diff options
Diffstat (limited to 'servers/rendering/renderer_rd')
10 files changed, 633 insertions, 159 deletions
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 2c63eed4f2..89e88f23b3 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -466,6 +466,10 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(SceneState::PushConstant)); uint32_t instance_count = surf->owner->instance_count > 1 ? surf->owner->instance_count : element_info.repeat; + if (surf->flags & GeometryInstanceSurfaceDataCache::FLAG_USES_PARTICLE_TRAILS) { + instance_count /= surf->owner->trail_steps; + } + RD::get_singleton()->draw_list_draw(draw_list, index_array_rd.is_valid(), instance_count); i += element_info.repeat - 1; //skip equal elements } @@ -2379,9 +2383,13 @@ void RenderForwardClustered::_geometry_instance_add_surface_with_material(Geomet flags |= GeometryInstanceSurfaceDataCache::FLAG_PASS_SHADOW; } + if (p_material->shader_data->uses_particle_trails) { + flags |= GeometryInstanceSurfaceDataCache::FLAG_USES_PARTICLE_TRAILS; + } + SceneShaderForwardClustered::MaterialData *material_shadow = nullptr; void *surface_shadow = nullptr; - if (!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_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); @@ -2550,7 +2558,7 @@ void RenderForwardClustered::_geometry_instance_update(GeometryInstance *p_geome } } - ginstance->instance_count = storage->particles_get_amount(ginstance->data->base); + ginstance->instance_count = storage->particles_get_amount(ginstance->data->base, ginstance->trail_steps); } break; @@ -2564,42 +2572,26 @@ void RenderForwardClustered::_geometry_instance_update(GeometryInstance *p_geome if (ginstance->data->base_type == RS::INSTANCE_MULTIMESH) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH; - uint32_t stride; if (storage->multimesh_get_transform_format(ginstance->data->base) == RS::MULTIMESH_TRANSFORM_2D) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_FORMAT_2D; - stride = 2; - } else { - stride = 3; } if (storage->multimesh_uses_colors(ginstance->data->base)) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_HAS_COLOR; - stride += 1; } if (storage->multimesh_uses_custom_data(ginstance->data->base)) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_HAS_CUSTOM_DATA; - stride += 1; } - ginstance->base_flags |= (stride << INSTANCE_DATA_FLAGS_MULTIMESH_STRIDE_SHIFT); ginstance->transforms_uniform_set = storage->multimesh_get_3d_uniform_set(ginstance->data->base, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); } else if (ginstance->data->base_type == RS::INSTANCE_PARTICLES) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH; - uint32_t stride; - if (false) { // 2D particles - ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_FORMAT_2D; - stride = 2; - } else { - stride = 3; - } ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_HAS_COLOR; - stride += 1; - ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH_HAS_CUSTOM_DATA; - stride += 1; - ginstance->base_flags |= (stride << INSTANCE_DATA_FLAGS_MULTIMESH_STRIDE_SHIFT); + //for particles, stride is the trail size + ginstance->base_flags |= (ginstance->trail_steps << INSTANCE_DATA_FLAGS_PARTICLE_TRAIL_SHIFT); if (!storage->particles_is_using_local_coords(ginstance->data->base)) { store_transform = false; @@ -2608,7 +2600,6 @@ void RenderForwardClustered::_geometry_instance_update(GeometryInstance *p_geome } else if (ginstance->data->base_type == RS::INSTANCE_MESH) { if (storage->skeleton_is_valid(ginstance->data->skeleton)) { - ginstance->base_flags |= INSTANCE_DATA_FLAG_SKELETON; ginstance->transforms_uniform_set = storage->skeleton_get_3d_uniform_set(ginstance->data->skeleton, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); if (ginstance->data->dirty_dependencies) { storage->skeleton_update_dependency(ginstance->data->skeleton, &ginstance->data->dependency_tracker); @@ -2643,6 +2634,7 @@ void RenderForwardClustered::_geometry_instance_dependency_changed(RendererStora switch (p_notification) { case RendererStorage::DEPENDENCY_CHANGED_MATERIAL: case RendererStorage::DEPENDENCY_CHANGED_MESH: + case RendererStorage::DEPENDENCY_CHANGED_PARTICLES: case RendererStorage::DEPENDENCY_CHANGED_MULTIMESH: case RendererStorage::DEPENDENCY_CHANGED_SKELETON_DATA: { static_cast<RenderForwardClustered *>(singleton)->_geometry_instance_mark_dirty(static_cast<GeometryInstance *>(p_tracker->userdata)); 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 72e84a6f24..05de10ae28 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h @@ -196,9 +196,9 @@ class RenderForwardClustered : public RendererSceneRenderRD { INSTANCE_DATA_FLAG_MULTIMESH_FORMAT_2D = 1 << 13, INSTANCE_DATA_FLAG_MULTIMESH_HAS_COLOR = 1 << 14, INSTANCE_DATA_FLAG_MULTIMESH_HAS_CUSTOM_DATA = 1 << 15, - INSTANCE_DATA_FLAGS_MULTIMESH_STRIDE_SHIFT = 16, - INSTANCE_DATA_FLAGS_MULTIMESH_STRIDE_MASK = 0x7, - INSTANCE_DATA_FLAG_SKELETON = 1 << 19, + INSTANCE_DATA_FLAGS_PARTICLE_TRAIL_SHIFT = 16, + INSTANCE_DATA_FLAGS_PARTICLE_TRAIL_MASK = 0xFF, + INSTANCE_DATA_FLAGS_NON_UNIFORM_SCALE = 1 << 24, }; struct SceneState { @@ -398,6 +398,7 @@ class RenderForwardClustered : public RendererSceneRenderRD { FLAG_USES_DEPTH_TEXTURE = 8192, FLAG_USES_NORMAL_TEXTURE = 16384, FLAG_USES_DOUBLE_SIDED_SHADOWS = 32768, + FLAG_USES_PARTICLE_TRAILS = 65536, }; union { @@ -453,6 +454,7 @@ class RenderForwardClustered : public RendererSceneRenderRD { uint32_t layer_mask = 1; RID transforms_uniform_set; uint32_t instance_count = 0; + uint32_t trail_steps = 1; RID mesh_instance; bool can_sdfgi = false; //used during setup 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 15982b4b29..df2931126b 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 @@ -73,6 +73,7 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { uses_time = false; writes_modelview_or_projection = false; uses_world_coordinates = false; + uses_particle_trails = false; int depth_drawi = DEPTH_DRAW_OPAQUE; @@ -101,6 +102,7 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { actions.render_mode_flags["unshaded"] = &unshaded; actions.render_mode_flags["wireframe"] = &wireframe; + actions.render_mode_flags["particle_trails"] = &uses_particle_trails; actions.usage_flag_pointers["ALPHA"] = &uses_alpha; actions.render_mode_flags["depth_prepass_alpha"] = &uses_depth_pre_pass; @@ -714,6 +716,7 @@ void SceneShaderForwardClustered::init(RendererStorageRD *p_storage, const Strin actions.render_mode_defines["ensure_correct_normals"] = "#define ENSURE_CORRECT_NORMALS\n"; actions.render_mode_defines["cull_front"] = "#define DO_SIDE_CHECK\n"; actions.render_mode_defines["cull_disabled"] = "#define DO_SIDE_CHECK\n"; + actions.render_mode_defines["particle_trails"] = "#define USE_PARTICLE_TRAILS\n"; bool force_lambert = GLOBAL_GET("rendering/shading/overrides/force_lambert_over_burley"); 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 953a5291c8..7c8879686b 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 @@ -126,6 +126,7 @@ public: bool uses_discard; bool uses_roughness; bool uses_normal; + bool uses_particle_trails; bool unshaded; bool uses_vertex; diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.cpp b/servers/rendering/renderer_rd/renderer_storage_rd.cpp index 250b694f74..47b9e33ca6 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_storage_rd.cpp @@ -3888,22 +3888,37 @@ bool RendererStorageRD::particles_get_emitting(RID p_particles) { } void RendererStorageRD::_particles_free_data(Particles *particles) { - if (!particles->particle_buffer.is_valid()) { - return; + if (particles->particle_buffer.is_valid()) { + RD::get_singleton()->free(particles->particle_buffer); + particles->particle_buffer = RID(); + RD::get_singleton()->free(particles->particle_instance_buffer); + particles->particle_instance_buffer = RID(); + } + + if (particles->frame_params_buffer.is_valid()) { + RD::get_singleton()->free(particles->frame_params_buffer); + particles->frame_params_buffer = RID(); } - RD::get_singleton()->free(particles->particle_buffer); - RD::get_singleton()->free(particles->frame_params_buffer); - RD::get_singleton()->free(particles->particle_instance_buffer); particles->particles_transforms_buffer_uniform_set = RID(); - particles->particle_buffer = RID(); + if (RD::get_singleton()->uniform_set_is_valid(particles->trail_bind_pose_uniform_set)) { + RD::get_singleton()->free(particles->trail_bind_pose_uniform_set); + } + particles->trail_bind_pose_uniform_set = RID(); + + if (particles->trail_bind_pose_buffer.is_valid()) { + RD::get_singleton()->free(particles->trail_bind_pose_buffer); + particles->trail_bind_pose_buffer = RID(); + } if (RD::get_singleton()->uniform_set_is_valid(particles->collision_textures_uniform_set)) { RD::get_singleton()->free(particles->collision_textures_uniform_set); } + particles->collision_textures_uniform_set = RID(); if (particles->particles_sort_buffer.is_valid()) { RD::get_singleton()->free(particles->particles_sort_buffer); particles->particles_sort_buffer = RID(); + particles->particles_sort_uniform_set = RID(); } if (particles->emission_buffer != nullptr) { @@ -3912,6 +3927,12 @@ void RendererStorageRD::_particles_free_data(Particles *particles) { RD::get_singleton()->free(particles->emission_storage_buffer); particles->emission_storage_buffer = RID(); } + + if (RD::get_singleton()->uniform_set_is_valid(particles->particles_material_uniform_set)) { + //will need to be re-created + RD::get_singleton()->free(particles->particles_material_uniform_set); + } + particles->particles_material_uniform_set = RID(); } void RendererStorageRD::particles_set_amount(RID p_particles, int p_amount) { @@ -3926,38 +3947,12 @@ void RendererStorageRD::particles_set_amount(RID p_particles, int p_amount) { particles->amount = p_amount; - if (particles->amount > 0) { - particles->particle_buffer = RD::get_singleton()->storage_buffer_create(sizeof(ParticleData) * p_amount); - particles->frame_params_buffer = RD::get_singleton()->storage_buffer_create(sizeof(ParticlesFrameParams) * 1); - particles->particle_instance_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * 4 * (3 + 1 + 1) * p_amount); - //needs to clear it - - { - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 1; - u.ids.push_back(particles->particle_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 2; - u.ids.push_back(particles->particle_instance_buffer); - uniforms.push_back(u); - } - - particles->particles_copy_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, particles_shader.copy_shader.version_get_shader(particles_shader.copy_shader_version, 0), 0); - } - } - particles->prev_ticks = 0; particles->phase = 0; particles->prev_phase = 0; particles->clear = true; + + particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); } void RendererStorageRD::particles_set_lifetime(RID p_particles, float p_lifetime) { @@ -4013,6 +4008,22 @@ void RendererStorageRD::particles_set_fixed_fps(RID p_particles, int p_fps) { ERR_FAIL_COND(!particles); particles->fixed_fps = p_fps; + + _particles_free_data(particles); + + particles->prev_ticks = 0; + particles->phase = 0; + particles->prev_phase = 0; + particles->clear = true; + + particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); +} + +void RendererStorageRD::particles_set_interpolate(RID p_particles, bool p_enable) { + Particles *particles = particles_owner.getornull(p_particles); + ERR_FAIL_COND(!particles); + + particles->interpolate = p_enable; } void RendererStorageRD::particles_set_fractional_delta(RID p_particles, bool p_enable) { @@ -4022,6 +4033,42 @@ void RendererStorageRD::particles_set_fractional_delta(RID p_particles, bool p_e 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); + ERR_FAIL_COND(!particles); + ERR_FAIL_COND(p_length < 0.1); + p_length = MIN(10.0, p_length); + + particles->trails_enabled = p_enable; + particles->trail_length = p_length; + + _particles_free_data(particles); + + particles->prev_ticks = 0; + particles->phase = 0; + particles->prev_phase = 0; + particles->clear = true; + + particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); +} + +void RendererStorageRD::particles_set_trail_bind_poses(RID p_particles, const Vector<Transform> &p_bind_poses) { + Particles *particles = particles_owner.getornull(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); + + particles->prev_ticks = 0; + particles->phase = 0; + particles->prev_phase = 0; + particles->clear = true; + } + particles->trail_bind_poses = p_bind_poses; + particles->trail_bind_poses_dirty = true; + + 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); ERR_FAIL_COND(!particles); @@ -4029,6 +4076,13 @@ void RendererStorageRD::particles_set_collision_base_size(RID p_particles, float 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); + 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); ERR_FAIL_COND(!particles); @@ -4152,8 +4206,13 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { const Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND_V(!particles, AABB()); + int total_amount = particles->amount; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + total_amount *= particles->trail_bind_poses.size(); + } + Vector<ParticleData> data; - data.resize(particles->amount); + data.resize(total_amount); Vector<uint8_t> buffer = RD::get_singleton()->buffer_get_data(particles->particle_buffer); @@ -4162,8 +4221,9 @@ AABB RendererStorageRD::particles_get_current_aabb(RID p_particles) { AABB aabb; if (buffer.size()) { bool first = true; + const ParticleData *particle_data = (const ParticleData *)data.ptr(); - for (int i = 0; i < particles->amount; i++) { + 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]); if (!particles->use_local_coords) { @@ -4224,14 +4284,12 @@ 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); 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); ERR_FAIL_COND(!particles); - particles->collisions.erase(p_particles_collision_instance); } @@ -4286,7 +4344,12 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta float new_phase = Math::fmod((float)p_particles->phase + (p_delta / p_particles->lifetime) * p_particles->speed_scale, (float)1.0); - ParticlesFrameParams &frame_params = p_particles->frame_params; + //move back history (if there is any) + for (uint32_t i = p_particles->frame_history.size() - 1; i > 0; i--) { + p_particles->frame_history[i] = p_particles->frame_history[i - 1]; + } + //update current frame + ParticlesFrameParams &frame_params = p_particles->frame_history[0]; if (p_particles->clear) { p_particles->cycle_number = 0; @@ -4317,6 +4380,10 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta } frame_params.cycle = p_particles->cycle_number; + frame_params.frame = p_particles->frame_counter++; + frame_params.pad0 = 0; + frame_params.pad1 = 0; + frame_params.pad2 = 0; { //collision and attractors @@ -4515,12 +4582,18 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta ParticlesShader::PushConstant push_constant; + int process_amount = p_particles->amount; + + if (p_particles->trails_enabled && p_particles->trail_bind_poses.size() > 1) { + process_amount *= p_particles->trail_bind_poses.size(); + } push_constant.clear = p_particles->clear; push_constant.total_particles = p_particles->amount; push_constant.lifetime = p_particles->lifetime; - push_constant.trail_size = 1; + push_constant.trail_size = p_particles->trail_params.size(); push_constant.use_fractional_delta = p_particles->fractional_delta; push_constant.sub_emitter_mode = !p_particles->emitting && p_particles->emission_buffer && (p_particles->emission_buffer->particle_count > 0 || p_particles->force_sub_emit); + push_constant.trail_pass = false; p_particles->force_sub_emit = false; //reset @@ -4553,7 +4626,17 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta p_particles->clear = false; - RD::get_singleton()->buffer_update(p_particles->frame_params_buffer, 0, sizeof(ParticlesFrameParams), &frame_params); + if (p_particles->trail_params.size() > 1) { + //fill the trail params + for (uint32_t i = 0; i < p_particles->trail_params.size(); i++) { + uint32_t src_idx = i * p_particles->frame_history.size() / p_particles->trail_params.size(); + p_particles->trail_params[i] = p_particles->frame_history[src_idx]; + } + } else { + p_particles->trail_params[0] = p_particles->frame_history[0]; + } + + RD::get_singleton()->buffer_update(p_particles->frame_params_buffer, 0, sizeof(ParticlesFrameParams) * p_particles->trail_params.size(), p_particles->trail_params.ptr()); ParticlesMaterialData *m = (ParticlesMaterialData *)material_get_data(p_particles->process_material, SHADER_TYPE_PARTICLES); if (!m) { @@ -4575,27 +4658,45 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ParticlesShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_particles->amount, 1, 1); + if (p_particles->trails_enabled && p_particles->trail_bind_poses.size() > 1) { + //trails requires two passes in order to catch particle starts + RD::get_singleton()->compute_list_dispatch_threads(compute_list, process_amount / p_particles->trail_bind_poses.size(), 1, 1); + + RD::get_singleton()->compute_list_add_barrier(compute_list); + + push_constant.trail_pass = true; + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ParticlesShader::PushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, process_amount - p_particles->amount, 1, 1); + } else { + RD::get_singleton()->compute_list_dispatch_threads(compute_list, process_amount, 1, 1); + } RD::get_singleton()->compute_list_end(); } -void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 &p_axis) { +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); ERR_FAIL_COND(!particles); - if (particles->draw_order != RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH) { - return; //uninteresting for other modes + 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) { + return; + } + + if (particles->particle_buffer.is_null()) { + return; //particles have not processed yet } + bool do_sort = particles->draw_order == RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH; + //copy to sort buffer - if (particles->particles_sort_buffer == RID()) { + if (do_sort && particles->particles_sort_buffer == RID()) { uint32_t size = particles->amount; if (size & 1) { size++; //make multiple of 16 } size *= sizeof(float) * 2; particles->particles_sort_buffer = RD::get_singleton()->storage_buffer_create(size); + { Vector<RD::Uniform> uniforms; @@ -4611,41 +4712,105 @@ void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 & } } + ParticlesShader::CopyPushConstant copy_push_constant; + + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + int fixed_fps = 60.0; + if (particles->fixed_fps > 0) { + fixed_fps = particles->fixed_fps; + } + + copy_push_constant.trail_size = particles->trail_bind_poses.size(); + copy_push_constant.trail_total = particles->frame_history.size(); + copy_push_constant.frame_delta = 1.0 / fixed_fps; + } else { + copy_push_constant.trail_size = 1; + copy_push_constant.trail_total = 1; + copy_push_constant.frame_delta = 0.0; + } + copy_push_constant.frame_remainder = particles->interpolate ? particles->frame_remainder : 0.0; + copy_push_constant.total_particles = particles->amount; + Vector3 axis = -p_axis; // cameras look to z negative if (particles->use_local_coords) { axis = particles->emission_transform.basis.xform_inv(axis).normalized(); } - ParticlesShader::CopyPushConstant copy_push_constant; - copy_push_constant.total_particles = particles->amount; copy_push_constant.sort_direction[0] = axis.x; copy_push_constant.sort_direction[1] = axis.y; copy_push_constant.sort_direction[2] = axis.z; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, particles_shader.copy_pipelines[ParticlesShader::COPY_MODE_FILL_SORT_BUFFER]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_copy_uniform_set, 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); - RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); + copy_push_constant.align_up[0] = p_up_axis.x; + copy_push_constant.align_up[1] = p_up_axis.y; + copy_push_constant.align_up[2] = p_up_axis.z; - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); + copy_push_constant.align_mode = particles->transform_align; - RD::get_singleton()->compute_list_end(); + if (do_sort) { + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, particles_shader.copy_pipelines[ParticlesShader::COPY_MODE_FILL_SORT_BUFFER]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_copy_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->trail_bind_pose_uniform_set, 2); + RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); + + RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); - effects.sort_buffer(particles->particles_sort_uniform_set, particles->amount); + RD::get_singleton()->compute_list_end(); + effects.sort_buffer(particles->particles_sort_uniform_set, particles->amount); + } + + copy_push_constant.total_particles *= copy_push_constant.total_particles; - compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, particles_shader.copy_pipelines[ParticlesShader::COPY_MODE_FILL_INSTANCES_WITH_SORT_BUFFER]); + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, particles_shader.copy_pipelines[do_sort ? ParticlesShader::COPY_MODE_FILL_INSTANCES_WITH_SORT_BUFFER : ParticlesShader::COPY_MODE_FILL_INSTANCES]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_copy_uniform_set, 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); + if (do_sort) { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); + } + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->trail_bind_pose_uniform_set, 2); + RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, copy_push_constant.total_particles, 1, 1); RD::get_singleton()->compute_list_end(); } +void RendererStorageRD::_particles_update_buffers(Particles *particles) { + if (particles->amount > 0 && particles->particle_buffer.is_null()) { + int total_amount = particles->amount; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + total_amount *= particles->trail_bind_poses.size(); + } + particles->particle_buffer = RD::get_singleton()->storage_buffer_create(sizeof(ParticleData) * total_amount); + particles->particle_instance_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * 4 * (3 + 1 + 1) * total_amount); + //needs to clear it + + { + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(particles->particle_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 2; + u.ids.push_back(particles->particle_instance_buffer); + uniforms.push_back(u); + } + + particles->particles_copy_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, particles_shader.copy_shader.version_get_shader(particles_shader.copy_shader_version, 0), 0); + } + } +} void RendererStorageRD::update_particles() { while (particle_update_list) { //use transform feedback to process particles @@ -4657,6 +4822,8 @@ void RendererStorageRD::update_particles() { particles->update_list = nullptr; particles->dirty = false; + _particles_update_buffers(particles); + if (particles->restart_request) { particles->prev_ticks = 0; particles->phase = 0; @@ -4688,12 +4855,81 @@ void RendererStorageRD::update_particles() { } } +#ifndef _MSC_VER +#warning Should use display refresh rate for all this +#endif + + float screen_hz = 60; + + int fixed_fps = 0; + if (particles->fixed_fps > 0) { + fixed_fps = particles->fixed_fps; + } else if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + fixed_fps = screen_hz; + } + { + //update trails + int history_size = 1; + int trail_steps = 1; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + history_size = MAX(1, int(particles->trail_length * fixed_fps)); + trail_steps = particles->trail_bind_poses.size(); + } + + if (uint32_t(history_size) != particles->frame_history.size()) { + particles->frame_history.resize(history_size); + memset(particles->frame_history.ptr(), 0, sizeof(ParticlesFrameParams) * history_size); + } + + if (uint32_t(trail_steps) != particles->trail_params.size() || particles->frame_params_buffer.is_null()) { + particles->trail_params.resize(trail_steps); + if (particles->frame_params_buffer.is_valid()) { + RD::get_singleton()->free(particles->frame_params_buffer); + } + particles->frame_params_buffer = RD::get_singleton()->storage_buffer_create(sizeof(ParticlesFrameParams) * trail_steps); + } + + if (particles->trail_bind_poses.size() > 1 && particles->trail_bind_pose_buffer.is_null()) { + particles->trail_bind_pose_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * 16 * particles->trail_bind_poses.size()); + particles->trail_bind_poses_dirty = true; + } + + if (particles->trail_bind_pose_uniform_set.is_null()) { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 0; + if (particles->trail_bind_pose_buffer.is_valid()) { + u.ids.push_back(particles->trail_bind_pose_buffer); + } else { + u.ids.push_back(default_rd_storage_buffer); + } + uniforms.push_back(u); + } + + particles->trail_bind_pose_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, particles_shader.copy_shader.version_get_shader(particles_shader.copy_shader_version, 0), 2); + } + + if (particles->trail_bind_pose_buffer.is_valid() && particles->trail_bind_poses_dirty) { + if (particles_shader.pose_update_buffer.size() < uint32_t(particles->trail_bind_poses.size()) * 16) { + particles_shader.pose_update_buffer.resize(particles->trail_bind_poses.size() * 16); + } + + for (int i = 0; i < particles->trail_bind_poses.size(); i++) { + store_transform(particles->trail_bind_poses[i], &particles_shader.pose_update_buffer[i * 16]); + } + + RD::get_singleton()->buffer_update(particles->trail_bind_pose_buffer, 0, particles->trail_bind_poses.size() * 16 * sizeof(float), particles_shader.pose_update_buffer.ptr()); + } + } + bool zero_time_scale = Engine::get_singleton()->get_time_scale() <= 0.0; if (particles->clear && particles->pre_process_time > 0.0) { float frame_time; - if (particles->fixed_fps > 0) { - frame_time = 1.0 / particles->fixed_fps; + if (fixed_fps > 0) { + frame_time = 1.0 / fixed_fps; } else { frame_time = 1.0 / 30.0; } @@ -4706,14 +4942,14 @@ void RendererStorageRD::update_particles() { } } - if (particles->fixed_fps > 0) { + if (fixed_fps > 0) { float frame_time; float decr; if (zero_time_scale) { frame_time = 0.0; - decr = 1.0 / particles->fixed_fps; + decr = 1.0 / fixed_fps; } else { - frame_time = 1.0 / particles->fixed_fps; + frame_time = 1.0 / fixed_fps; decr = frame_time; } float delta = RendererCompositorRD::singleton->get_frame_delta_time(); @@ -4741,16 +4977,39 @@ void RendererStorageRD::update_particles() { //copy particles to instance buffer - if (particles->draw_order != RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH) { + 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) { + //does not need view dependent operation, do copy here ParticlesShader::CopyPushConstant copy_push_constant; - copy_push_constant.total_particles = particles->amount; + + int total_amount = particles->amount; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + total_amount *= particles->trail_bind_poses.size(); + } + + copy_push_constant.total_particles = total_amount; + copy_push_constant.frame_remainder = particles->interpolate ? particles->frame_remainder : 0.0; + copy_push_constant.align_mode = particles->transform_align; + copy_push_constant.align_up[0] = 0; + copy_push_constant.align_up[1] = 0; + copy_push_constant.align_up[2] = 0; + + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + copy_push_constant.trail_size = particles->trail_bind_poses.size(); + copy_push_constant.trail_total = particles->frame_history.size(); + copy_push_constant.frame_delta = 1.0 / fixed_fps; + } else { + copy_push_constant.trail_size = 1; + copy_push_constant.trail_total = 1; + copy_push_constant.frame_delta = 0.0; + } RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, particles_shader.copy_pipelines[ParticlesShader::COPY_MODE_FILL_INSTANCES]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_copy_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->trail_bind_pose_uniform_set, 2); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, total_amount, 1, 1); RD::get_singleton()->compute_list_end(); } @@ -8324,9 +8583,10 @@ bool RendererStorageRD::free(RID p_rid) { light_owner.free(p_rid); } else if (particles_owner.owns(p_rid)) { + update_particles(); Particles *particles = particles_owner.getornull(p_rid); - _particles_free_data(particles); 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); @@ -8863,14 +9123,14 @@ RendererStorageRD::RendererStorageRD() { actions.renames["COLOR"] = "PARTICLE.color"; actions.renames["VELOCITY"] = "PARTICLE.velocity"; //actions.renames["MASS"] = "mass"; ? - actions.renames["ACTIVE"] = "PARTICLE.is_active"; + actions.renames["ACTIVE"] = "particle_active"; actions.renames["RESTART"] = "restart"; actions.renames["CUSTOM"] = "PARTICLE.custom"; actions.renames["TRANSFORM"] = "PARTICLE.xform"; actions.renames["TIME"] = "FRAME.time"; actions.renames["LIFETIME"] = "params.lifetime"; actions.renames["DELTA"] = "local_delta"; - actions.renames["NUMBER"] = "particle"; + actions.renames["NUMBER"] = "particle_number"; actions.renames["INDEX"] = "index"; //actions.renames["GRAVITY"] = "current_gravity"; actions.renames["EMISSION_TRANSFORM"] = "FRAME.emission_transform"; diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.h b/servers/rendering/renderer_rd/renderer_storage_rd.h index 6405bb75b0..961bdfb178 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.h +++ b/servers/rendering/renderer_rd/renderer_storage_rd.h @@ -660,6 +660,11 @@ private: float time; float delta; + uint32_t frame; + uint32_t pad0; + uint32_t pad1; + uint32_t pad2; + uint32_t random_seed; uint32_t attractor_count; uint32_t collider_count; @@ -704,10 +709,16 @@ private: AABB custom_aabb = AABB(Vector3(-4, -4, -4), Vector3(8, 8, 8)); bool use_local_coords = true; RID process_material; + uint32_t frame_counter = 0; + RS::ParticlesTransformAlign transform_align = RS::PARTICLES_TRANSFORM_ALIGN_DISABLED; RS::ParticlesDrawOrder draw_order = RS::PARTICLES_DRAW_ORDER_INDEX; Vector<RID> draw_passes; + Vector<Transform> trail_bind_poses; + bool trail_bind_poses_dirty = false; + RID trail_bind_pose_buffer; + RID trail_bind_pose_uniform_set; RID particle_buffer; RID particle_instance_buffer; @@ -739,7 +750,8 @@ private: float speed_scale = 1.0; - int fixed_fps = 0; + int fixed_fps = 30; + bool interpolate = true; bool fractional_delta = false; float frame_remainder = 0; float collision_base_size = 0.01; @@ -759,12 +771,19 @@ private: Dependency dependency; - ParticlesFrameParams frame_params; + float trail_length = 1.0; + bool trails_enabled = false; + LocalVector<ParticlesFrameParams> frame_history; + LocalVector<ParticlesFrameParams> trail_params; + + Particles() { + } }; void _particles_process(Particles *p_particles, float p_delta); void _particles_allocate_emission_buffer(Particles *particles); void _particles_free_data(Particles *particles); + void _particles_update_buffers(Particles *particles); struct ParticlesShader { struct PushConstant { @@ -776,7 +795,7 @@ private: uint32_t use_fractional_delta; uint32_t sub_emitter_mode; uint32_t can_emit; - uint32_t pad; + uint32_t trail_pass; }; ParticlesShaderRD shader; @@ -791,6 +810,14 @@ private: struct CopyPushConstant { float sort_direction[3]; uint32_t total_particles; + + uint32_t trail_size; + uint32_t trail_total; + float frame_delta; + float frame_remainder; + + float align_up[3]; + uint32_t align_mode; }; enum { @@ -804,6 +831,8 @@ private: RID copy_shader_version; RID copy_pipelines[COPY_MODE_MAX]; + LocalVector<float> pose_update_buffer; + } particles_shader; Particles *particle_update_list = nullptr; @@ -2076,10 +2105,17 @@ public: 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_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_trail_bind_poses(RID p_particles, const Vector<Transform> &p_bind_poses); + void particles_restart(RID p_particles); void particles_emit(RID p_particles, const Transform &p_transform, const Vector3 &p_velocity, const Color &p_color, const Color &p_custom, uint32_t p_emit_flags); + void particles_set_subemitter(RID p_particles, RID p_subemitter_particles); void particles_set_draw_order(RID p_particles, RS::ParticlesDrawOrder p_order); @@ -2097,15 +2133,21 @@ public: int particles_get_draw_passes(RID p_particles) const; RID particles_get_draw_pass_mesh(RID p_particles, int p_pass) const; - void particles_set_view_axis(RID p_particles, const Vector3 &p_axis); + void particles_set_view_axis(RID p_particles, const Vector3 &p_axis, const Vector3 &p_up_axis); virtual bool particles_is_inactive(RID p_particles) const; - _FORCE_INLINE_ uint32_t particles_get_amount(RID p_particles) { + _FORCE_INLINE_ uint32_t particles_get_amount(RID p_particles, uint32_t &r_trail_divisor) { Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND_V(!particles, 0); - return particles->amount; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + r_trail_divisor = particles->trail_bind_poses.size(); + } else { + r_trail_divisor = 1; + } + + return particles->amount * r_trail_divisor; } _FORCE_INLINE_ uint32_t particles_is_using_local_coords(RID p_particles) { @@ -2119,6 +2161,8 @@ public: Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND_V(!particles, RID()); if (particles->particles_transforms_buffer_uniform_set.is_null()) { + _particles_update_buffers(particles); + Vector<RD::Uniform> uniforms; { diff --git a/servers/rendering/renderer_rd/shaders/particles.glsl b/servers/rendering/renderer_rd/shaders/particles.glsl index 3712220dc4..beaff10793 100644 --- a/servers/rendering/renderer_rd/shaders/particles.glsl +++ b/servers/rendering/renderer_rd/shaders/particles.glsl @@ -76,6 +76,11 @@ struct FrameParams { float time; float delta; + uint frame; + uint pad0; + uint pad1; + uint pad2; + uint random_seed; uint attractor_count; uint collider_count; @@ -92,10 +97,16 @@ layout(set = 1, binding = 0, std430) restrict buffer FrameHistory { } frame_history; +#define PARTICLE_FLAG_ACTIVE uint(1) +#define PARTICLE_FLAG_STARTED uint(2) +#define PARTICLE_FLAG_TRAILED uint(4) +#define PARTICLE_FRAME_MASK uint(0xFFFF) +#define PARTICLE_FRAME_SHIFT uint(16) + struct ParticleData { mat4 xform; vec3 velocity; - bool is_active; + uint flags; vec4 color; vec4 custom; }; @@ -162,7 +173,7 @@ layout(push_constant, binding = 0, std430) uniform Params { bool use_fractional_delta; bool sub_emitter_mode; bool can_emit; - uint pad; + bool trail_pass; } params; @@ -201,6 +212,14 @@ bool emit_subparticle(mat4 p_xform, vec3 p_velocity, vec4 p_color, vec4 p_custom void main() { uint particle = gl_GlobalInvocationID.x; + if (params.trail_size > 1) { + if (params.trail_pass) { + particle += (particle / (params.trail_size - 1)) + 1; + } else { + particle *= params.trail_size; + } + } + if (particle >= params.total_particles * params.trail_size) { return; //discard } @@ -229,7 +248,7 @@ void main() { PARTICLE.color = vec4(1.0); PARTICLE.custom = vec4(0.0); PARTICLE.velocity = vec3(0.0); - PARTICLE.is_active = false; + PARTICLE.flags = 0; PARTICLE.xform = mat4( vec4(1.0, 0.0, 0.0, 0.0), vec4(0.0, 1.0, 0.0, 0.0), @@ -237,6 +256,29 @@ void main() { vec4(0.0, 0.0, 0.0, 1.0)); } + //clear started flag if set + + if (params.trail_pass) { + //trail started + uint src_idx = index * params.trail_size; + if (bool(particles.data[src_idx].flags & PARTICLE_FLAG_STARTED)) { + //save start conditions for trails + PARTICLE.color = particles.data[src_idx].color; + PARTICLE.custom = particles.data[src_idx].custom; + PARTICLE.velocity = particles.data[src_idx].velocity; + PARTICLE.flags = PARTICLE_FLAG_TRAILED | ((frame_history.data[0].frame & PARTICLE_FRAME_MASK) << PARTICLE_FRAME_SHIFT); //mark it as trailed, save in which frame it will start + PARTICLE.xform = particles.data[src_idx].xform; + } + + if (bool(PARTICLE.flags & PARTICLE_FLAG_TRAILED) && ((PARTICLE.flags >> PARTICLE_FRAME_SHIFT) == (FRAME.frame & PARTICLE_FRAME_MASK))) { //check this is trailed and see if it should start now + // we just assume that this is the first frame of the particle, the rest is deterministic + PARTICLE.flags = PARTICLE_FLAG_ACTIVE | (particles.data[src_idx].flags & (PARTICLE_FRAME_MASK << PARTICLE_FRAME_SHIFT)); + return; //- this appears like it should be correct, but it seems not to be.. wonder why. + } + } else { + PARTICLE.flags &= ~PARTICLE_FLAG_STARTED; + } + bool collided = false; vec3 collision_normal = vec3(0.0); float collision_depth = 0.0; @@ -245,19 +287,17 @@ void main() { #if !defined(DISABLE_VELOCITY) - if (PARTICLE.is_active) { + if (bool(PARTICLE.flags & PARTICLE_FLAG_ACTIVE)) { PARTICLE.xform[3].xyz += PARTICLE.velocity * local_delta; } #endif - /* Process physics if active */ - - if (params.sub_emitter_mode) { - if (!PARTICLE.is_active) { + if (!params.trail_pass && params.sub_emitter_mode) { + if (!bool(PARTICLE.flags & PARTICLE_FLAG_ACTIVE)) { int src_index = atomicAdd(src_particles.particle_count, -1) - 1; if (src_index >= 0) { - PARTICLE.is_active = true; + PARTICLE.flags = (PARTICLE_FLAG_ACTIVE | PARTICLE_FLAG_STARTED | (FRAME.cycle << PARTICLE_FRAME_SHIFT)); restart = true; if (bool(src_particles.data[src_index].flags & EMISSION_FLAG_HAS_POSITION)) { @@ -339,16 +379,12 @@ void main() { } } - uint current_cycle = FRAME.cycle; - - if (FRAME.system_phase < restart_phase) { - current_cycle -= uint(1); + if (params.trail_pass) { + restart = false; } - uint particle_number = current_cycle * uint(params.total_particles) + particle; - if (restart) { - PARTICLE.is_active = FRAME.emitting; + PARTICLE.flags = FRAME.emitting ? (PARTICLE_FLAG_ACTIVE | PARTICLE_FLAG_STARTED | (FRAME.cycle << PARTICLE_FRAME_SHIFT)) : 0; restart_position = true; restart_rotation_scale = true; restart_velocity = true; @@ -357,11 +393,15 @@ void main() { } } - if (restart && PARTICLE.is_active) { + bool particle_active = bool(PARTICLE.flags & PARTICLE_FLAG_ACTIVE); + + uint particle_number = (PARTICLE.flags >> PARTICLE_FRAME_SHIFT) * uint(params.total_particles) + index; + + if (restart && particle_active) { #CODE : START } - if (PARTICLE.is_active) { + if (particle_active) { for (uint i = 0; i < FRAME.attractor_count; i++) { vec3 dir; float amount; @@ -539,7 +579,12 @@ void main() { } } - if (PARTICLE.is_active) { + if (particle_active) { #CODE : PROCESS } + + PARTICLE.flags &= ~PARTICLE_FLAG_ACTIVE; + if (particle_active) { + PARTICLE.flags |= PARTICLE_FLAG_ACTIVE; + } } diff --git a/servers/rendering/renderer_rd/shaders/particles_copy.glsl b/servers/rendering/renderer_rd/shaders/particles_copy.glsl index 80adb49619..7804d66d1c 100644 --- a/servers/rendering/renderer_rd/shaders/particles_copy.glsl +++ b/servers/rendering/renderer_rd/shaders/particles_copy.glsl @@ -6,10 +6,14 @@ layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in; +#define PARTICLE_FLAG_ACTIVE uint(1) +#define PARTICLE_FLAG_STARTED uint(2) +#define PARTICLE_FLAG_TRAILED uint(4) + struct ParticleData { mat4 xform; vec3 velocity; - bool is_active; + uint flags; vec4 color; vec4 custom; }; @@ -33,12 +37,30 @@ sort_buffer; #endif // USE_SORT_BUFFER +layout(set = 2, binding = 0, std430) restrict readonly buffer TrailBindPoses { + mat4 data[]; +} +trail_bind_poses; + layout(push_constant, binding = 0, std430) uniform Params { vec3 sort_direction; uint total_particles; + + uint trail_size; + uint trail_total; + float frame_delta; + float frame_remainder; + + vec3 align_up; + uint align_mode; } params; +#define TRANSFORM_ALIGN_DISABLED 0 +#define TRANSFORM_ALIGN_Z_BILLBOARD 1 +#define TRANSFORM_ALIGN_Y_TO_VELOCITY 2 +#define TRANSFORM_ALIGN_Z_BILLBOARD_Y_TO_VELOCITY 3 + void main() { #ifdef MODE_FILL_SORT_BUFFER @@ -47,7 +69,11 @@ void main() { return; //discard } - sort_buffer.data[particle].x = dot(params.sort_direction, particles.data[particle].xform[3].xyz); + uint src_particle = particle; + if (params.trail_size > 1) { + src_particle = src_particle * params.trail_size + params.trail_size / 2; //use trail center for sorting + } + sort_buffer.data[particle].x = dot(params.sort_direction, particles.data[src_particle].xform[3].xyz); sort_buffer.data[particle].y = float(particle); #endif @@ -61,13 +87,78 @@ void main() { } #ifdef USE_SORT_BUFFER - particle = uint(sort_buffer.data[particle].y); //use index from sort buffer + + if (params.trail_size > 1) { + particle = uint(sort_buffer.data[particle / params.trail_size].y) + (particle % params.trail_size); + } else { + particle = uint(sort_buffer.data[particle].y); //use index from sort buffer + } #endif mat4 txform; - if (particles.data[particle].is_active) { - txform = transpose(particles.data[particle].xform); + 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 + // 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; + + txform[3].xyz -= particles.data[particle].velocity * natural_ofs; + } + + switch (params.align_mode) { + case TRANSFORM_ALIGN_DISABLED: { + } break; //nothing + case TRANSFORM_ALIGN_Z_BILLBOARD: { + mat3 local = mat3(normalize(cross(params.align_up, params.sort_direction)), params.align_up, params.sort_direction); + local = local * mat3(txform); + txform[0].xyz = local[0]; + txform[1].xyz = local[1]; + txform[2].xyz = local[2]; + + } break; + case TRANSFORM_ALIGN_Y_TO_VELOCITY: { + vec3 v = particles.data[particle].velocity; + float s = (length(txform[0]) + length(txform[1]) + length(txform[2])) / 3.0; + if (length(v) > 0.0) { + txform[1].xyz = normalize(v); + } else { + txform[1].xyz = normalize(txform[1].xyz); + } + + txform[0].xyz = normalize(cross(txform[1].xyz, txform[2].xyz)); + txform[2].xyz = vec3(0.0, 0.0, 1.0) * s; + txform[0].xyz *= s; + txform[1].xyz *= s; + } break; + case TRANSFORM_ALIGN_Z_BILLBOARD_Y_TO_VELOCITY: { + vec3 v = particles.data[particle].velocity; + vec3 sv = v - params.sort_direction * dot(params.sort_direction, v); //screen velocity + float s = (length(txform[0]) + length(txform[1]) + length(txform[2])) / 3.0; + + if (length(sv) == 0) { + sv = params.align_up; + } + + sv = normalize(sv); + + txform[0].xyz = normalize(cross(sv, params.sort_direction)) * s; + txform[1].xyz = sv * s; + txform[2].xyz = params.sort_direction * s; + + } break; + } + + txform[3].xyz += particles.data[particle].velocity * params.frame_remainder; + + if (params.trail_size > 1) { + uint part_ofs = particle % params.trail_size; + txform = txform * trail_bind_poses.data[part_ofs]; + } + + txform = transpose(txform); } else { txform = mat4(vec4(0.0), vec4(0.0), vec4(0.0), vec4(0.0)); //zero scale, becomes invisible } diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl index fce17c47e8..f0dbe4eedf 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl @@ -48,11 +48,11 @@ layout(location = 8) in vec4 custom2_attrib; layout(location = 9) in vec4 custom3_attrib; #endif -#if defined(BONES_USED) +#if defined(BONES_USED) || defined(USE_PARTICLE_TRAILS) layout(location = 10) in uvec4 bone_attrib; #endif -#if defined(WEIGHTS_USED) +#if defined(WEIGHTS_USED) || defined(USE_PARTICLE_TRAILS) layout(location = 11) in vec4 weight_attrib; #endif @@ -125,10 +125,72 @@ void main() { if (is_multimesh) { //multimesh, instances are for it - uint offset = (instances.data[instance_index].flags >> INSTANCE_FLAGS_MULTIMESH_STRIDE_SHIFT) & INSTANCE_FLAGS_MULTIMESH_STRIDE_MASK; - offset *= gl_InstanceIndex; mat4 matrix; + +#ifdef USE_PARTICLE_TRAILS + uint trail_size = (instances.data[instance_index].flags >> INSTANCE_FLAGS_PARTICLE_TRAIL_SHIFT) & INSTANCE_FLAGS_PARTICLE_TRAIL_MASK; + uint stride = 3 + 1 + 1; //particles always uses this format + + uint offset = trail_size * stride * gl_InstanceIndex; + +#ifdef COLOR_USED + vec4 pcolor; +#endif + { + uint boffset = offset + bone_attrib.x * stride; + matrix = mat4(transforms.data[boffset + 0], transforms.data[boffset + 1], transforms.data[boffset + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weight_attrib.x; +#ifdef COLOR_USED + pcolor = transforms.data[boffset + 3] * weight_attrib.x; +#endif + } + if (weight_attrib.y > 0.001) { + uint boffset = offset + bone_attrib.y * stride; + matrix += mat4(transforms.data[boffset + 0], transforms.data[boffset + 1], transforms.data[boffset + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weight_attrib.y; +#ifdef COLOR_USED + pcolor += transforms.data[boffset + 3] * weight_attrib.y; +#endif + } + if (weight_attrib.z > 0.001) { + uint boffset = offset + bone_attrib.z * stride; + matrix += mat4(transforms.data[boffset + 0], transforms.data[boffset + 1], transforms.data[boffset + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weight_attrib.z; +#ifdef COLOR_USED + pcolor += transforms.data[boffset + 3] * weight_attrib.z; +#endif + } + if (weight_attrib.w > 0.001) { + uint boffset = offset + bone_attrib.w * stride; + matrix += mat4(transforms.data[boffset + 0], transforms.data[boffset + 1], transforms.data[boffset + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weight_attrib.w; +#ifdef COLOR_USED + pcolor += transforms.data[boffset + 3] * weight_attrib.w; +#endif + } + + instance_custom = transforms.data[offset + 4]; + +#ifdef COLOR_USED + color_interp *= pcolor; +#endif + +#else + uint stride = 0; + { + //TODO implement a small lookup table for the stride + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_FORMAT_2D)) { + stride += 2; + } else { + stride += 3; + } + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_HAS_COLOR)) { + stride += 1; + } + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_HAS_CUSTOM_DATA)) { + stride += 1; + } + } + + uint offset = stride * gl_InstanceIndex; + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_FORMAT_2D)) { matrix = mat4(transforms.data[offset + 0], transforms.data[offset + 1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0)); offset += 2; @@ -148,6 +210,7 @@ void main() { instance_custom = transforms.data[offset]; } +#endif //transpose matrix = transpose(matrix); world_matrix = world_matrix * matrix; @@ -165,32 +228,6 @@ void main() { vec3 binormal = normalize(cross(normal, tangent) * binormalf); #endif -#if 0 - if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_SKELETON)) { - //multimesh, instances are for it - - uvec2 bones_01 = uvec2(bone_attrib.x & 0xFFFF, bone_attrib.x >> 16) * 3; - uvec2 bones_23 = uvec2(bone_attrib.y & 0xFFFF, bone_attrib.y >> 16) * 3; - vec2 weights_01 = unpackUnorm2x16(bone_attrib.z); - vec2 weights_23 = unpackUnorm2x16(bone_attrib.w); - - mat4 m = mat4(transforms.data[bones_01.x], transforms.data[bones_01.x + 1], transforms.data[bones_01.x + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weights_01.x; - m += mat4(transforms.data[bones_01.y], transforms.data[bones_01.y + 1], transforms.data[bones_01.y + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weights_01.y; - m += mat4(transforms.data[bones_23.x], transforms.data[bones_23.x + 1], transforms.data[bones_23.x + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weights_23.x; - m += mat4(transforms.data[bones_23.y], transforms.data[bones_23.y + 1], transforms.data[bones_23.y + 2], vec4(0.0, 0.0, 0.0, 1.0)) * weights_23.y; - - //reverse order because its transposed - vertex = (vec4(vertex, 1.0) * m).xyz; - normal = (vec4(normal, 0.0) * m).xyz; - -#if defined(TANGENT_USED) || defined(NORMAL_MAP_USED) || defined(LIGHT_ANISOTROPY_USED) - - tangent = (vec4(tangent, 0.0) * m).xyz; - binormal = (vec4(binormal, 0.0) * m).xyz; -#endif - } -#endif - #ifdef UV_USED uv_interp = uv_attrib; #endif 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 48e331714d..83423cdc24 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl @@ -61,12 +61,11 @@ layout(set = 0, binding = 2) uniform sampler shadow_sampler; #define INSTANCE_FLAGS_MULTIMESH_FORMAT_2D (1 << 13) #define INSTANCE_FLAGS_MULTIMESH_HAS_COLOR (1 << 14) #define INSTANCE_FLAGS_MULTIMESH_HAS_CUSTOM_DATA (1 << 15) -#define INSTANCE_FLAGS_MULTIMESH_STRIDE_SHIFT 16 +#define INSTANCE_FLAGS_PARTICLE_TRAIL_SHIFT 16 //3 bits of stride -#define INSTANCE_FLAGS_MULTIMESH_STRIDE_MASK 0x7 +#define INSTANCE_FLAGS_PARTICLE_TRAIL_MASK 0xFF -#define INSTANCE_FLAGS_SKELETON (1 << 19) -#define INSTANCE_FLAGS_NON_UNIFORM_SCALE (1 << 20) +#define INSTANCE_FLAGS_NON_UNIFORM_SCALE (1 << 24) layout(set = 0, binding = 3, std430) restrict readonly buffer OmniLights { LightData data[]; |