0.0; // Because an cone mesh is not a perfect cone, we need to enlarge it to cover the actual cone area. RID box_vertex_buffer; RID box_vertex_array; @@ -73,6 +73,7 @@ class ClusterBuilderSharedDataRD { ClusterRenderShaderRD cluster_render_shader; RID shader_version; RID shader; + enum PipelineVersion { PIPELINE_NORMAL, PIPELINE_MSAA, @@ -85,10 +86,11 @@ class ClusterBuilderSharedDataRD { struct ClusterStore { struct PushConstant { uint32_t cluster_render_data_size; // how much data for a single cluster takes - uint32_t max_render_element_count_div_32; //divided by 32 + uint32_t max_render_element_count_div_32; // divided by 32 uint32_t cluster_screen_size[2]; - uint32_t render_element_count_div_32; //divided by 32 - uint32_t max_cluster_element_count_div_32; //divided by 32 + uint32_t render_element_count_div_32; // divided by 32 + uint32_t max_cluster_element_count_div_32; // divided by 32 + uint32_t pad1; uint32_t pad2; }; @@ -111,6 +113,7 @@ class ClusterBuilderSharedDataRD { uint32_t orthogonal; uint32_t max_cluster_element_count_div_32; + uint32_t pad1; uint32_t pad2; }; @@ -128,6 +131,8 @@ public: class ClusterBuilderRD { public: + static constexpr float WIDE_SPOT_ANGLE_THRESHOLD_DEG = 60.0f; + enum LightType { LIGHT_TYPE_OMNI, LIGHT_TYPE_SPOT @@ -144,21 +149,20 @@ public: ELEMENT_TYPE_DECAL, ELEMENT_TYPE_REFLECTION_PROBE, ELEMENT_TYPE_MAX, - }; private: ClusterBuilderSharedDataRD *shared = nullptr; struct RenderElementData { - uint32_t type; //0-4 + uint32_t type; // 0-4 uint32_t touches_near; uint32_t touches_far; uint32_t original_index; - float transform_inv[12]; //transposed transform for less space + float transform_inv[12]; // Transposed transform for less space. float scale[3]; - uint32_t pad; - }; + uint32_t has_wide_spot_angle; + }; // Keep aligned to 32 bytes. uint32_t cluster_count_by_type[ELEMENT_TYPE_MAX] = {}; uint32_t max_elements_by_type = 0; @@ -172,7 +176,7 @@ private: Projection projection; float z_far = 0; float z_near = 0; - bool orthogonal = false; + bool camera_orthogonal = false; enum Divisor { DIVISOR_1, @@ -188,26 +192,27 @@ private: Size2i cluster_screen_size; RID framebuffer; - RID cluster_render_buffer; //used for creating - RID cluster_buffer; //used for rendering - RID element_buffer; //used for storing, to hint element touches far plane or near plane + RID cluster_render_buffer; // Used for creating. + RID cluster_buffer; // Used for rendering. + RID element_buffer; // Used for storing, to hint element touches far plane or near plane. uint32_t cluster_render_buffer_size = 0; uint32_t cluster_buffer_size = 0; RID cluster_render_uniform_set; RID cluster_store_uniform_set; - //persistent data + // Persistent data. void _clear(); struct StateUniform { float projection[16]; float inv_z_far; - uint32_t screen_to_clusters_shift; // shift to obtain coordinates in block indices - uint32_t cluster_screen_width; // - uint32_t cluster_data_size; // how much data for a single cluster takes + uint32_t screen_to_clusters_shift; // Shift to obtain coordinates in block indices. + uint32_t cluster_screen_width; + uint32_t cluster_data_size; // How much data is needed for a single cluster. uint32_t cluster_depth_offset; + uint32_t pad0; uint32_t pad1; uint32_t pad2; @@ -224,10 +229,10 @@ public: _FORCE_INLINE_ void add_light(LightType p_type, const Transform3D &p_transform, float p_radius, float p_spot_aperture) { if (p_type == LIGHT_TYPE_OMNI && cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } if (p_type == LIGHT_TYPE_SPOT && cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } RenderElementData &e = render_elements[render_element_count]; @@ -242,15 +247,14 @@ public: radius *= p_radius; if (p_type == LIGHT_TYPE_OMNI) { - radius *= shared->sphere_overfit; // overfit icosphere + radius *= shared->sphere_overfit; // Overfit icosphere. - //omni float depth = -xform.origin.z; - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = (depth - radius) < z_near; } else { - //contains camera inside light - float radius2 = radius * shared->sphere_overfit; // overfit again for outer size (camera may be outside actual sphere but behind an icosphere vertex) + // Contains camera inside light. + float radius2 = radius * shared->sphere_overfit; // Overfit again for outer size (camera may be outside actual sphere but behind an icosphere vertex) e.touches_near = xform.origin.length_squared() < radius2 * radius2; } @@ -265,12 +269,11 @@ public: cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT]++; - } else { - //spot - radius *= shared->cone_overfit; // overfit icosphere + } else /*LIGHT_TYPE_SPOT */ { + radius *= shared->cone_overfit; // Overfit icosphere real_t len = Math::tan(Math::deg_to_rad(p_spot_aperture)) * radius; - //approximate, probably better to use a cone support function + // Approximate, probably better to use a cone support function. float max_d = -1e20; float min_d = 1e20; #define CONE_MINMAX(m_x, m_y) \ @@ -285,14 +288,13 @@ public: CONE_MINMAX(-1, -1); CONE_MINMAX(1, -1); - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = min_d < z_near; } else { - //contains camera inside light Plane base_plane(-xform.basis.get_column(Vector3::AXIS_Z), xform.origin); float dist = base_plane.distance_to(Vector3()); if (dist >= 0 && dist < radius) { - //inside, check angle + // Contains camera inside light, check angle. float angle = Math::rad_to_deg(Math::acos((-xform.origin.normalized()).dot(-xform.basis.get_column(Vector3::AXIS_Z)))); e.touches_near = angle < p_spot_aperture * 1.05; //overfit aperture a little due to cone overfit } else { @@ -302,12 +304,23 @@ public: e.touches_far = max_d > z_far; - e.scale[0] = len * shared->cone_overfit; - e.scale[1] = len * shared->cone_overfit; - e.scale[2] = radius; + // If the spot angle is above the threshold, use a sphere instead of a cone for building the clusters + // since the cone gets too flat/large (spot angle close to 90 degrees) or + // can't even cover the affected area of the light (spot angle above 90 degrees). + if (p_spot_aperture > WIDE_SPOT_ANGLE_THRESHOLD_DEG) { + e.scale[0] = radius; + e.scale[1] = radius; + e.scale[2] = radius; + e.has_wide_spot_angle = true; + } else { + e.scale[0] = len * shared->cone_overfit; + e.scale[1] = len * shared->cone_overfit; + e.scale[2] = radius; + e.has_wide_spot_angle = false; + } e.type = ELEMENT_TYPE_SPOT_LIGHT; - e.original_index = cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]; //use omni since they share index + e.original_index = cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]; // Use omni light since they share index. RendererRD::MaterialStorage::store_transform_transposed_3x4(xform, e.transform_inv); @@ -319,16 +332,16 @@ public: _FORCE_INLINE_ void add_box(BoxType p_box_type, const Transform3D &p_transform, const Vector3 &p_half_extents) { if (p_box_type == BOX_TYPE_DECAL && cluster_count_by_type[ELEMENT_TYPE_DECAL] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } if (p_box_type == BOX_TYPE_REFLECTION_PROBE && cluster_count_by_type[ELEMENT_TYPE_REFLECTION_PROBE] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } RenderElementData &e = render_elements[render_element_count]; Transform3D xform = view_xform * p_transform; - //extract scale and scale the matrix by it, makes things simpler + // Extract scale and scale the matrix by it, makes things simpler. Vector3 scale = p_half_extents; for (uint32_t i = 0; i < 3; i++) { float s = xform.basis.rows[i].length(); @@ -339,10 +352,10 @@ public: float box_depth = Math::abs(xform.basis.xform_inv(Vector3(0, 0, -1)).dot(scale)); float depth = -xform.origin.z; - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = depth - box_depth < z_near; } else { - //contains camera inside box + // Contains camera inside box. Vector3 inside = xform.xform_inv(Vector3(0, 0, 0)).abs(); e.touches_near = inside.x < scale.x && inside.y < scale.y && inside.z < scale.z; } diff --git a/servers/rendering/renderer_rd/environment/gi.cpp b/servers/rendering/renderer_rd/environment/gi.cpp index a52716cd78..08133bf8d6 100644 --- a/servers/rendering/renderer_rd/environment/gi.cpp +++ b/servers/rendering/renderer_rd/environment/gi.cpp @@ -540,9 +540,7 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re occlusion_texture = RD::get_singleton()->texture_create_shared(tv, occlusion_data); } - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; - + for (SDFGI::Cascade &cascade : cascades) { /* 3D Textures */ cascade.sdf_tex = RD::get_singleton()->texture_create(tf_sdf, RD::TextureView()); @@ -743,9 +741,7 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re } //direct light - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; - + for (SDFGI::Cascade &cascade : cascades) { Vector<RD::Uniform> uniforms; { RD::Uniform u; @@ -1134,8 +1130,7 @@ void GI::SDFGI::free_data() { } GI::SDFGI::~SDFGI() { - for (uint32_t i = 0; i < cascades.size(); i++) { - const SDFGI::Cascade &c = cascades[i]; + for (const SDFGI::Cascade &c : cascades) { RD::get_singleton()->free(c.light_data); RD::get_singleton()->free(c.light_aniso_0_tex); RD::get_singleton()->free(c.light_aniso_1_tex); @@ -1198,8 +1193,7 @@ void GI::SDFGI::update(RID p_env, const Vector3 &p_world_position) { int32_t drag_margin = (cascade_size / SDFGI::PROBE_DIVISOR) / 2; - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; + for (SDFGI::Cascade &cascade : cascades) { cascade.dirty_regions = Vector3i(); Vector3 probe_half_size = Vector3(1, 1, 1) * cascade.cell_size * float(cascade_size / SDFGI::PROBE_DIVISOR) * 0.5; diff --git a/servers/rendering/renderer_rd/environment/sky.cpp b/servers/rendering/renderer_rd/environment/sky.cpp index 7e02f98ce9..1dbad79477 100644 --- a/servers/rendering/renderer_rd/environment/sky.cpp +++ b/servers/rendering/renderer_rd/environment/sky.cpp @@ -347,7 +347,10 @@ void SkyRD::ReflectionData::update_reflection_data(int p_size, int p_mipmaps, bo tf.texture_type = RD::TEXTURE_TYPE_CUBE; tf.array_layers = 6; tf.mipmaps = p_low_quality ? 7 : mipmaps - 1; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + if (RendererSceneRenderRD::get_singleton()->_render_buffers_can_be_storage()) { + tf.usage_bits |= RD::TEXTURE_USAGE_STORAGE_BIT; + } downsampled_radiance_cubemap = RD::get_singleton()->texture_create(tf, RD::TextureView()); RD::get_singleton()->set_resource_name(downsampled_radiance_cubemap, "downsampled radiance cubemap"); @@ -1627,7 +1630,10 @@ void SkyRD::update_dirty_skys() { tf.mipmaps = mipmaps; tf.width = w; tf.height = h; - tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT; + if (RendererSceneRenderRD::get_singleton()->_render_buffers_can_be_storage()) { + tf.usage_bits |= RD::TEXTURE_USAGE_STORAGE_BIT; + } sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); 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 93c741fd34..ad63d3fd4b 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -1365,8 +1365,8 @@ void RenderForwardClustered::_pre_opaque_render(RenderDataRD *p_render_data, boo } //cube shadows are rendered in their own way - for (uint32_t i = 0; i < p_render_data->cube_shadows.size(); i++) { - _render_shadow_pass(p_render_data->render_shadows[p_render_data->cube_shadows[i]].light, p_render_data->shadow_atlas, p_render_data->render_shadows[p_render_data->cube_shadows[i]].pass, p_render_data->render_shadows[p_render_data->cube_shadows[i]].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); + for (const int &index : p_render_data->cube_shadows) { + _render_shadow_pass(p_render_data->render_shadows[index].light, p_render_data->shadow_atlas, p_render_data->render_shadows[index].pass, p_render_data->render_shadows[index].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); } if (p_render_data->directional_shadows.size()) { @@ -1480,7 +1480,7 @@ void RenderForwardClustered::_pre_opaque_render(RenderDataRD *p_render_data, boo uint32_t directional_light_count = 0; uint32_t positional_light_count = 0; light_storage->update_light_buffers(p_render_data, *p_render_data->lights, p_render_data->scene_data->cam_transform, p_render_data->shadow_atlas, using_shadows, directional_light_count, positional_light_count, p_render_data->directional_light_soft_shadows); - texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform.affine_inverse()); + texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform); p_render_data->directional_light_count = directional_light_count; @@ -2415,8 +2415,7 @@ void RenderForwardClustered::_render_shadow_process() { void RenderForwardClustered::_render_shadow_end(uint32_t p_barrier) { RD::get_singleton()->draw_command_begin_label("Shadow Render"); - for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { - SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + for (SceneState::ShadowPass &shadow_pass : scene_state.shadow_passes) { 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, 0, true, false, shadow_pass.rp_uniform_set, false, Vector2(), shadow_pass.lod_distance_multiplier, shadow_pass.screen_mesh_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); } @@ -3376,9 +3375,7 @@ int RenderForwardClustered::sdfgi_get_pending_region_count(const Ref<RenderScene Ref<RendererRD::GI::SDFGI> sdfgi = rb->get_custom_data(RB_SCOPE_SDFGI); int dirty_count = 0; - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - const RendererRD::GI::SDFGI::Cascade &c = sdfgi->cascades[i]; - + for (const RendererRD::GI::SDFGI::Cascade &c : sdfgi->cascades) { if (c.dirty_regions == RendererRD::GI::SDFGI::Cascade::DIRTY_ALL) { dirty_count++; } else { @@ -4007,11 +4004,11 @@ RenderForwardClustered::~RenderForwardClustered() { RSG::light_storage->directional_shadow_atlas_set_size(0); { - for (uint32_t i = 0; i < scene_state.uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.uniform_buffers[i]); + for (const RID &rid : scene_state.uniform_buffers) { + RD::get_singleton()->free(rid); } - for (uint32_t i = 0; i < scene_state.implementation_uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.implementation_uniform_buffers[i]); + for (const RID &rid : scene_state.implementation_uniform_buffers) { + RD::get_singleton()->free(rid); } RD::get_singleton()->free(scene_state.lightmap_buffer); RD::get_singleton()->free(scene_state.lightmap_capture_buffer); 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 816248567b..25204f1abf 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp @@ -579,8 +579,8 @@ void RenderForwardMobile::_pre_opaque_render(RenderDataRD *p_render_data) { } //cube shadows are rendered in their own way - for (uint32_t i = 0; i < p_render_data->cube_shadows.size(); i++) { - _render_shadow_pass(p_render_data->render_shadows[p_render_data->cube_shadows[i]].light, p_render_data->shadow_atlas, p_render_data->render_shadows[p_render_data->cube_shadows[i]].pass, p_render_data->render_shadows[p_render_data->cube_shadows[i]].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); + for (const int &index : p_render_data->cube_shadows) { + _render_shadow_pass(p_render_data->render_shadows[index].light, p_render_data->shadow_atlas, p_render_data->render_shadows[index].pass, p_render_data->render_shadows[index].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); } if (p_render_data->directional_shadows.size()) { @@ -632,7 +632,7 @@ void RenderForwardMobile::_pre_opaque_render(RenderDataRD *p_render_data) { uint32_t directional_light_count = 0; uint32_t positional_light_count = 0; light_storage->update_light_buffers(p_render_data, *p_render_data->lights, p_render_data->scene_data->cam_transform, p_render_data->shadow_atlas, using_shadows, directional_light_count, positional_light_count, p_render_data->directional_light_soft_shadows); - texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform.affine_inverse()); + texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform); p_render_data->directional_light_count = directional_light_count; } @@ -1340,8 +1340,7 @@ void RenderForwardMobile::_render_shadow_process() { void RenderForwardMobile::_render_shadow_end(uint32_t p_barrier) { RD::get_singleton()->draw_command_begin_label("Shadow Render"); - for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { - SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + for (SceneState::ShadowPass &shadow_pass : scene_state.shadow_passes) { 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.lod_distance_multiplier, shadow_pass.screen_mesh_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); } @@ -2810,8 +2809,8 @@ RenderForwardMobile::~RenderForwardMobile() { } { - for (uint32_t i = 0; i < scene_state.uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.uniform_buffers[i]); + for (const RID &rid : scene_state.uniform_buffers) { + RD::get_singleton()->free(rid); } RD::get_singleton()->free(scene_state.lightmap_buffer); RD::get_singleton()->free(scene_state.lightmap_capture_buffer); diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index 462b925134..638fe44266 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -455,7 +455,7 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend light_count++; - if (light_count == MAX_LIGHTS_PER_ITEM) { + if (light_count == MAX_LIGHTS_PER_ITEM - 1) { break; } } diff --git a/servers/rendering/renderer_rd/shader_rd.cpp b/servers/rendering/renderer_rd/shader_rd.cpp index 2d5263a3e2..533a912a34 100644 --- a/servers/rendering/renderer_rd/shader_rd.cpp +++ b/servers/rendering/renderer_rd/shader_rd.cpp @@ -165,8 +165,7 @@ void ShaderRD::_clear_version(Version *p_version) { } void ShaderRD::_build_variant_code(StringBuilder &builder, uint32_t p_variant, const Version *p_version, const StageTemplate &p_template) { - for (uint32_t i = 0; i < p_template.chunks.size(); i++) { - const StageTemplate::Chunk &chunk = p_template.chunks[i]; + for (const StageTemplate::Chunk &chunk : p_template.chunks) { switch (chunk.type) { case StageTemplate::Chunk::TYPE_VERSION_DEFINES: { builder.append("\n"); //make sure defines begin at newline diff --git a/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl b/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl index 134aae5ce7..b1ff46dd3b 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl @@ -56,7 +56,7 @@ vec4 screen_space_to_view_space_depth(vec4 p_depth) { float depth_linearize_mul = params.z_near; float depth_linearize_add = params.z_far; - // Optimised version of "-cameraClipNear / (cameraClipFar - projDepth * (cameraClipFar - cameraClipNear)) * cameraClipFar" + // Optimized version of "-cameraClipNear / (cameraClipFar - projDepth * (cameraClipFar - cameraClipNear)) * cameraClipFar" // Set your depth_linearize_mul and depth_linearize_add to: // depth_linearize_mul = ( cameraClipFar * cameraClipNear) / ( cameraClipFar - cameraClipNear ); diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao.glsl index 2a87e273bc..ffaa6872c9 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao.glsl @@ -221,7 +221,7 @@ void SSAOTap(const int p_quality_level, inout float r_obscurance_sum, inout floa // snap to pixel center (more correct obscurance math, avoids artifacts) sample_offset = round(sample_offset); - // calculate MIP based on the sample distance from the centre, similar to as described + // calculate MIP based on the sample distance from the center, similar to as described // in http://graphics.cs.williams.edu/papers/SAOHPG12/. float mip_level = (p_quality_level < SSAO_DEPTH_MIPS_ENABLE_AT_QUALITY_PRESET) ? (0) : (sample_pow_2_len + p_mip_offset); @@ -259,7 +259,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o // get this pixel's viewspace depth pix_z = valuesUL.y; - // get left right top bottom neighbouring pixels for edge detection (gets compiled out on quality_level == 0) + // get left right top bottom neighboring pixels for edge detection (gets compiled out on quality_level == 0) pix_left_z = valuesUL.x; pix_top_z = valuesUL.z; pix_right_z = valuesBR.z; @@ -304,7 +304,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o float obscurance_sum = 0.0; float weight_sum = 0.0; - // edge mask for between this and left/right/top/bottom neighbour pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) + // edge mask for between this and left/right/top/bottom neighbor pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) vec4 edgesLRTB = vec4(1.0, 1.0, 1.0, 1.0); // Move center pixel slightly towards camera to avoid imprecision artifacts due to using of 16bit depth buffer; a lot smaller offsets needed when using 32bit floats @@ -318,7 +318,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o if (!p_adaptive_base && (p_quality_level >= SSAO_DETAIL_AO_ENABLE_AT_QUALITY_PRESET)) { // disable in case of quality level 4 (reference) if (p_quality_level != 4) { - //approximate neighbouring pixels positions (actually just deltas or "positions - pix_center_pos" ) + //approximate neighboring pixels positions (actually just deltas or "positions - pix_center_pos" ) vec3 normalized_viewspace_dir = vec3(pix_center_pos.xy / pix_center_pos.zz, 1.0); vec3 pixel_left_delta = vec3(-pixel_size_at_center.x, 0.0, 0.0) + normalized_viewspace_dir * (pix_left_z - pix_center_pos.z); vec3 pixel_right_delta = vec3(+pixel_size_at_center.x, 0.0, 0.0) + normalized_viewspace_dir * (pix_right_z - pix_center_pos.z); diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl index 04f98964e8..d234ab4417 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl @@ -80,7 +80,7 @@ void main() { #ifdef PROCESS_MAPA vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; @@ -98,7 +98,7 @@ void main() { #ifdef PROCESS_MAPB vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl index f6a9a92fac..45cc62d361 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl @@ -60,8 +60,8 @@ void main() { int mx = int(pix_pos.x % 2); int my = int(pix_pos.y % 2); int index_center = mx + my * 2; // center index - int index_horizontal = (1 - mx) + my * 2; // neighbouring, horizontal - int index_vertical = mx + (1 - my) * 2; // neighbouring, vertical + int index_horizontal = (1 - mx) + my * 2; // neighboring, horizontal + int index_vertical = mx + (1 - my) * 2; // neighboring, vertical int index_diagonal = (1 - mx) + (1 - my) * 2; // diagonal vec2 center_val = texelFetch(source_texture, ivec3(pix_pos / uvec2(params.size_modifier), index_center), 0).xy; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil.glsl index 513791dfbf..de7b97953f 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil.glsl @@ -234,7 +234,7 @@ void SSILTap(const int p_quality_level, inout vec3 r_color_sum, inout float r_ob // snap to pixel center (more correct obscurance math, avoids artifacts) sample_offset = round(sample_offset); - // calculate MIP based on the sample distance from the centre, similar to as described + // calculate MIP based on the sample distance from the center, similar to as described // in http://graphics.cs.williams.edu/papers/SAOHPG12/. float mip_level = (p_quality_level < SSIL_DEPTH_MIPS_ENABLE_AT_QUALITY_PRESET) ? (0) : (sample_pow_2_len + p_mip_offset); @@ -272,7 +272,7 @@ void generate_SSIL(out vec3 r_color, out vec4 r_edges, out float r_obscurance, o // get this pixel's viewspace depth pix_z = valuesUL.y; - // get left right top bottom neighbouring pixels for edge detection (gets compiled out on quality_level == 0) + // get left right top bottom neighboring pixels for edge detection (gets compiled out on quality_level == 0) pix_left_z = valuesUL.x; pix_top_z = valuesUL.z; pix_right_z = valuesBR.z; @@ -318,7 +318,7 @@ void generate_SSIL(out vec3 r_color, out vec4 r_edges, out float r_obscurance, o float obscurance_sum = 0.0; float weight_sum = 0.0; - // edge mask for between this and left/right/top/bottom neighbour pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) + // edge mask for between this and left/right/top/bottom neighbor pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) vec4 edgesLRTB = vec4(1.0, 1.0, 1.0, 1.0); // Move center pixel slightly towards camera to avoid imprecision artifacts due to using of 16bit depth buffer; a lot smaller offsets needed when using 32bit floats diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl index 47c56571f6..f48e6c4341 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl @@ -124,14 +124,14 @@ void main() { vec2 uv = (vec2(gl_GlobalInvocationID.xy) + vec2(0.5, 0.5)) * params.half_screen_pixel_size; - vec4 centre = textureLod(source_ssil, uv, 0.0); + vec4 center = textureLod(source_ssil, uv, 0.0); vec4 value = textureLod(source_ssil, vec2(uv + vec2(-half_pixel.x * 3, -half_pixel.y)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(+half_pixel.x, -half_pixel.y * 3)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(-half_pixel.x, +half_pixel.y * 3)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(+half_pixel.x * 3, +half_pixel.y)), 0.0) * 0.2; - vec4 sampled = value + centre * 0.2; + vec4 sampled = value + center * 0.2; #else #ifdef MODE_SMART diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl index 6b6b02739d..193e3458ab 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl @@ -82,7 +82,7 @@ void main() { #ifdef PROCESS_MAPA vec2 uv = (vec2(ssC) + 0.5) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; @@ -100,7 +100,7 @@ void main() { #ifdef PROCESS_MAPB vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl index 9e86ac0cf0..ed85b8ee4c 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl @@ -62,8 +62,8 @@ void main() { int mx = int(pix_pos.x % 2); int my = int(pix_pos.y % 2); int index_center = mx + my * 2; // center index - int index_horizontal = (1 - mx) + my * 2; // neighbouring, horizontal - int index_vertical = mx + (1 - my) * 2; // neighbouring, vertical + int index_horizontal = (1 - mx) + my * 2; // neighboring, horizontal + int index_vertical = mx + (1 - my) * 2; // neighboring, vertical int index_diagonal = (1 - mx) + (1 - my) * 2; // diagonal vec4 color = texelFetch(source_texture, ivec3(pix_pos / uvec2(params.size_modifier), index_center), 0); diff --git a/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl b/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl index 9f7449b8aa..06709f65d3 100644 --- a/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl +++ b/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl @@ -24,7 +24,7 @@ struct ProcessVoxel { 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 + //total neighbors: 26 }; #ifdef MODE_PROCESS_STATIC @@ -443,10 +443,10 @@ void main() { imageStore(dst_aniso1, positioni, vec4(aniso1, 0.0, 0.0)); imageStore(dst_light, positioni, uvec4(light_total_rgbe)); - //also fill neighbours, so light interpolation during the indirect pass works + //also fill neighbors, so light interpolation during the indirect pass works - //recover the neighbour list from the leftover bits - uint neighbours = (voxel_albedo >> 21) | ((voxel_position >> 21) << 11) | ((process_voxels.data[voxel_index].light >> 30) << 22) | ((process_voxels.data[voxel_index].light_aniso >> 30) << 24); + //recover the neighbor list from the leftover bits + uint neighbors = (voxel_albedo >> 21) | ((voxel_position >> 21) << 11) | ((process_voxels.data[voxel_index].light >> 30) << 22) | ((process_voxels.data[voxel_index].light_aniso >> 30) << 24); const uint max_neighbours = 26; const ivec3 neighbour_positions[max_neighbours] = ivec3[]( @@ -478,7 +478,7 @@ void main() { ivec3(1, 1, 1)); for (uint i = 0; i < max_neighbours; i++) { - if (bool(neighbours & (1 << i))) { + if (bool(neighbors & (1 << i))) { ivec3 neighbour_pos = positioni + neighbour_positions[i]; imageStore(dst_light, neighbour_pos, uvec4(light_total_rgbe)); imageStore(dst_aniso0, neighbour_pos, aniso0); diff --git a/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl b/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl index bce98f4054..dd35ae3b73 100644 --- a/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl +++ b/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl @@ -102,10 +102,10 @@ dispatch_data; struct ProcessVoxel { 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 neighbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours - uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours - //total neighbours: 26 + 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 neighbors: 26 }; layout(set = 0, binding = 11, std430) restrict buffer writeonly ProcessVoxels { @@ -135,10 +135,10 @@ dispatch_data; struct ProcessVoxel { 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 neighbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours - uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours - //total neighbours: 26 + 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 neighbors: 26 }; layout(set = 0, binding = 6, std430) restrict buffer readonly ProcessVoxels { @@ -1016,14 +1016,14 @@ void main() { store_positions[index].albedo = rgb >> 1; //store as it comes (555) to avoid precision loss (and move away the alpha bit) store_positions[index].albedo |= (facing & 0x3F) << 15; // store facing in bits 15-21 - store_positions[index].albedo |= neighbour_bits << 21; //store lower 11 bits of neighbours with remaining albedo - store_positions[index].position |= (neighbour_bits >> 11) << 21; //store 11 bits more of neighbours with position + store_positions[index].albedo |= neighbour_bits << 21; //store lower 11 bits of neighbors with remaining albedo + store_positions[index].position |= (neighbour_bits >> 11) << 21; //store 11 bits more of neighbors with position store_positions[index].light = imageLoad(src_light, pos).r; store_positions[index].light_aniso = imageLoad(src_light_aniso, pos).r; - //add neighbours - store_positions[index].light |= (neighbour_bits >> 22) << 30; //store 2 bits more of neighbours with light - store_positions[index].light_aniso |= (neighbour_bits >> 24) << 30; //store 2 bits more of neighbours with aniso + //add neighbors + store_positions[index].light |= (neighbour_bits >> 22) << 30; //store 2 bits more of neighbors with light + store_positions[index].light_aniso |= (neighbour_bits >> 24) << 30; //store 2 bits more of neighbors with aniso } groupMemoryBarrier(); diff --git a/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl index d32e6d717f..91c2000c1a 100644 --- a/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl @@ -1305,24 +1305,26 @@ void fragment_shader(in SceneData scene_data) { } if (sc_use_forward_gi && bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_VOXEL_GI)) { // process voxel_gi_instances - uint index1 = instances.data[instance_index].gi_offset & 0xFFFF; - vec3 ref_vec = normalize(reflect(-view, normal)); - ref_vec = mix(ref_vec, normal, roughness * roughness); + // Make vertex orientation the world one, but still align to camera. + vec3 cam_pos = mat3(scene_data.inv_view_matrix) * vertex; + vec3 cam_normal = mat3(scene_data.inv_view_matrix) * normal; + vec3 ref_vec = mat3(scene_data.inv_view_matrix) * normalize(reflect(-view, normal)); + //find arbitrary tangent and bitangent, then build a matrix - 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)); - mat3 normal_mat = mat3(tangent, bitangent, normal); + vec3 v0 = abs(cam_normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); + vec3 tangent = normalize(cross(v0, cam_normal)); + vec3 bitangent = normalize(cross(tangent, cam_normal)); + mat3 normal_mat = mat3(tangent, bitangent, cam_normal); vec4 amb_accum = vec4(0.0); vec4 spec_accum = vec4(0.0); - voxel_gi_compute(index1, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); + voxel_gi_compute(index1, cam_pos, cam_normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); uint index2 = instances.data[instance_index].gi_offset >> 16; if (index2 != 0xFFFF) { - voxel_gi_compute(index2, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); + voxel_gi_compute(index2, cam_pos, cam_normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); } if (amb_accum.a > 0.0) { @@ -1351,8 +1353,8 @@ void fragment_shader(in SceneData scene_data) { #endif // USE_MULTIVIEW for (int i = 0; i < 4; i++) { - const vec2 neighbours[4] = vec2[](vec2(-1, 0), vec2(1, 0), vec2(0, -1), vec2(0, 1)); - vec2 neighbour_coord = base_coord + neighbours[i] * scene_data.screen_pixel_size; + const vec2 neighbors[4] = vec2[](vec2(-1, 0), vec2(1, 0), vec2(0, -1), vec2(0, 1)); + vec2 neighbour_coord = base_coord + neighbors[i] * scene_data.screen_pixel_size; #ifdef USE_MULTIVIEW float neighbour_ang = dot(normal, textureLod(sampler2DArray(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), vec3(neighbour_coord, ViewIndex), 0.0).xyz * 2.0 - 1.0); #else // USE_MULTIVIEW @@ -1798,7 +1800,7 @@ void fragment_shader(in SceneData scene_data) { shadow = float(shadow1 >> ((i - 4u) * 8u) & 0xFFu) / 255.0; } - shadow = shadow * directional_lights.data[i].shadow_opacity + 1.0 - directional_lights.data[i].shadow_opacity; + shadow = mix(1.0, shadow, directional_lights.data[i].shadow_opacity); #endif blur_shadow(shadow); @@ -2082,7 +2084,7 @@ void fragment_shader(in SceneData scene_data) { float sRed = floor((cRed / pow(2.0f, exps - B - N)) + 0.5f); float sGreen = floor((cGreen / pow(2.0f, exps - B - N)) + 0.5f); float sBlue = floor((cBlue / pow(2.0f, exps - B - N)) + 0.5f); - //store as 8985 to have 2 extra neighbour bits + //store as 8985 to have 2 extra neighbor bits uint light_rgbe = ((uint(sRed) & 0x1FFu) >> 1) | ((uint(sGreen) & 0x1FFu) << 8) | (((uint(sBlue) & 0x1FFu) >> 1) << 17) | ((uint(exps) & 0x1Fu) << 25); imageStore(emission_grid, grid_pos, uvec4(light_rgbe)); diff --git a/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl b/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl index 5e64d4e651..2966a2ff65 100644 --- a/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl +++ b/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl @@ -1515,6 +1515,8 @@ void main() { } else { shadow = float(shadow1 >> ((i - 4) * 8) & 0xFF) / 255.0; } + + shadow = mix(1.0, shadow, directional_lights.data[i].shadow_opacity); #endif blur_shadow(shadow); 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 b30b0c8169..9dda62c28d 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl @@ -794,8 +794,13 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v float light_length = length(light_rel_vec); float spot_attenuation = get_omni_attenuation(light_length, spot_lights.data[idx].inv_radius, spot_lights.data[idx].attenuation); vec3 spot_dir = spot_lights.data[idx].direction; - float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[idx].cone_angle); - float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[idx].cone_angle)); + + // This conversion to a highp float is crucial to prevent light leaking + // due to precision errors in the following calculations (cone angle is mediump). + highp float cone_angle = spot_lights.data[idx].cone_angle; + float scos = max(dot(-normalize(light_rel_vec), spot_dir), cone_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - cone_angle)); + spot_attenuation *= 1.0 - pow(spot_rim, spot_lights.data[idx].cone_attenuation); float light_attenuation = spot_attenuation; vec3 color = spot_lights.data[idx].color; diff --git a/servers/rendering/renderer_rd/storage_rd/light_storage.cpp b/servers/rendering/renderer_rd/storage_rd/light_storage.cpp index 673fc25595..33e35a7a64 100644 --- a/servers/rendering/renderer_rd/storage_rd/light_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/light_storage.cpp @@ -568,8 +568,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged r_directional_light_count = 0; r_positional_light_count = 0; - Plane camera_plane(-p_camera_transform.basis.get_column(Vector3::AXIS_Z).normalized(), p_camera_transform.origin); - omni_light_count = 0; spot_light_count = 0; @@ -720,7 +718,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged } Transform3D light_transform = light_instance->transform; - const real_t distance = camera_plane.distance_to(light_transform.origin); + const real_t distance = p_camera_transform.origin.distance_to(light_transform.origin); if (light->distance_fade) { const float fade_begin = light->distance_fade_begin; @@ -745,7 +743,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged } Transform3D light_transform = light_instance->transform; - const real_t distance = camera_plane.distance_to(light_transform.origin); + const real_t distance = p_camera_transform.origin.distance_to(light_transform.origin); if (light->distance_fade) { const float fade_begin = light->distance_fade_begin; @@ -787,6 +785,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged RS::LightType type = (i < omni_light_count) ? RS::LIGHT_OMNI : RS::LIGHT_SPOT; LightInstance *light_instance = (i < omni_light_count) ? omni_light_sort[index].light_instance : spot_light_sort[index].light_instance; Light *light = (i < omni_light_count) ? omni_light_sort[index].light : spot_light_sort[index].light; + real_t distance = (i < omni_light_count) ? omni_light_sort[index].depth : spot_light_sort[index].depth; if (using_forward_ids) { forward_id_storage->map_forward_id(type == RS::LIGHT_OMNI ? RendererRD::FORWARD_ID_TYPE_OMNI_LIGHT : RendererRD::FORWARD_ID_TYPE_SPOT_LIGHT, light_instance->forward_id, index); @@ -803,7 +802,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged float fade_begin = 0.0; float fade_shadow = 0.0; float fade_length = 0.0; - real_t distance = 0.0; float fade = 1.0; float shadow_opacity_fade = 1.0; @@ -811,7 +809,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged fade_begin = light->distance_fade_begin; fade_shadow = light->distance_fade_shadow; fade_length = light->distance_fade_length; - distance = camera_plane.distance_to(light_transform.origin); // Use `smoothstep()` to make opacity changes more gradual and less noticeable to the player. if (distance > fade_begin) { diff --git a/servers/rendering/renderer_rd/storage_rd/light_storage.h b/servers/rendering/renderer_rd/storage_rd/light_storage.h index f7c8b8833b..d359219128 100644 --- a/servers/rendering/renderer_rd/storage_rd/light_storage.h +++ b/servers/rendering/renderer_rd/storage_rd/light_storage.h @@ -226,7 +226,7 @@ private: Color ambient_color; float ambient_color_energy = 1.0; float max_distance = 0; - Vector3 extents = Vector3(1, 1, 1); + Vector3 extents = Vector3(10, 10, 10); Vector3 origin_offset; bool interior = false; bool box_projection = false; diff --git a/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp b/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp index 62da62403f..46a0d85ba8 100644 --- a/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp @@ -842,15 +842,15 @@ void MeshStorage::mesh_instance_set_blend_shape_weight(RID p_mesh_instance, int } void MeshStorage::_mesh_instance_clear(MeshInstance *mi) { - for (uint32_t i = 0; i < mi->surfaces.size(); i++) { - if (mi->surfaces[i].versions) { - for (uint32_t j = 0; j < mi->surfaces[i].version_count; j++) { - RD::get_singleton()->free(mi->surfaces[i].versions[j].vertex_array); + for (const RendererRD::MeshStorage::MeshInstance::Surface surface : mi->surfaces) { + if (surface.versions) { + for (uint32_t j = 0; j < surface.version_count; j++) { + RD::get_singleton()->free(surface.versions[j].vertex_array); } - memfree(mi->surfaces[i].versions); + memfree(surface.versions); } - if (mi->surfaces[i].vertex_buffer.is_valid()) { - RD::get_singleton()->free(mi->surfaces[i].vertex_buffer); + if (surface.vertex_buffer.is_valid()) { + RD::get_singleton()->free(surface.vertex_buffer); } } mi->surfaces.clear(); @@ -866,8 +866,8 @@ void MeshStorage::_mesh_instance_clear(MeshInstance *mi) { void MeshStorage::_mesh_instance_add_surface(MeshInstance *mi, Mesh *mesh, uint32_t p_surface) { if (mesh->blend_shape_count > 0 && mi->blend_weights_buffer.is_null()) { mi->blend_weights.resize(mesh->blend_shape_count); - for (uint32_t i = 0; i < mi->blend_weights.size(); i++) { - mi->blend_weights[i] = 0; + for (float &weight : mi->blend_weights) { + weight = 0; } mi->blend_weights_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * mi->blend_weights.size(), mi->blend_weights.to_byte_array()); mi->weights_dirty = true; diff --git a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp index e4149f6bbd..00fb8acca8 100644 --- a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp @@ -851,9 +851,9 @@ void ParticlesStorage::_particles_process(Particles *p_particles, double p_delta collision_heightmap_texture = p_particles->sdf_collision_texture; //replace in all other history frames where used because parameters are no longer valid if screen moves - for (uint32_t i = 1; i < p_particles->frame_history.size(); i++) { - if (p_particles->frame_history[i].collider_count > 0 && p_particles->frame_history[i].colliders[0].type == ParticlesFrameParams::COLLISION_TYPE_2D_SDF) { - p_particles->frame_history[i].colliders[0] = frame_params.colliders[0]; + for (ParticlesFrameParams ¶ms : p_particles->frame_history) { + if (params.collider_count > 0 && params.colliders[0].type == ParticlesFrameParams::COLLISION_TYPE_2D_SDF) { + params.colliders[0] = frame_params.colliders[0]; } } } @@ -1145,7 +1145,7 @@ void ParticlesStorage::particles_set_view_axis(RID p_particles, const Vector3 &p return; } - if (particles->particle_buffer.is_null()) { + if (particles->particle_buffer.is_null() || particles->trail_bind_pose_uniform_set.is_null()) { return; //particles have not processed yet } diff --git a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp index 31377a10a0..7771f75a65 100644 --- a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp +++ b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp @@ -251,7 +251,7 @@ RID RenderSceneBuffersRD::create_texture(const StringName &p_context, const Stri } RID RenderSceneBuffersRD::create_texture_from_format(const StringName &p_context, const StringName &p_texture_name, const RD::TextureFormat &p_texture_format, RD::TextureView p_view, bool p_unique) { - // TODO p_unique, if p_unique is true, this is a texture that can be shared. This will be implemented later as an optimisation. + // TODO p_unique, if p_unique is true, this is a texture that can be shared. This will be implemented later as an optimization. NTKey key(p_context, p_texture_name); diff --git a/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp b/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp index 06fda8fa9e..2eaa7824fb 100644 --- a/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp @@ -422,8 +422,7 @@ TextureStorage::TextureStorage() { tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT | RD::TEXTURE_USAGE_VRS_ATTACHMENT_BIT; tformat.texture_type = RD::TEXTURE_TYPE_2D; if (!RD::get_singleton()->has_feature(RD::SUPPORTS_ATTACHMENT_VRS)) { - tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT; - tformat.format = RD::DATA_FORMAT_R8_UNORM; + tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT; } Vector<uint8_t> pv; @@ -2206,6 +2205,12 @@ void TextureStorage::decal_instance_set_transform(RID p_decal_instance, const Tr di->transform = p_transform; } +void TextureStorage::decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) { + DecalInstance *di = decal_instance_owner.get_or_null(p_decal_instance); + ERR_FAIL_COND(!di); + di->sorting_offset = p_sorting_offset; +} + /* DECAL DATA API */ void TextureStorage::free_decal_data() { @@ -2233,7 +2238,7 @@ void TextureStorage::set_max_decals(const uint32_t p_max_decals) { decal_buffer = RD::get_singleton()->storage_buffer_create(decal_buffer_size); } -void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_inverse_xform) { +void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_xform) { ForwardIDStorage *forward_id_storage = ForwardIDStorage::get_singleton(); Transform3D uv_xform; @@ -2257,7 +2262,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D xform = decal_instance->transform; - real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; + real_t distance = p_camera_xform.origin.distance_to(xform.origin); if (decal->distance_fade) { float fade_begin = decal->distance_fade_begin; @@ -2272,7 +2277,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const decal_sort[decal_count].decal_instance = decal_instance; decal_sort[decal_count].decal = decal; - decal_sort[decal_count].depth = distance; + decal_sort[decal_count].depth = distance - decal_instance->sorting_offset; decal_count++; } @@ -2292,11 +2297,10 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const decal_instance->cull_mask = decal->cull_mask; - Transform3D xform = decal_instance->transform; float fade = 1.0; if (decal->distance_fade) { - const real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; + const real_t distance = decal_sort[i].depth + decal_instance->sorting_offset; const float fade_begin = decal->distance_fade_begin; const float fade_length = decal->distance_fade_length; @@ -2312,11 +2316,16 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D scale_xform; scale_xform.basis.scale(decal_extents); - Transform3D to_decal_xform = (p_camera_inverse_xform * xform * scale_xform * uv_xform).affine_inverse(); + + Transform3D xform = decal_instance->transform; + + Transform3D camera_inverse_xform = p_camera_xform.affine_inverse(); + + Transform3D to_decal_xform = (camera_inverse_xform * xform * scale_xform * uv_xform).affine_inverse(); MaterialStorage::store_transform(to_decal_xform, dd.xform); Vector3 normal = xform.basis.get_column(Vector3::AXIS_Y).normalized(); - normal = p_camera_inverse_xform.basis.xform(normal); //camera is normalized, so fine + normal = camera_inverse_xform.basis.xform(normal); //camera is normalized, so fine dd.normal[0] = normal.x; dd.normal[1] = normal.y; @@ -2350,7 +2359,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const dd.normal_rect[2] = rect.size.x; dd.normal_rect[3] = rect.size.y; - Basis normal_xform = p_camera_inverse_xform.basis * xform.basis.orthonormalized(); + Basis normal_xform = camera_inverse_xform.basis * xform.basis.orthonormalized(); MaterialStorage::store_basis_3x4(normal_xform, dd.normal_xform); } else { dd.normal_rect[0] = 0; diff --git a/servers/rendering/renderer_rd/storage_rd/texture_storage.h b/servers/rendering/renderer_rd/storage_rd/texture_storage.h index 1558342c3b..ea0df0b459 100644 --- a/servers/rendering/renderer_rd/storage_rd/texture_storage.h +++ b/servers/rendering/renderer_rd/storage_rd/texture_storage.h @@ -258,6 +258,7 @@ private: struct DecalInstance { RID decal; Transform3D transform; + float sorting_offset = 0.0; uint32_t cull_mask = 0; RendererRD::ForwardID forward_id = -1; }; @@ -646,6 +647,7 @@ public: virtual RID decal_instance_create(RID p_decal) override; virtual void decal_instance_free(RID p_decal_instance) override; virtual void decal_instance_set_transform(RID p_decal_instance, const Transform3D &p_transform) override; + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) override; _FORCE_INLINE_ RID decal_instance_get_base(RID p_decal_instance) const { DecalInstance *di = decal_instance_owner.get_or_null(p_decal_instance); @@ -677,7 +679,7 @@ public: void free_decal_data(); void set_max_decals(const uint32_t p_max_decals); RID get_decal_buffer() { return decal_buffer; } - void update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_inverse_xform); + void update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_xform); /* RENDER TARGET API */ diff --git a/servers/rendering/renderer_scene_cull.cpp b/servers/rendering/renderer_scene_cull.cpp index 218bb7b736..9011cdd98a 100644 --- a/servers/rendering/renderer_scene_cull.cpp +++ b/servers/rendering/renderer_scene_cull.cpp @@ -694,6 +694,7 @@ void RendererSceneCull::instance_set_base(RID p_instance, RID p_base) { instance->base_data = decal; decal->instance = RSG::texture_storage->decal_instance_create(p_base); + RSG::texture_storage->decal_instance_set_sorting_offset(decal->instance, instance->sorting_offset); } break; case RS::INSTANCE_LIGHTMAP: { InstanceLightmapData *lightmap_data = memnew(InstanceLightmapData); @@ -871,6 +872,9 @@ void RendererSceneCull::instance_set_pivot_data(RID p_instance, float p_sorting_ InstanceGeometryData *geom = static_cast<InstanceGeometryData *>(instance->base_data); ERR_FAIL_NULL(geom->geometry_instance); geom->geometry_instance->set_pivot_data(p_sorting_offset, p_use_aabb_center); + } else if (instance->base_type == RS::INSTANCE_DECAL && instance->base_data) { + InstanceDecalData *decal = static_cast<InstanceDecalData *>(instance->base_data); + RSG::texture_storage->decal_instance_set_sorting_offset(decal->instance, instance->sorting_offset); } } @@ -2824,7 +2828,9 @@ void RendererSceneCull::_scene_cull(CullData &cull_data, InstanceCullResult &cul InstanceGeometryData *geom = static_cast<InstanceGeometryData *>(idata.instance->base_data); ERR_FAIL_NULL(geom->geometry_instance); + cull_data.cull->lock.lock(); geom->geometry_instance->set_softshadow_projector_pairing(geom->softshadow_count > 0, geom->projector_count > 0); + cull_data.cull->lock.unlock(); idata.flags &= ~uint32_t(InstanceData::FLAG_GEOM_PROJECTOR_SOFTSHADOW_DIRTY); } @@ -2891,7 +2897,9 @@ void RendererSceneCull::_scene_cull(CullData &cull_data, InstanceCullResult &cul sh[j] = sh[j].lerp(target_sh[j], MIN(1.0, lightmap_probe_update_speed)); } ERR_FAIL_NULL(geom->geometry_instance); + cull_data.cull->lock.lock(); geom->geometry_instance->set_lightmap_capture(sh); + cull_data.cull->lock.unlock(); idata.instance->last_frame_pass = frame_number; } @@ -3088,15 +3096,15 @@ void RendererSceneCull::_render_scene(const RendererSceneRender::CameraData *p_c #endif if (cull_to > thread_cull_threshold) { //multiple threads - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].clear(); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.clear(); } WorkerThreadPool::GroupID group_task = WorkerThreadPool::get_singleton()->add_template_group_task(this, &RendererSceneCull::_scene_cull_threaded, &cull_data, scene_cull_result_threads.size(), -1, true, SNAME("RenderCullInstances")); WorkerThreadPool::get_singleton()->wait_for_group_task_completion(group_task); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result.append_from(scene_cull_result_threads[i]); + for (InstanceCullResult &thread : scene_cull_result_threads) { + scene_cull_result.append_from(thread); } } else { @@ -4126,8 +4134,8 @@ RendererSceneCull::RendererSceneCull() { scene_cull_result.init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); scene_cull_result_threads.resize(WorkerThreadPool::get_singleton()->get_thread_count()); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); } indexer_update_iterations = GLOBAL_GET("rendering/limits/spatial_indexer/update_iterations_per_frame"); @@ -4155,8 +4163,8 @@ RendererSceneCull::~RendererSceneCull() { } scene_cull_result.reset(); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].reset(); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.reset(); } scene_cull_result_threads.clear(); diff --git a/servers/rendering/shader_language.cpp b/servers/rendering/shader_language.cpp index 5ecc38a63b..ba7ffda00a 100644 --- a/servers/rendering/shader_language.cpp +++ b/servers/rendering/shader_language.cpp @@ -5402,7 +5402,7 @@ ShaderLanguage::Node *ShaderLanguage::_parse_expression(BlockNode *p_block, cons if (identifier == "SCREEN_TEXTURE" || identifier == "DEPTH_TEXTURE" || identifier == "NORMAL_ROUGHNESS_TEXTURE") { String name = String(identifier); String name_lower = name.to_lower(); - _set_error(vformat(RTR("%s has been removed in favor of using hint_%s with a uniform.\nTo continue with minimal code changes add 'uniform sampler2D %s : hint_%s, filter_linear_mipmaps;' near the top of your shader."), name, name_lower, name, name_lower)); + _set_error(vformat(RTR("%s has been removed in favor of using hint_%s with a uniform.\nTo continue with minimal code changes add 'uniform sampler2D %s : hint_%s, filter_linear_mipmap;' near the top of your shader."), name, name_lower, name, name_lower)); return nullptr; } _set_error(vformat(RTR("Unknown identifier in expression: '%s'."), String(identifier))); @@ -8697,14 +8697,17 @@ Error ShaderLanguage::_parse_shader(const HashMap<StringName, FunctionInfo> &p_f case TK_HINT_SCREEN_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_SCREEN_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_HINT_NORMAL_ROUGHNESS_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_NORMAL_ROUGHNESS_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_HINT_DEPTH_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_DEPTH_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_FILTER_NEAREST: { new_filter = FILTER_NEAREST; diff --git a/servers/rendering/shader_preprocessor.cpp b/servers/rendering/shader_preprocessor.cpp index ccbf5defa2..97efd2df94 100644 --- a/servers/rendering/shader_preprocessor.cpp +++ b/servers/rendering/shader_preprocessor.cpp @@ -330,8 +330,8 @@ String ShaderPreprocessor::vector_to_string(const LocalVector<char32_t> &p_v, in String ShaderPreprocessor::tokens_to_string(const LocalVector<Token> &p_tokens) { LocalVector<char32_t> result; - for (uint32_t i = 0; i < p_tokens.size(); i++) { - result.push_back(p_tokens[i].text); + for (const Token &token : p_tokens) { + result.push_back(token.text); } return vector_to_string(result); } diff --git a/servers/rendering/storage/texture_storage.h b/servers/rendering/storage/texture_storage.h index 92149b0064..4c4a84d04e 100644 --- a/servers/rendering/storage/texture_storage.h +++ b/servers/rendering/storage/texture_storage.h @@ -127,6 +127,7 @@ public: virtual RID decal_instance_create(RID p_decal) = 0; virtual void decal_instance_free(RID p_decal_instance) = 0; virtual void decal_instance_set_transform(RID p_decal_instance, const Transform3D &p_transform) = 0; + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) = 0; /* RENDER TARGET */ |