diff options
Diffstat (limited to 'servers/rendering/renderer_rd')
19 files changed, 352 insertions, 123 deletions
diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.cpp b/servers/rendering/renderer_rd/cluster_builder_rd.cpp index 64950c6991..7554e478bb 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.cpp +++ b/servers/rendering/renderer_rd/cluster_builder_rd.cpp @@ -58,6 +58,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { afs.push_back(RD::AttachmentFormat()); afs.write[0].usage_flags = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; fb_format = RD::get_singleton()->framebuffer_format_create(afs); + blend_state = RD::PipelineColorBlendState::create_blend(); defines = "\n#define USE_ATTACHMENT\n"; } @@ -94,7 +95,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { 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 }; static const uint32_t icosphere_triangle_count = 80; - static const uint32_t icosphere_triangle_indices[icosphere_triangle_count * 3] = { + static const uint16_t icosphere_triangle_indices[icosphere_triangle_count * 3] = { 0, 13, 12, 1, 13, 15, 0, 12, 17, 0, 17, 19, 0, 19, 16, 1, 15, 22, 2, 14, 24, 3, 18, 26, 4, 20, 28, 5, 21, 30, 1, 22, 25, 2, 24, 27, 3, 26, 29, 4, 28, 31, 5, 30, 23, 6, 32, 37, 7, 33, 39, 8, 34, 40, 9, 35, 41, 10, 36, 38, 38, 41, 11, 38, 36, 41, 36, 9, 41, 41, 40, 11, 41, 35, 40, 35, 8, 40, 40, 39, 11, 40, 34, 39, 34, 7, 39, 39, 37, 11, 39, 33, 37, 33, 6, 37, 37, 38, 11, 37, 32, 38, 32, 10, 38, 23, 36, 10, 23, 30, 36, 30, 9, 36, 31, 35, 9, 31, 28, 35, 28, 8, 35, 29, 34, 8, 29, 26, 34, 26, 7, 34, 27, 33, 7, 27, 24, 33, 24, 6, 33, 25, 32, 6, 25, 22, 32, 22, 10, 32, 30, 31, 9, 30, 21, 31, 21, 4, 31, 28, 29, 8, 28, 20, 29, 20, 3, 29, 26, 27, 7, 26, 18, 27, 18, 2, 27, 24, 25, 6, 24, 14, 25, 14, 1, 25, 22, 23, 10, 22, 15, 23, 15, 5, 23, 16, 21, 5, 16, 19, 21, 19, 4, 21, 19, 20, 4, 19, 17, 20, 17, 3, 20, 17, 18, 3, 17, 12, 18, 12, 2, 18, 15, 16, 5, 15, 13, 16, 13, 0, 16, 12, 14, 2, 12, 13, 14, 13, 1, 14 }; @@ -105,10 +106,10 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { sphere_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); Vector<uint8_t> index_data; - index_data.resize(sizeof(uint32_t) * icosphere_triangle_count * 3); + index_data.resize(sizeof(uint16_t) * icosphere_triangle_count * 3); memcpy(index_data.ptrw(), icosphere_triangle_indices, index_data.size()); - sphere_index_buffer = RD::get_singleton()->index_buffer_create(icosphere_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + sphere_index_buffer = RD::get_singleton()->index_buffer_create(icosphere_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT16, index_data); Vector<RID> buffers; buffers.push_back(sphere_vertex_buffer); @@ -138,7 +139,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { 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 }; static const uint32_t cone_triangle_count = 62; - static const uint32_t cone_triangle_indices[cone_triangle_count * 3] = { + static const uint16_t cone_triangle_indices[cone_triangle_count * 3] = { 0, 23, 1, 1, 23, 2, 2, 23, 3, 3, 23, 4, 4, 23, 5, 5, 23, 6, 6, 23, 7, 7, 23, 8, 8, 23, 9, 9, 23, 10, 10, 23, 11, 11, 23, 12, 12, 23, 13, 13, 23, 14, 14, 23, 15, 15, 23, 16, 16, 23, 17, 17, 23, 18, 18, 23, 19, 19, 23, 20, 20, 23, 21, 21, 23, 22, 22, 23, 24, 24, 23, 25, 25, 23, 26, 26, 23, 27, 27, 23, 28, 28, 23, 29, 29, 23, 30, 30, 23, 31, 31, 23, 32, 32, 23, 0, 7, 15, 24, 32, 0, 1, 1, 2, 3, 3, 4, 5, 5, 6, 3, 6, 7, 3, 7, 8, 9, 9, 10, 7, 10, 11, 7, 11, 12, 15, 12, 13, 15, 13, 14, 15, 15, 16, 17, 17, 18, 19, 19, 20, 24, 20, 21, 24, 21, 22, 24, 24, 25, 26, 26, 27, 28, 28, 29, 30, 30, 31, 32, 32, 1, 3, 15, 17, 24, 17, 19, 24, 24, 26, 32, 26, 28, 32, 28, 30, 32, 32, 3, 7, 7, 11, 15, 32, 7, 24 }; @@ -149,10 +150,10 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { cone_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); Vector<uint8_t> index_data; - index_data.resize(sizeof(uint32_t) * cone_triangle_count * 3); + index_data.resize(sizeof(uint16_t) * cone_triangle_count * 3); memcpy(index_data.ptrw(), cone_triangle_indices, index_data.size()); - cone_index_buffer = RD::get_singleton()->index_buffer_create(cone_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + cone_index_buffer = RD::get_singleton()->index_buffer_create(cone_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT16, index_data); Vector<RID> buffers; buffers.push_back(cone_vertex_buffer); @@ -192,7 +193,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { -1, -1, -1, -1, -1, 1, -1, 1, -1, -1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, -1, 1, 1, 1 }; static const uint32_t box_triangle_count = 12; - static const uint32_t box_triangle_indices[box_triangle_count * 3] = { + static const uint16_t box_triangle_indices[box_triangle_count * 3] = { 1, 2, 0, 3, 6, 2, 7, 4, 6, 5, 0, 4, 6, 0, 2, 3, 5, 7, 1, 3, 2, 3, 7, 6, 7, 5, 4, 5, 1, 0, 6, 4, 0, 3, 1, 5 }; @@ -203,10 +204,10 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { box_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); Vector<uint8_t> index_data; - index_data.resize(sizeof(uint32_t) * box_triangle_count * 3); + index_data.resize(sizeof(uint16_t) * box_triangle_count * 3); memcpy(index_data.ptrw(), box_triangle_indices, index_data.size()); - box_index_buffer = RD::get_singleton()->index_buffer_create(box_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + box_index_buffer = RD::get_singleton()->index_buffer_create(box_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT16, index_data); Vector<RID> buffers; buffers.push_back(box_vertex_buffer); diff --git a/servers/rendering/renderer_rd/effects/debug_effects.cpp b/servers/rendering/renderer_rd/effects/debug_effects.cpp index 8cd3c22483..357d035ae9 100644 --- a/servers/rendering/renderer_rd/effects/debug_effects.cpp +++ b/servers/rendering/renderer_rd/effects/debug_effects.cpp @@ -51,6 +51,17 @@ DebugEffects::DebugEffects() { raster_state.wireframe = true; shadow_frustum.pipelines[SFP_WIREFRAME].setup(shadow_frustum.shader.version_get_shader(shadow_frustum.shader_version, 0), RD::RENDER_PRIMITIVE_LINES, raster_state, RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); } + + { + // Motion Vectors debug shader. + Vector<String> modes; + modes.push_back(""); + + motion_vectors.shader.initialize(modes); + motion_vectors.shader_version = motion_vectors.shader.version_create(); + + motion_vectors.pipeline.setup(motion_vectors.shader.version_get_shader(motion_vectors.shader_version, 0), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_blend(), 0); + } } void DebugEffects::_create_frustum_arrays() { @@ -74,7 +85,7 @@ void DebugEffects::_create_frustum_arrays() { } if (frustum.index_buffer.is_null()) { - uint32_t indices[6 * 2 * 3] = { + uint16_t indices[6 * 2 * 3] = { // Far 0, 1, 2, // FLT, FLB, FRT 1, 3, 2, // FLB, FRB, FRT @@ -100,19 +111,19 @@ void DebugEffects::_create_frustum_arrays() { data.resize(6 * 2 * 3 * 4); { uint8_t *w = data.ptrw(); - int *p32 = (int *)w; + uint16_t *p16 = (uint16_t *)w; for (int i = 0; i < 6 * 2 * 3; i++) { - *p32 = indices[i]; - p32++; + *p16 = indices[i]; + p16++; } } - frustum.index_buffer = RD::get_singleton()->index_buffer_create(6 * 2 * 3, RenderingDevice::INDEX_BUFFER_FORMAT_UINT32, data); + frustum.index_buffer = RD::get_singleton()->index_buffer_create(6 * 2 * 3, RenderingDevice::INDEX_BUFFER_FORMAT_UINT16, data); frustum.index_array = RD::get_singleton()->index_array_create(frustum.index_buffer, 0, 6 * 2 * 3); } if (frustum.lines_buffer.is_null()) { - uint32_t indices[12 * 2] = { + uint16_t indices[12 * 2] = { 0, 1, // FLT - FLB 1, 3, // FLB - FRB 3, 2, // FRB - FRT @@ -134,14 +145,14 @@ void DebugEffects::_create_frustum_arrays() { data.resize(12 * 2 * 4); { uint8_t *w = data.ptrw(); - int *p32 = (int *)w; + uint16_t *p16 = (uint16_t *)w; for (int i = 0; i < 12 * 2; i++) { - *p32 = indices[i]; - p32++; + *p16 = indices[i]; + p16++; } } - frustum.lines_buffer = RD::get_singleton()->index_buffer_create(12 * 2, RenderingDevice::INDEX_BUFFER_FORMAT_UINT32, data); + frustum.lines_buffer = RD::get_singleton()->index_buffer_create(12 * 2, RenderingDevice::INDEX_BUFFER_FORMAT_UINT16, data); frustum.lines_array = RD::get_singleton()->index_array_create(frustum.lines_buffer, 0, 12 * 2); } } @@ -163,6 +174,8 @@ DebugEffects::~DebugEffects() { if (frustum.lines_buffer.is_valid()) { RD::get_singleton()->free(frustum.lines_buffer); // Array gets freed as dependency. } + + motion_vectors.shader.version_free(motion_vectors.shader_version); } void DebugEffects::draw_shadow_frustum(RID p_light, const Projection &p_cam_projection, const Transform3D &p_cam_transform, RID p_dest_fb, const Rect2 p_rect) { @@ -326,3 +339,26 @@ void DebugEffects::draw_shadow_frustum(RID p_light, const Projection &p_cam_proj } } } + +void DebugEffects::draw_motion_vectors(RID p_velocity, RID p_dest_fb, Size2i p_velocity_size) { + MaterialStorage *material_storage = MaterialStorage::get_singleton(); + ERR_FAIL_NULL(material_storage); + + UniformSetCacheRD *uniform_set_cache = UniformSetCacheRD::get_singleton(); + ERR_FAIL_NULL(uniform_set_cache); + + RID default_sampler = material_storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + RD::Uniform u_source_velocity(RD::UNIFORM_TYPE_SAMPLER_WITH_TEXTURE, 0, Vector<RID>({ default_sampler, p_velocity })); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_fb, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, motion_vectors.pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_fb), false, RD::get_singleton()->draw_list_get_current_pass())); + + motion_vectors.push_constant.velocity_resolution[0] = p_velocity_size.width; + motion_vectors.push_constant.velocity_resolution[1] = p_velocity_size.height; + + RID shader = motion_vectors.shader.version_get_shader(motion_vectors.shader_version, 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, uniform_set_cache->get_cache(shader, 0, u_source_velocity), 0); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &motion_vectors.push_constant, sizeof(MotionVectorsPushConstant)); + RD::get_singleton()->draw_list_draw(draw_list, false, 1u, 3u); + RD::get_singleton()->draw_list_end(); +} diff --git a/servers/rendering/renderer_rd/effects/debug_effects.h b/servers/rendering/renderer_rd/effects/debug_effects.h index 21b7b03f84..ae32d94912 100644 --- a/servers/rendering/renderer_rd/effects/debug_effects.h +++ b/servers/rendering/renderer_rd/effects/debug_effects.h @@ -32,6 +32,7 @@ #define DEBUG_EFFECTS_RD_H #include "servers/rendering/renderer_rd/pipeline_cache_rd.h" +#include "servers/rendering/renderer_rd/shaders/effects/motion_vectors.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/effects/shadow_frustum.glsl.gen.h" #include "servers/rendering/renderer_scene_render.h" @@ -70,6 +71,18 @@ private: PipelineCacheRD pipelines[SFP_MAX]; } shadow_frustum; + struct MotionVectorsPushConstant { + float velocity_resolution[2]; + float pad[2]; + }; + + struct { + MotionVectorsShaderRD shader; + RID shader_version; + PipelineCacheRD pipeline; + MotionVectorsPushConstant push_constant; + } motion_vectors; + void _create_frustum_arrays(); protected: @@ -78,6 +91,7 @@ public: ~DebugEffects(); void draw_shadow_frustum(RID p_light, const Projection &p_cam_projection, const Transform3D &p_cam_transform, RID p_dest_fb, const Rect2 p_rect); + void draw_motion_vectors(RID p_velocity, RID p_dest_fb, Size2i p_velocity_size); }; } // namespace RendererRD diff --git a/servers/rendering/renderer_rd/environment/gi.cpp b/servers/rendering/renderer_rd/environment/gi.cpp index 991ccf984e..d623553273 100644 --- a/servers/rendering/renderer_rd/environment/gi.cpp +++ b/servers/rendering/renderer_rd/environment/gi.cpp @@ -395,6 +395,16 @@ Dependency *GI::voxel_gi_get_dependency(RID p_voxel_gi) const { //////////////////////////////////////////////////////////////////////////////// // SDFGI +static RID create_clear_texture(const RD::TextureFormat &p_format, const String &p_name) { + RID texture = RD::get_singleton()->texture_create(p_format, RD::TextureView()); + ERR_FAIL_COND_V_MSG(texture.is_null(), RID(), String("Cannot create texture: ") + p_name); + + RD::get_singleton()->set_resource_name(texture, p_name); + RD::get_singleton()->texture_clear(texture, Color(0, 0, 0, 0), 0, p_format.mipmaps, 0, p_format.array_layers); + + return texture; +} + void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size, GI *p_gi) { RendererRD::TextureStorage *texture_storage = RendererRD::TextureStorage::get_singleton(); RendererRD::MaterialStorage *material_storage = RendererRD::MaterialStorage::get_singleton(); @@ -424,39 +434,31 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re { RD::TextureFormat tf_render = tf_sdf; tf_render.format = RD::DATA_FORMAT_R16_UINT; - render_albedo = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_albedo, "VoxelGI Render Albedo"); + render_albedo = create_clear_texture(tf_render, "SDFGI Render Albedo"); + tf_render.format = RD::DATA_FORMAT_R32_UINT; - render_emission = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_emission, "VoxelGI Render Emission"); - render_emission_aniso = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_emission_aniso, "VoxelGI Render Emission Aniso"); + render_emission = create_clear_texture(tf_render, "SDFGI Render Emission"); + render_emission_aniso = create_clear_texture(tf_render, "SDFGI Render Emission Aniso"); tf_render.format = RD::DATA_FORMAT_R8_UNORM; //at least its easy to visualize for (int i = 0; i < 8; i++) { - render_occlusion[i] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_occlusion[i], String("VoxelGI Render Occlusion ") + itos(i)); + render_occlusion[i] = create_clear_texture(tf_render, String("SDFGI Render Occlusion ") + itos(i)); } tf_render.format = RD::DATA_FORMAT_R32_UINT; - render_geom_facing = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_geom_facing, "VoxelGI Render Geometry Facing"); + render_geom_facing = create_clear_texture(tf_render, "SDFGI Render Geometry Facing"); tf_render.format = RD::DATA_FORMAT_R8G8B8A8_UINT; - render_sdf[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_sdf[0], "VoxelGI Render SDF 0"); - render_sdf[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_sdf[1], "VoxelGI Render SDF 1"); + render_sdf[0] = create_clear_texture(tf_render, "SDFGI Render SDF 0"); + render_sdf[1] = create_clear_texture(tf_render, "SDFGI Render SDF 1"); tf_render.width /= 2; tf_render.height /= 2; tf_render.depth /= 2; - render_sdf_half[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_sdf_half[0], "VoxelGI Render SDF Half 0"); - render_sdf_half[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - RD::get_singleton()->set_resource_name(render_sdf_half[1], "VoxelGI Render SDF Half 1"); + render_sdf_half[0] = create_clear_texture(tf_render, "SDFGI Render SDF Half 0"); + render_sdf_half[1] = create_clear_texture(tf_render, "SDFGI Render SDF Half 1"); } RD::TextureFormat tf_occlusion = tf_sdf; @@ -496,10 +498,8 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re tf_probe_average.format = RD::DATA_FORMAT_R32G32B32A32_SINT; //signed integer because SH are signed tf_probe_average.texture_type = RD::TEXTURE_TYPE_2D; - lightprobe_history_scroll = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); - RD::get_singleton()->set_resource_name(lightprobe_history_scroll, "VoxelGI LightProbe History Scroll"); - lightprobe_average_scroll = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); - RD::get_singleton()->set_resource_name(lightprobe_average_scroll, "VoxelGI LightProbe Average Scroll"); + lightprobe_history_scroll = create_clear_texture(tf_probe_history, "SDFGI LightProbe History Scroll"); + lightprobe_average_scroll = create_clear_texture(tf_probe_average, "SDFGI LightProbe Average Scroll"); { //octahedral lightprobes @@ -512,8 +512,7 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re tf_octprobes.shareable_formats.push_back(RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32); //lightprobe texture is an octahedral texture - lightprobe_data = RD::get_singleton()->texture_create(tf_octprobes, RD::TextureView()); - RD::get_singleton()->set_resource_name(lightprobe_data, "VoxelGI LightProbe Data"); + lightprobe_data = create_clear_texture(tf_octprobes, "SDFGI LightProbe Data"); RD::TextureView tv; tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; lightprobe_texture = RD::get_singleton()->texture_create_shared(tv, lightprobe_data); @@ -526,14 +525,12 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re tf_ambient.height = probe_axis_count; tf_ambient.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; //lightprobe texture is an octahedral texture - ambient_texture = RD::get_singleton()->texture_create(tf_ambient, RD::TextureView()); - RD::get_singleton()->set_resource_name(ambient_texture, "VoxelGI Ambient Texture"); + ambient_texture = create_clear_texture(tf_ambient, "SDFGI Ambient Texture"); } cascades_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(SDFGI::Cascade::UBO) * SDFGI::MAX_CASCADES); - occlusion_data = RD::get_singleton()->texture_create(tf_occlusion, RD::TextureView()); - RD::get_singleton()->set_resource_name(occlusion_data, "VoxelGI Occlusion Data"); + occlusion_data = create_clear_texture(tf_occlusion, "SDFGI Occlusion Data"); { RD::TextureView tv; tv.format_override = RD::DATA_FORMAT_R4G4B4A4_UNORM_PACK16; @@ -543,25 +540,17 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re for (SDFGI::Cascade &cascade : cascades) { /* 3D Textures */ - cascade.sdf_tex = RD::get_singleton()->texture_create(tf_sdf, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.sdf_tex, "VoxelGI Cascade SDF Texture"); + cascade.sdf_tex = create_clear_texture(tf_sdf, "SDFGI Cascade SDF Texture"); - cascade.light_data = RD::get_singleton()->texture_create(tf_light, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.light_data, "VoxelGI Cascade Light Data"); + cascade.light_data = create_clear_texture(tf_light, "SDFGI Cascade Light Data"); - cascade.light_aniso_0_tex = RD::get_singleton()->texture_create(tf_aniso0, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.light_aniso_0_tex, "VoxelGI Cascade Light Aniso 0 Texture"); - cascade.light_aniso_1_tex = RD::get_singleton()->texture_create(tf_aniso1, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.light_aniso_1_tex, "VoxelGI Cascade Light Aniso 1 Texture"); + cascade.light_aniso_0_tex = create_clear_texture(tf_aniso0, "SDFGI Cascade Light Aniso 0 Texture"); + cascade.light_aniso_1_tex = create_clear_texture(tf_aniso1, "SDFGI Cascade Light Aniso 1 Texture"); { RD::TextureView tv; tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; cascade.light_tex = RD::get_singleton()->texture_create_shared(tv, cascade.light_data); - - RD::get_singleton()->texture_clear(cascade.light_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); - RD::get_singleton()->texture_clear(cascade.light_aniso_0_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); - RD::get_singleton()->texture_clear(cascade.light_aniso_1_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); } cascade.cell_size = base_cell_size; @@ -579,11 +568,11 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re /* Probe History */ cascade.lightprobe_history_tex = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.lightprobe_history_tex, "VoxelGI Cascade LightProbe History Texture"); + RD::get_singleton()->set_resource_name(cascade.lightprobe_history_tex, "SDFGI Cascade LightProbe History Texture"); RD::get_singleton()->texture_clear(cascade.lightprobe_history_tex, Color(0, 0, 0, 0), 0, 1, 0, tf_probe_history.array_layers); //needs to be cleared for average to work cascade.lightprobe_average_tex = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); - RD::get_singleton()->set_resource_name(cascade.lightprobe_average_tex, "VoxelGI Cascade LightProbe Average Texture"); + RD::get_singleton()->set_resource_name(cascade.lightprobe_average_tex, "SDFGI Cascade LightProbe Average Texture"); RD::get_singleton()->texture_clear(cascade.lightprobe_average_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); //needs to be cleared for average to work /* Buffers */ 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 c98752fb39..2e64946fd2 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -252,6 +252,7 @@ bool RenderForwardClustered::free(RID p_rid) { template <RenderForwardClustered::PassMode p_pass_mode, uint32_t p_color_pass_flags> void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { RendererRD::MeshStorage *mesh_storage = RendererRD::MeshStorage::get_singleton(); + RendererRD::ParticlesStorage *particles_storage = RendererRD::ParticlesStorage::get_singleton(); RD::DrawListID draw_list = p_draw_list; RD::FramebufferFormatID framebuffer_format = p_framebuffer_Format; @@ -477,7 +478,9 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p prev_material_uniform_set = material_uniform_set; } - if ((surf->owner->base_flags & (INSTANCE_DATA_FLAG_MULTIMESH | INSTANCE_DATA_FLAG_PARTICLES)) == INSTANCE_DATA_FLAG_MULTIMESH) { + if (surf->owner->base_flags & INSTANCE_DATA_FLAG_PARTICLES) { + particles_storage->particles_get_instance_buffer_motion_vectors_offsets(surf->owner->data->base, push_constant.multimesh_motion_vectors_current_offset, push_constant.multimesh_motion_vectors_previous_offset); + } else if (surf->owner->base_flags & INSTANCE_DATA_FLAG_MULTIMESH) { mesh_storage->_multimesh_get_motion_vectors_offsets(surf->owner->data->base, push_constant.multimesh_motion_vectors_current_offset, push_constant.multimesh_motion_vectors_previous_offset); } else { push_constant.multimesh_motion_vectors_current_offset = 0; @@ -3716,6 +3719,10 @@ void RenderForwardClustered::_geometry_instance_update(RenderGeometryInstance *p // Particles haven't been cleared or updated, update once now to ensure they are ready to render. particles_storage->update_particles(); } + + if (ginstance->data->dirty_dependencies) { + particles_storage->particles_update_dependency(ginstance->data->base, &ginstance->data->dependency_tracker); + } } else if (ginstance->data->base_type == RS::INSTANCE_MESH) { if (mesh_storage->skeleton_is_valid(ginstance->data->skeleton)) { ginstance->transforms_uniform_set = mesh_storage->skeleton_get_3d_uniform_set(ginstance->data->skeleton, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); @@ -3755,6 +3762,7 @@ void RenderForwardClustered::_geometry_instance_dependency_changed(Dependency::D case Dependency::DEPENDENCY_CHANGED_MATERIAL: case Dependency::DEPENDENCY_CHANGED_MESH: case Dependency::DEPENDENCY_CHANGED_PARTICLES: + case Dependency::DEPENDENCY_CHANGED_PARTICLES_INSTANCES: case Dependency::DEPENDENCY_CHANGED_MULTIMESH: case Dependency::DEPENDENCY_CHANGED_SKELETON_DATA: { static_cast<RenderGeometryInstance *>(p_tracker->userdata)->_mark_dirty(); 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 12f8f6a366..509072bbec 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 @@ -707,6 +707,7 @@ void SceneShaderForwardClustered::init(const String p_defines) { actions.render_mode_defines["shadow_to_opacity"] = "#define USE_SHADOW_TO_OPACITY\n"; actions.render_mode_defines["unshaded"] = "#define MODE_UNSHADED\n"; actions.render_mode_defines["debug_shadow_splits"] = "#define DEBUG_DRAW_PSSM_SPLITS\n"; + actions.render_mode_defines["fog_disabled"] = "#define FOG_DISABLED\n"; actions.base_texture_binding_index = 1; actions.texture_layout_set = RenderForwardClustered::MATERIAL_UNIFORM_SET; 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 90d770399e..3f01ffa3c9 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp @@ -752,7 +752,8 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color bool draw_sky = false; bool draw_sky_fog_only = false; // We invert luminance_multiplier for sky so that we can combine it with exposure value. - float sky_energy_multiplier = 1.0 / _render_buffers_get_luminance_multiplier(); + float inverse_luminance_multiplier = 1.0 / _render_buffers_get_luminance_multiplier(); + float sky_energy_multiplier = inverse_luminance_multiplier; Color clear_color = p_default_bg_color; bool keep_color = false; @@ -899,14 +900,14 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color // regular forward for now Vector<Color> c; { - Color cc = clear_color.srgb_to_linear(); + Color cc = clear_color.srgb_to_linear() * inverse_luminance_multiplier; if (rb_data.is_valid()) { cc.a = 0; // For transparent viewport backgrounds. } c.push_back(cc); // Our render buffer. if (rb_data.is_valid()) { if (p_render_data->render_buffers->get_msaa_3d() != RS::VIEWPORT_MSAA_DISABLED) { - c.push_back(clear_color.srgb_to_linear()); // Our resolve buffer. + c.push_back(clear_color.srgb_to_linear() * inverse_luminance_multiplier); // Our resolve buffer. } if (using_subpass_post_process) { c.push_back(Color()); // Our 2D buffer we're copying into. @@ -2614,6 +2615,10 @@ void RenderForwardMobile::_geometry_instance_update(RenderGeometryInstance *p_ge // Particles haven't been cleared or updated, update once now to ensure they are ready to render. particles_storage->update_particles(); } + + if (ginstance->data->dirty_dependencies) { + particles_storage->particles_update_dependency(ginstance->data->base, &ginstance->data->dependency_tracker); + } } else if (ginstance->data->base_type == RS::INSTANCE_MESH) { if (mesh_storage->skeleton_is_valid(ginstance->data->skeleton)) { ginstance->transforms_uniform_set = mesh_storage->skeleton_get_3d_uniform_set(ginstance->data->skeleton, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); diff --git a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp index 32d2289f75..ffbe7f7e59 100644 --- a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp @@ -610,6 +610,7 @@ void SceneShaderForwardMobile::init(const String p_defines) { actions.render_mode_defines["shadow_to_opacity"] = "#define USE_SHADOW_TO_OPACITY\n"; actions.render_mode_defines["unshaded"] = "#define MODE_UNSHADED\n"; actions.render_mode_defines["debug_shadow_splits"] = "#define DEBUG_DRAW_PSSM_SPLITS\n"; + actions.render_mode_defines["fog_disabled"] = "#define FOG_DISABLED\n"; actions.base_texture_binding_index = 1; actions.texture_layout_set = RenderForwardMobile::MATERIAL_UNIFORM_SET; diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index d8c035a51c..b1ad7e16ed 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -2587,18 +2587,18 @@ RendererCanvasRenderRD::RendererCanvasRenderRD() { { // default index buffer Vector<uint8_t> pv; - pv.resize(6 * 4); + pv.resize(6 * 2); { uint8_t *w = pv.ptrw(); - int *p32 = (int *)w; - p32[0] = 0; - p32[1] = 1; - p32[2] = 2; - p32[3] = 0; - p32[4] = 2; - p32[5] = 3; + uint16_t *p16 = (uint16_t *)w; + p16[0] = 0; + p16[1] = 1; + p16[2] = 2; + p16[3] = 0; + p16[4] = 2; + p16[5] = 3; } - shader.quad_index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT32, pv); + shader.quad_index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT16, pv); shader.quad_index_array = RD::get_singleton()->index_array_create(shader.quad_index_buffer, 0, 6); } diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp index 4ccd2aa322..7eb8cbd02f 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp @@ -126,18 +126,18 @@ void RendererCompositorRD::initialize() { //create index array for copy shader Vector<uint8_t> pv; - pv.resize(6 * 4); + pv.resize(6 * 2); { uint8_t *w = pv.ptrw(); - int *p32 = (int *)w; - p32[0] = 0; - p32[1] = 1; - p32[2] = 2; - p32[3] = 0; - p32[4] = 2; - p32[5] = 3; + uint16_t *p16 = (uint16_t *)w; + p16[0] = 0; + p16[1] = 1; + p16[2] = 2; + p16[3] = 0; + p16[4] = 2; + p16[5] = 3; } - blit.index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT32, pv); + blit.index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT16, pv); blit.array = RD::get_singleton()->index_array_create(blit.index_buffer, 0, 6); blit.sampler = RD::get_singleton()->sampler_create(RD::SamplerState()); diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp index 9d4d266a7a..20e24dba0e 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp @@ -745,8 +745,7 @@ void RendererSceneRenderRD::_render_buffers_debug_draw(const RenderDataRD *p_ren } if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_MOTION_VECTORS && _render_buffers_get_velocity_texture(rb).is_valid()) { - Size2i rtsize = texture_storage->render_target_get_size(render_target); - copy_effects->copy_to_fb_rect(_render_buffers_get_velocity_texture(rb), texture_storage->render_target_get_rd_framebuffer(render_target), Rect2(Vector2(), rtsize), false, false); + debug_effects->draw_motion_vectors(_render_buffers_get_velocity_texture(rb), texture_storage->render_target_get_rd_framebuffer(render_target), rb->get_internal_size()); } } diff --git a/servers/rendering/renderer_rd/shaders/effects/motion_vectors.glsl b/servers/rendering/renderer_rd/shaders/effects/motion_vectors.glsl new file mode 100644 index 0000000000..80e4f51565 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/effects/motion_vectors.glsl @@ -0,0 +1,80 @@ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +layout(location = 0) out vec2 uv_interp; + +void main() { + vec2 base_arr[3] = vec2[](vec2(-1.0, -1.0), vec2(-1.0, 3.0), vec2(3.0, -1.0)); + gl_Position = vec4(base_arr[gl_VertexIndex], 0.0, 1.0); + uv_interp = clamp(gl_Position.xy, vec2(0.0, 0.0), vec2(1.0, 1.0)) * 2.0; // saturate(x) * 2.0 +} + +#[fragment] + +#version 450 + +#VERSION_DEFINES + +layout(location = 0) in vec2 uv_interp; + +layout(set = 0, binding = 0) uniform sampler2D source_velocity; + +layout(location = 0) out vec4 frag_color; + +layout(push_constant, std430) uniform Params { + vec2 resolution; +} +params; + +// Based on distance to line segment from https://www.shadertoy.com/view/3tdSDj + +float line_segment(in vec2 p, in vec2 a, in vec2 b) { + vec2 aspect = vec2(params.resolution.x / params.resolution.y, 1.0f); + vec2 ba = (b - a) * aspect; + vec2 pa = (p - a) * aspect; + float h = clamp(dot(pa, ba) / dot(ba, ba), 0.0f, 1.0f); + return length(pa - h * ba) * (params.resolution.y / 2.0f); +} + +void main() { + // Retrieve motion vector data. + float cell_size = 32.0f; + float circle_radius = 2.0f; + vec3 nan_color = vec3(1.0f, 0.0f, 0.0f); + vec3 active_color = vec3(1.0f, 0.8f, 0.1f); + vec3 inactive_color = vec3(0.5f, 0.5f, 0.5f); + vec2 pos_pixel = uv_interp * params.resolution; + vec2 cell_pos_pixel = floor(pos_pixel / cell_size) * cell_size + (cell_size * 0.5f); + vec2 cell_pos_uv = cell_pos_pixel / params.resolution; + vec2 cell_pos_previous_uv = cell_pos_uv + textureLod(source_velocity, cell_pos_uv, 0.0f).xy; + + // Draw the shapes. + float epsilon = 1e-6f; + vec2 cell_pos_delta_uv = cell_pos_uv - cell_pos_previous_uv; + bool motion_active = length(cell_pos_delta_uv) > epsilon; + vec3 color; + if (any(isnan(cell_pos_delta_uv))) { + color = nan_color; + } else if (motion_active) { + color = active_color; + } else { + color = inactive_color; + } + + float alpha; + if (length(cell_pos_pixel - pos_pixel) <= circle_radius) { + // Circle center. + alpha = 1.0f; + } else if (motion_active) { + // Motion vector line. + alpha = 1.0f - line_segment(uv_interp, cell_pos_uv, cell_pos_previous_uv); + } else { + // Ignore pixel. + alpha = 0.0f; + } + + frag_color = vec4(color, alpha); +} diff --git a/servers/rendering/renderer_rd/shaders/effects/taa_resolve.glsl b/servers/rendering/renderer_rd/shaders/effects/taa_resolve.glsl index 02566d8e35..e50397cc5d 100644 --- a/servers/rendering/renderer_rd/shaders/effects/taa_resolve.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/taa_resolve.glsl @@ -320,7 +320,7 @@ vec3 temporal_antialiasing(uvec2 pos_group_top_left, uvec2 pos_group, uvec2 pos_ vec2 velocity = imageLoad(velocity_buffer, ivec2(pos_screen)).xy; // Get reprojected uv - vec2 uv_reprojected = uv - velocity; + vec2 uv_reprojected = uv + velocity; // Get input color vec3 color_input = load_color(pos_group); 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 d1cfda515f..cfba408fe1 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 @@ -802,7 +802,9 @@ void fragment_shader(in SceneData scene_data) { float clearcoat_roughness = 0.0; float anisotropy = 0.0; vec2 anisotropy_flow = vec2(1.0, 0.0); +#ifndef FOG_DISABLED vec4 fog = vec4(0.0); +#endif // !FOG_DISABLED #if defined(CUSTOM_RADIANCE_USED) vec4 custom_radiance = vec4(0.0); #endif @@ -962,6 +964,7 @@ void fragment_shader(in SceneData scene_data) { /////////////////////// FOG ////////////////////// #ifndef MODE_RENDER_DEPTH +#ifndef FOG_DISABLED #ifndef CUSTOM_FOG_USED // fog must be processed as early as possible and then packed. // to maximize VGPR usage @@ -997,6 +1000,7 @@ void fragment_shader(in SceneData scene_data) { uint fog_rg = packHalf2x16(fog.rg); uint fog_ba = packHalf2x16(fog.ba); +#endif //!FOG_DISABLED #endif //!MODE_RENDER_DEPTH /////////////////////// DECALS //////////////////////////////// @@ -2250,8 +2254,10 @@ void fragment_shader(in SceneData scene_data) { diffuse_light *= 1.0 - metallic; ambient_light *= 1.0 - metallic; +#ifndef FOG_DISABLED //restore fog fog = vec4(unpackHalf2x16(fog_rg), unpackHalf2x16(fog_ba)); +#endif //!FOG_DISABLED #ifdef MODE_SEPARATE_SPECULAR @@ -2268,8 +2274,10 @@ void fragment_shader(in SceneData scene_data) { specular_buffer = vec4(specular_light, metallic); #endif +#ifndef FOG_DISABLED diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); +#endif //!FOG_DISABLED #else //MODE_SEPARATE_SPECULAR @@ -2280,8 +2288,10 @@ void fragment_shader(in SceneData scene_data) { //frag_color = vec4(1.0); #endif //USE_NO_SHADING +#ifndef FOG_DISABLED // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); +#endif //!FOG_DISABLED #endif //MODE_SEPARATE_SPECULAR @@ -2293,7 +2303,7 @@ void fragment_shader(in SceneData scene_data) { vec2 position_uv = position_clip * vec2(0.5, 0.5); vec2 prev_position_uv = prev_position_clip * vec2(0.5, 0.5); - motion_vector = position_uv - prev_position_uv; + motion_vector = prev_position_uv - position_uv; #endif } 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 b63eea1401..cdf81bb6ec 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 @@ -694,7 +694,9 @@ void main() { float clearcoat_roughness = 0.0; float anisotropy = 0.0; vec2 anisotropy_flow = vec2(1.0, 0.0); +#ifndef FOG_DISABLED vec4 fog = vec4(0.0); +#endif // !FOG_DISABLED #if defined(CUSTOM_RADIANCE_USED) vec4 custom_radiance = vec4(0.0); #endif @@ -860,6 +862,7 @@ void main() { /////////////////////// FOG ////////////////////// #ifndef MODE_RENDER_DEPTH +#ifndef FOG_DISABLED #ifndef CUSTOM_FOG_USED // fog must be processed as early as possible and then packed. // to maximize VGPR usage @@ -874,6 +877,7 @@ void main() { uint fog_rg = packHalf2x16(fog.rg); uint fog_ba = packHalf2x16(fog.ba); +#endif //!FOG_DISABLED #endif //!MODE_RENDER_DEPTH /////////////////////// DECALS //////////////////////////////// @@ -899,6 +903,10 @@ void main() { break; } + if (!bool(decals.data[decal_index].mask & draw_call.layer_mask)) { + continue; //not masked + } + vec3 uv_local = (decals.data[decal_index].xform * vec4(vertex, 1.0)).xyz; if (any(lessThan(uv_local, vec3(0.0, -1.0, 0.0))) || any(greaterThan(uv_local, vec3(1.0)))) { continue; //out of decal @@ -1740,8 +1748,10 @@ void main() { diffuse_light *= 1.0 - metallic; ambient_light *= 1.0 - metallic; +#ifndef FOG_DISABLED //restore fog fog = vec4(unpackHalf2x16(fog_rg), unpackHalf2x16(fog_ba)); +#endif // !FOG_DISABLED #ifdef MODE_MULTIPLE_RENDER_TARGETS @@ -1758,8 +1768,10 @@ void main() { specular_buffer = vec4(specular_light, metallic); #endif // MODE_UNSHADED +#ifndef FOG_DISABLED diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); +#endif // !FOG_DISABLED #else //MODE_MULTIPLE_RENDER_TARGETS @@ -1769,8 +1781,10 @@ void main() { frag_color = vec4(emission + ambient_light + diffuse_light + specular_light, alpha); #endif // MODE_UNSHADED +#ifndef FOG_DISABLED // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); +#endif // !FOG_DISABLED // On mobile we use a UNORM buffer with 10bpp which results in a range from 0.0 - 1.0 resulting in HDR breaking // We divide by sc_luminance_multiplier to support a range from 0.0 - 2.0 both increasing precision on bright and darker images diff --git a/servers/rendering/renderer_rd/shaders/particles_copy.glsl b/servers/rendering/renderer_rd/shaders/particles_copy.glsl index 67a894b86f..93c4d1009d 100644 --- a/servers/rendering/renderer_rd/shaders/particles_copy.glsl +++ b/servers/rendering/renderer_rd/shaders/particles_copy.glsl @@ -45,6 +45,9 @@ layout(set = 2, binding = 0, std430) restrict readonly buffer TrailBindPoses { } trail_bind_poses; +#define PARAMS_FLAG_ORDER_BY_LIFETIME 1 +#define PARAMS_FLAG_COPY_MODE_2D 2 + layout(push_constant, std430) uniform Params { vec3 sort_direction; uint total_particles; @@ -57,10 +60,10 @@ layout(push_constant, std430) uniform Params { vec3 align_up; uint align_mode; - bool order_by_lifetime; uint lifetime_split; bool lifetime_reverse; - bool copy_mode_2d; + uint motion_vectors_current_offset; + uint flags; mat4 inv_emission_transform; } @@ -103,7 +106,7 @@ void main() { particle = uint(sort_buffer.data[particle].y); //use index from sort buffer } #else - if (params.order_by_lifetime) { + if (bool(params.flags & PARAMS_FLAG_ORDER_BY_LIFETIME)) { if (params.trail_size > 1) { uint limit = (params.total_particles / params.trail_size) - params.lifetime_split; @@ -179,7 +182,6 @@ void main() { 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; @@ -187,9 +189,9 @@ void main() { 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; + txform[0].xyz = normalize(cross(sv, params.sort_direction)) * length(txform[0]); + txform[1].xyz = sv * length(txform[1]); + txform[2].xyz = params.sort_direction * length(txform[2]); } break; } @@ -201,7 +203,7 @@ void main() { txform = txform * trail_bind_poses.data[part_ofs]; } - if (params.copy_mode_2d) { + if (bool(params.flags & PARAMS_FLAG_COPY_MODE_2D)) { // In global mode, bring 2D particles to local coordinates // as they will be drawn with the node position as origin. txform = params.inv_emission_transform * txform; @@ -213,15 +215,16 @@ void main() { } txform = transpose(txform); - if (params.copy_mode_2d) { - uint write_offset = gl_GlobalInvocationID.x * (2 + 1 + 1); //xform + color + custom + uint instance_index = gl_GlobalInvocationID.x + params.motion_vectors_current_offset; + if (bool(params.flags & PARAMS_FLAG_COPY_MODE_2D)) { + uint write_offset = instance_index * (2 + 1 + 1); //xform + color + custom instances.data[write_offset + 0] = txform[0]; instances.data[write_offset + 1] = txform[1]; instances.data[write_offset + 2] = particles.data[particle].color; instances.data[write_offset + 3] = particles.data[particle].custom; } else { - uint write_offset = gl_GlobalInvocationID.x * (3 + 1 + 1); //xform + color + custom + uint write_offset = instance_index * (3 + 1 + 1); //xform + color + custom instances.data[write_offset + 0] = txform[0]; instances.data[write_offset + 1] = txform[1]; diff --git a/servers/rendering/renderer_rd/storage_rd/material_storage.cpp b/servers/rendering/renderer_rd/storage_rd/material_storage.cpp index fda341bbc9..5c4fa1a47c 100644 --- a/servers/rendering/renderer_rd/storage_rd/material_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/material_storage.cpp @@ -1191,18 +1191,18 @@ MaterialStorage::MaterialStorage() { // buffers { //create index array for copy shaders Vector<uint8_t> pv; - pv.resize(6 * 4); + pv.resize(6 * 2); { uint8_t *w = pv.ptrw(); - int *p32 = (int *)w; - p32[0] = 0; - p32[1] = 1; - p32[2] = 2; - p32[3] = 0; - p32[4] = 2; - p32[5] = 3; + uint16_t *p16 = (uint16_t *)w; + p16[0] = 0; + p16[1] = 1; + p16[2] = 2; + p16[3] = 0; + p16[4] = 2; + p16[5] = 3; } - quad_index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT32, pv); + quad_index_buffer = RD::get_singleton()->index_buffer_create(6, RenderingDevice::INDEX_BUFFER_FORMAT_UINT16, pv); quad_index_array = RD::get_singleton()->index_array_create(quad_index_buffer, 0, 6); } diff --git a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp index 0a91672544..3a415e97e0 100644 --- a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp @@ -666,6 +666,19 @@ RID ParticlesStorage::particles_get_draw_pass_mesh(RID p_particles, int p_pass) return particles->draw_passes[p_pass]; } +void ParticlesStorage::particles_update_dependency(RID p_particles, DependencyTracker *p_instance) { + Particles *particles = particles_owner.get_or_null(p_particles); + ERR_FAIL_COND(!particles); + p_instance->update_dependency(&particles->dependency); +} + +void ParticlesStorage::particles_get_instance_buffer_motion_vectors_offsets(RID p_particles, uint32_t &r_current_offset, uint32_t &r_prev_offset) { + Particles *particles = particles_owner.get_or_null(p_particles); + ERR_FAIL_COND(!particles); + r_current_offset = particles->instance_motion_vectors_current_offset; + r_prev_offset = particles->instance_motion_vectors_previous_offset; +} + void ParticlesStorage::particles_add_collision(RID p_particles, RID p_particles_collision_instance) { Particles *particles = particles_owner.get_or_null(p_particles); ERR_FAIL_COND(!particles); @@ -1185,6 +1198,7 @@ void ParticlesStorage::particles_set_view_axis(RID p_particles, const Vector3 &p copy_push_constant.order_by_lifetime = (particles->draw_order == RS::PARTICLES_DRAW_ORDER_LIFETIME || particles->draw_order == RS::PARTICLES_DRAW_ORDER_REVERSE_LIFETIME); copy_push_constant.lifetime_split = (MIN(int(particles->amount * particles->phase), particles->amount - 1) + 1) % particles->amount; copy_push_constant.lifetime_reverse = particles->draw_order == RS::PARTICLES_DRAW_ORDER_REVERSE_LIFETIME; + copy_push_constant.motion_vectors_current_offset = particles->instance_motion_vectors_current_offset; copy_push_constant.frame_remainder = particles->interpolate ? particles->frame_remainder : 0.0; copy_push_constant.total_particles = particles->amount; @@ -1252,28 +1266,50 @@ void ParticlesStorage::_particles_update_buffers(Particles *particles) { userdata_count = particle_shader_data->userdata_count; } + bool uses_motion_vectors = RSG::viewport->get_num_viewports_with_motion_vectors() > 0; + bool index_draw_order = particles->draw_order == RS::ParticlesDrawOrder::PARTICLES_DRAW_ORDER_INDEX; + bool enable_motion_vectors = uses_motion_vectors && index_draw_order && !particles->instance_motion_vectors_enabled; + bool only_instances_changed = false; + if (userdata_count != particles->userdata_count) { - // Mismatch userdata, re-create buffers. + // Mismatch userdata, re-create all buffers. _particles_free_data(particles); + } else if (enable_motion_vectors) { + // Only motion vectors are required, release the transforms buffer and uniform set. + if (particles->particle_instance_buffer.is_valid()) { + RD::get_singleton()->free(particles->particle_instance_buffer); + particles->particle_instance_buffer = RID(); + } + + particles->particles_transforms_buffer_uniform_set = RID(); + only_instances_changed = true; + } else if (!particles->particle_buffer.is_null()) { + // No operation is required because a buffer already exists, return early. + return; } - if (particles->amount > 0 && particles->particle_buffer.is_null()) { + if (particles->amount > 0) { int total_amount = particles->amount; if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { total_amount *= particles->trail_bind_poses.size(); } uint32_t xform_size = particles->mode == RS::PARTICLES_MODE_2D ? 2 : 3; - - particles->particle_buffer = RD::get_singleton()->storage_buffer_create((sizeof(ParticleData) + userdata_count * sizeof(float) * 4) * total_amount); - - particles->userdata_count = userdata_count; + if (particles->particle_buffer.is_null()) { + particles->particle_buffer = RD::get_singleton()->storage_buffer_create((sizeof(ParticleData) + userdata_count * sizeof(float) * 4) * total_amount); + particles->userdata_count = userdata_count; + } PackedByteArray data; - data.resize_zeroed(sizeof(float) * 4 * (xform_size + 1 + 1) * total_amount); + uint32_t particle_instance_buffer_size = total_amount * (xform_size + 1 + 1) * sizeof(float) * 4; + if (uses_motion_vectors) { + particle_instance_buffer_size *= 2; + particles->instance_motion_vectors_enabled = true; + } - particles->particle_instance_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * 4 * (xform_size + 1 + 1) * total_amount, data); - //needs to clear it + data.resize_zeroed(particle_instance_buffer_size); + + particles->particle_instance_buffer = RD::get_singleton()->storage_buffer_create(particle_instance_buffer_size, data); { Vector<RD::Uniform> uniforms; @@ -1295,9 +1331,20 @@ void ParticlesStorage::_particles_update_buffers(Particles *particles) { 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->instance_motion_vectors_current_offset = 0; + particles->instance_motion_vectors_previous_offset = 0; + particles->instance_motion_vectors_last_change = -1; + + if (only_instances_changed) { + // Notify the renderer the instances uniform must be retrieved again, as it's the only element that has been changed because motion vectors were enabled. + particles->dependency.changed_notify(Dependency::DEPENDENCY_CHANGED_PARTICLES_INSTANCES); + } } } void ParticlesStorage::update_particles() { + uint32_t frame = RSG::rasterizer->get_frame_number(); + bool uses_motion_vectors = RSG::viewport->get_num_viewports_with_motion_vectors() > 0; while (particle_update_list) { //use transform feedback to process particles @@ -1461,16 +1508,25 @@ void ParticlesStorage::update_particles() { // Ensure that memory is initialized (the code above should ensure that _particles_process is always called at least once upon clearing). DEV_ASSERT(!particles->clear); + int total_amount = particles->amount; + if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { + total_amount *= particles->trail_bind_poses.size(); + } + + // Swap offsets for motion vectors. Motion vectors can only be used when the draw order keeps the indices consistent across frames. + bool index_draw_order = particles->draw_order == RS::ParticlesDrawOrder::PARTICLES_DRAW_ORDER_INDEX; + particles->instance_motion_vectors_previous_offset = particles->instance_motion_vectors_current_offset; + if (uses_motion_vectors && index_draw_order && particles->instance_motion_vectors_enabled && (frame - particles->instance_motion_vectors_last_change) == 1) { + particles->instance_motion_vectors_current_offset = total_amount - particles->instance_motion_vectors_current_offset; + } + + particles->instance_motion_vectors_last_change = frame; + // Copy particles to instance buffer. 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; - int total_amount = particles->amount; - if (particles->trails_enabled && particles->trail_bind_poses.size() > 1) { - total_amount *= particles->trail_bind_poses.size(); - } - // Affect 2D only. if (particles->use_local_coords) { // In local mode, particle positions are calculated locally (relative to the node position) @@ -1506,6 +1562,7 @@ void ParticlesStorage::update_particles() { copy_push_constant.order_by_lifetime = (particles->draw_order == RS::PARTICLES_DRAW_ORDER_LIFETIME || particles->draw_order == RS::PARTICLES_DRAW_ORDER_REVERSE_LIFETIME); copy_push_constant.lifetime_split = (MIN(int(particles->amount * particles->phase), particles->amount - 1) + 1) % particles->amount; copy_push_constant.lifetime_reverse = particles->draw_order == RS::PARTICLES_DRAW_ORDER_REVERSE_LIFETIME; + copy_push_constant.motion_vectors_current_offset = particles->instance_motion_vectors_current_offset; RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); copy_push_constant.copy_mode_2d = particles->mode == RS::PARTICLES_MODE_2D ? 1 : 0; diff --git a/servers/rendering/renderer_rd/storage_rd/particles_storage.h b/servers/rendering/renderer_rd/storage_rd/particles_storage.h index 612420a36c..7cea706a8c 100644 --- a/servers/rendering/renderer_rd/storage_rd/particles_storage.h +++ b/servers/rendering/renderer_rd/storage_rd/particles_storage.h @@ -225,6 +225,11 @@ private: double frame_remainder = 0; real_t collision_base_size = 0.01; + uint32_t instance_motion_vectors_current_offset = 0; + uint32_t instance_motion_vectors_previous_offset = 0; + uint64_t instance_motion_vectors_last_change = -1; + bool instance_motion_vectors_enabled = false; + bool clear = true; bool force_sub_emit = false; @@ -288,10 +293,13 @@ private: float align_up[3]; uint32_t align_mode; - uint32_t order_by_lifetime; uint32_t lifetime_split; uint32_t lifetime_reverse; - uint32_t copy_mode_2d; + uint32_t motion_vectors_current_offset; + struct { + uint32_t order_by_lifetime : 1; + uint32_t copy_mode_2d : 1; + }; float inv_emission_transform[16]; }; @@ -521,12 +529,15 @@ public: return particles->particles_transforms_buffer_uniform_set; } + void particles_get_instance_buffer_motion_vectors_offsets(RID p_particles, uint32_t &r_current_offset, uint32_t &r_prev_offset); + virtual void particles_add_collision(RID p_particles, RID p_particles_collision_instance) override; virtual void particles_remove_collision(RID p_particles, RID p_particles_collision_instance) override; void particles_set_canvas_sdf_collision(RID p_particles, bool p_enable, const Transform2D &p_xform, const Rect2 &p_to_screen, RID p_texture); virtual void update_particles() override; + void particles_update_dependency(RID p_particles, DependencyTracker *p_instance); Dependency *particles_get_dependency(RID p_particles) const; /* Particles Collision */ |