bevy/pipelined/bevy_pbr2/src/render/mod.rs

851 lines
32 KiB
Rust
Raw Normal View History

2021-06-02 02:59:17 +00:00
mod light;
pub use light::*;
use crate::{NotShadowCaster, NotShadowReceiver, StandardMaterial, StandardMaterialUniformData};
2021-06-02 02:59:17 +00:00
use bevy_asset::{Assets, Handle};
use bevy_core_pipeline::Transparent3dPhase;
2021-06-02 02:59:17 +00:00
use bevy_ecs::{prelude::*, system::SystemState};
use bevy_math::Mat4;
use bevy_render2::{
mesh::Mesh,
render_asset::RenderAssets,
2021-06-02 02:59:17 +00:00
render_graph::{Node, NodeRunError, RenderGraphContext},
render_phase::{Draw, DrawFunctions, Drawable, RenderPhase, TrackedRenderPass},
2021-06-21 23:28:52 +00:00
render_resource::*,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
renderer::{RenderContext, RenderDevice, RenderQueue},
2021-06-21 23:28:52 +00:00
shader::Shader,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
texture::{BevyDefault, GpuImage, Image, TextureFormatPixelInfo},
view::{ExtractedView, ViewMeta, ViewUniformOffset},
2021-06-02 02:59:17 +00:00
};
use bevy_transform::components::GlobalTransform;
use bevy_utils::slab::{FrameSlabMap, FrameSlabMapKey};
use crevice::std140::AsStd140;
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
use wgpu::{
Extent3d, ImageCopyTexture, ImageDataLayout, Origin3d, TextureDimension, TextureFormat,
TextureViewDescriptor,
};
2021-06-21 23:28:52 +00:00
2021-06-02 02:59:17 +00:00
pub struct PbrShaders {
2021-06-21 23:28:52 +00:00
pipeline: RenderPipeline,
view_layout: BindGroupLayout,
material_layout: BindGroupLayout,
mesh_layout: BindGroupLayout,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
// This dummy white texture is to be used in place of optional StandardMaterial textures
dummy_white_gpu_image: GpuImage,
2021-06-02 02:59:17 +00:00
}
// TODO: this pattern for initializing the shaders / pipeline isn't ideal. this should be handled by the asset system
impl FromWorld for PbrShaders {
fn from_world(world: &mut World) -> Self {
2021-06-21 23:28:52 +00:00
let render_device = world.get_resource::<RenderDevice>().unwrap();
let shader = Shader::from_wgsl(include_str!("pbr.wgsl"));
let shader_module = render_device.create_shader_module(&shader);
2021-06-21 23:28:52 +00:00
// TODO: move this into ViewMeta?
let view_layout = render_device.create_bind_group_layout(&BindGroupLayoutDescriptor {
entries: &[
// View
BindGroupLayoutEntry {
binding: 0,
visibility: ShaderStage::VERTEX | ShaderStage::FRAGMENT,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: true,
// TODO: change this to ViewUniform::std140_size_static once crevice fixes this!
// Context: https://github.com/LPGhatguy/crevice/issues/29
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
min_binding_size: BufferSize::new(144),
2021-06-21 23:28:52 +00:00
},
count: None,
2021-06-02 02:59:17 +00:00
},
2021-06-21 23:28:52 +00:00
// Lights
BindGroupLayoutEntry {
binding: 1,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: true,
// TODO: change this to GpuLights::std140_size_static once crevice fixes this!
// Context: https://github.com/LPGhatguy/crevice/issues/29
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
min_binding_size: BufferSize::new(1424),
2021-06-21 23:28:52 +00:00
},
count: None,
2021-06-02 02:59:17 +00:00
},
// Point Shadow Texture Cube Array
2021-06-21 23:28:52 +00:00
BindGroupLayoutEntry {
binding: 2,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Depth,
2021-07-01 23:48:55 +00:00
view_dimension: TextureViewDimension::CubeArray,
2021-06-21 23:28:52 +00:00
},
count: None,
},
// Point Shadow Texture Array Sampler
2021-06-21 23:28:52 +00:00
BindGroupLayoutEntry {
binding: 3,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: true,
filtering: true,
},
count: None,
2021-06-02 02:59:17 +00:00
},
// Directional Shadow Texture Array
BindGroupLayoutEntry {
binding: 4,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Depth,
view_dimension: TextureViewDimension::D2Array,
},
count: None,
},
// Directional Shadow Texture Array Sampler
BindGroupLayoutEntry {
binding: 5,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: true,
filtering: true,
},
count: None,
},
2021-06-02 02:59:17 +00:00
],
2021-06-21 23:28:52 +00:00
label: None,
});
let material_layout = render_device.create_bind_group_layout(&BindGroupLayoutDescriptor {
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
entries: &[
BindGroupLayoutEntry {
binding: 0,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: BufferSize::new(
StandardMaterialUniformData::std140_size_static() as u64,
),
},
count: None,
2021-06-21 23:28:52 +00:00
},
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
// Base Color Texture
BindGroupLayoutEntry {
binding: 1,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Float { filterable: true },
view_dimension: TextureViewDimension::D2,
},
count: None,
},
// Base Color Texture Sampler
BindGroupLayoutEntry {
binding: 2,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: false,
filtering: true,
},
count: None,
},
// Emissive Texture
BindGroupLayoutEntry {
binding: 3,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Float { filterable: true },
view_dimension: TextureViewDimension::D2,
},
count: None,
},
// Emissive Texture Sampler
BindGroupLayoutEntry {
binding: 4,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: false,
filtering: true,
},
count: None,
},
// Metallic Roughness Texture
BindGroupLayoutEntry {
binding: 5,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Float { filterable: true },
view_dimension: TextureViewDimension::D2,
},
count: None,
},
// Metallic Roughness Texture Sampler
BindGroupLayoutEntry {
binding: 6,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: false,
filtering: true,
},
count: None,
},
// Occlusion Texture
BindGroupLayoutEntry {
binding: 7,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Float { filterable: true },
view_dimension: TextureViewDimension::D2,
},
count: None,
},
// Occlusion Texture Sampler
BindGroupLayoutEntry {
binding: 8,
visibility: ShaderStage::FRAGMENT,
ty: BindingType::Sampler {
comparison: false,
filtering: true,
},
count: None,
},
],
2021-06-21 23:28:52 +00:00
label: None,
});
2021-06-02 02:59:17 +00:00
let mesh_layout = render_device.create_bind_group_layout(&BindGroupLayoutDescriptor {
entries: &[BindGroupLayoutEntry {
binding: 0,
visibility: ShaderStage::VERTEX | ShaderStage::FRAGMENT,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: true,
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
// TODO: change this to MeshUniform::std140_size_static once crevice fixes this!
// Context: https://github.com/LPGhatguy/crevice/issues/29
min_binding_size: BufferSize::new(144),
},
count: None,
}],
label: None,
});
2021-06-21 23:28:52 +00:00
let pipeline_layout = render_device.create_pipeline_layout(&PipelineLayoutDescriptor {
label: None,
push_constant_ranges: &[],
bind_group_layouts: &[&view_layout, &material_layout, &mesh_layout],
2021-06-21 23:28:52 +00:00
});
let pipeline = render_device.create_render_pipeline(&RenderPipelineDescriptor {
label: None,
vertex: VertexState {
buffers: &[VertexBufferLayout {
array_stride: 32,
step_mode: InputStepMode::Vertex,
attributes: &[
// Position (GOTCHA! Vertex_Position isn't first in the buffer due to how Mesh sorts attributes (alphabetically))
VertexAttribute {
format: VertexFormat::Float32x3,
offset: 12,
shader_location: 0,
},
// Normal
VertexAttribute {
format: VertexFormat::Float32x3,
offset: 0,
shader_location: 1,
},
// Uv
VertexAttribute {
format: VertexFormat::Float32x2,
offset: 24,
shader_location: 2,
},
],
}],
module: &shader_module,
entry_point: "vertex",
2021-06-21 23:28:52 +00:00
},
fragment: Some(FragmentState {
module: &shader_module,
entry_point: "fragment",
2021-06-21 23:28:52 +00:00
targets: &[ColorTargetState {
format: TextureFormat::bevy_default(),
blend: Some(BlendState {
color: BlendComponent {
src_factor: BlendFactor::SrcAlpha,
dst_factor: BlendFactor::OneMinusSrcAlpha,
operation: BlendOperation::Add,
},
alpha: BlendComponent {
src_factor: BlendFactor::One,
dst_factor: BlendFactor::One,
operation: BlendOperation::Add,
},
}),
write_mask: ColorWrite::ALL,
}],
}),
2021-06-02 02:59:17 +00:00
depth_stencil: Some(DepthStencilState {
format: TextureFormat::Depth32Float,
depth_write_enabled: true,
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
depth_compare: CompareFunction::Greater,
2021-06-02 02:59:17 +00:00
stencil: StencilState {
front: StencilFaceState::IGNORE,
back: StencilFaceState::IGNORE,
read_mask: 0,
write_mask: 0,
},
bias: DepthBiasState {
constant: 0,
slope_scale: 0.0,
clamp: 0.0,
},
}),
2021-06-21 23:28:52 +00:00
layout: Some(&pipeline_layout),
multisample: MultisampleState::default(),
primitive: PrimitiveState {
topology: PrimitiveTopology::TriangleList,
strip_index_format: None,
front_face: FrontFace::Ccw,
cull_mode: Some(Face::Back),
polygon_mode: PolygonMode::Fill,
clamp_depth: false,
conservative: false,
},
});
2021-06-02 02:59:17 +00:00
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
// A 1x1x1 'all 1.0' texture to use as a dummy texture to use in place of optional StandardMaterial textures
let dummy_white_gpu_image = {
let image = Image::new_fill(
Extent3d::default(),
TextureDimension::D2,
&[255u8; 4],
TextureFormat::bevy_default(),
);
let texture = render_device.create_texture(&image.texture_descriptor);
let sampler = render_device.create_sampler(&image.sampler_descriptor);
let format_size = image.texture_descriptor.format.pixel_size();
let render_queue = world.get_resource_mut::<RenderQueue>().unwrap();
render_queue.write_texture(
ImageCopyTexture {
texture: &texture,
mip_level: 0,
origin: Origin3d::ZERO,
},
&image.data,
ImageDataLayout {
offset: 0,
bytes_per_row: Some(
std::num::NonZeroU32::new(
image.texture_descriptor.size.width * format_size as u32,
)
.unwrap(),
),
rows_per_image: None,
},
image.texture_descriptor.size,
);
let texture_view = texture.create_view(&TextureViewDescriptor::default());
GpuImage {
texture,
texture_view,
sampler,
}
};
2021-06-02 02:59:17 +00:00
PbrShaders {
pipeline,
2021-06-21 23:28:52 +00:00
view_layout,
material_layout,
mesh_layout,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
dummy_white_gpu_image,
2021-06-02 02:59:17 +00:00
}
}
}
struct ExtractedMesh {
transform: Mat4,
mesh: Handle<Mesh>,
transform_binding_offset: u32,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
material_handle: Handle<StandardMaterial>,
casts_shadows: bool,
receives_shadows: bool,
2021-06-02 02:59:17 +00:00
}
pub struct ExtractedMeshes {
meshes: Vec<ExtractedMesh>,
}
pub fn extract_meshes(
mut commands: Commands,
meshes: Res<Assets<Mesh>>,
materials: Res<Assets<StandardMaterial>>,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
images: Res<Assets<Image>>,
query: Query<(
&GlobalTransform,
&Handle<Mesh>,
&Handle<StandardMaterial>,
Option<&NotShadowCaster>,
Option<&NotShadowReceiver>,
)>,
2021-06-02 02:59:17 +00:00
) {
let mut extracted_meshes = Vec::new();
for (
transform,
mesh_handle,
material_handle,
maybe_not_shadow_caster,
maybe_not_shadow_receiver,
) in query.iter()
{
if !meshes.contains(mesh_handle) {
continue;
}
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
if let Some(material) = materials.get(material_handle) {
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
if let Some(ref image) = material.base_color_texture {
if !images.contains(image) {
continue;
}
}
if let Some(ref image) = material.emissive_texture {
if !images.contains(image) {
continue;
}
}
if let Some(ref image) = material.metallic_roughness_texture {
if !images.contains(image) {
continue;
}
}
if let Some(ref image) = material.occlusion_texture {
if !images.contains(image) {
continue;
}
2021-06-02 02:59:17 +00:00
}
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
extracted_meshes.push(ExtractedMesh {
transform: transform.compute_matrix(),
mesh: mesh_handle.clone_weak(),
transform_binding_offset: 0,
material_handle: material_handle.clone_weak(),
// NOTE: Double-negative is so that meshes cast and receive shadows by default
// Not not shadow caster means that this mesh is a shadow caster
casts_shadows: maybe_not_shadow_caster.is_none(),
receives_shadows: maybe_not_shadow_receiver.is_none(),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
});
} else {
continue;
2021-06-02 02:59:17 +00:00
}
}
commands.insert_resource(ExtractedMeshes {
meshes: extracted_meshes,
});
}
struct MeshDrawInfo {
// TODO: compare cost of doing this vs cloning the BindGroup?
material_bind_group_key: FrameSlabMapKey<BufferId, BindGroup>,
}
#[derive(Debug, AsStd140)]
pub struct MeshUniform {
model: Mat4,
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
inverse_transpose_model: Mat4,
flags: u32,
}
// NOTE: These must match the bit flags in bevy_pbr2/src/render/pbr.wgsl!
bitflags::bitflags! {
#[repr(transparent)]
struct MeshFlags: u32 {
const SHADOW_RECEIVER = (1 << 0);
const NONE = 0;
const UNINITIALIZED = 0xFFFF;
}
}
2021-06-02 02:59:17 +00:00
#[derive(Default)]
pub struct MeshMeta {
transform_uniforms: DynamicUniformVec<MeshUniform>,
material_bind_groups: FrameSlabMap<BufferId, BindGroup>,
mesh_transform_bind_group: FrameSlabMap<BufferId, BindGroup>,
mesh_transform_bind_group_key: Option<FrameSlabMapKey<BufferId, BindGroup>>,
mesh_draw_info: Vec<MeshDrawInfo>,
2021-06-02 02:59:17 +00:00
}
pub fn prepare_meshes(
2021-06-21 23:28:52 +00:00
render_device: Res<RenderDevice>,
2021-06-02 02:59:17 +00:00
mut mesh_meta: ResMut<MeshMeta>,
mut extracted_meshes: ResMut<ExtractedMeshes>,
) {
mesh_meta
.transform_uniforms
2021-06-21 23:28:52 +00:00
.reserve_and_clear(extracted_meshes.meshes.len(), &render_device);
2021-06-02 02:59:17 +00:00
for extracted_mesh in extracted_meshes.meshes.iter_mut() {
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
let model = extracted_mesh.transform;
let inverse_transpose_model = model.inverse().transpose();
let flags = if extracted_mesh.receives_shadows {
MeshFlags::SHADOW_RECEIVER
} else {
MeshFlags::NONE
};
extracted_mesh.transform_binding_offset = mesh_meta.transform_uniforms.push(MeshUniform {
Use the infinite reverse right-handed perspective projection (#2543) # Objective Forward perspective projections have poor floating point precision distribution over the depth range. Reverse projections fair much better, and instead of having to have a far plane, with the reverse projection, using an infinite far plane is not a problem. The infinite reverse perspective projection has become the industry standard. The renderer rework is a great time to migrate to it. ## Solution All perspective projections, including point lights, have been moved to using `glam::Mat4::perspective_infinite_reverse_rh()` and so have no far plane. As various depth textures are shared between orthographic and perspective projections, a quirk of this PR is that the near and far planes of the orthographic projection are swapped when the Mat4 is computed. This has no impact on 2D/3D orthographic projection usage, and provides consistency in shaders, texture clear values, etc. throughout the codebase. ## Known issues For some reason, when looking along -Z, all geometry is black. The camera can be translated up/down / strafed left/right and geometry will still be black. Moving forward/backward or rotating the camera away from looking exactly along -Z causes everything to work as expected. I have tried to debug this issue but both in macOS and Windows I get crashes when doing pixel debugging. If anyone could reproduce this and debug it I would be very grateful. Otherwise I will have to try to debug it further without pixel debugging, though the projections and such all looked fine to me.
2021-08-27 20:15:09 +00:00
model,
inverse_transpose_model,
flags: flags.bits,
});
2021-06-02 02:59:17 +00:00
}
mesh_meta
.transform_uniforms
2021-06-21 23:28:52 +00:00
.write_to_staging_buffer(&render_device);
2021-06-02 02:59:17 +00:00
}
2021-06-21 23:28:52 +00:00
pub struct MeshViewBindGroups {
view: BindGroup,
}
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
fn image_handle_to_view_sampler<'a>(
pbr_shaders: &'a PbrShaders,
gpu_images: &'a RenderAssets<Image>,
image_option: &Option<Handle<Image>>,
) -> (&'a TextureView, &'a Sampler) {
image_option.as_ref().map_or(
(
&pbr_shaders.dummy_white_gpu_image.texture_view,
&pbr_shaders.dummy_white_gpu_image.sampler,
),
|image_handle| {
let gpu_image = gpu_images
.get(image_handle)
.expect("only materials with valid textures should be drawn");
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
(&gpu_image.texture_view, &gpu_image.sampler)
},
)
}
2021-07-02 01:05:20 +00:00
#[allow(clippy::too_many_arguments)]
2021-06-02 02:59:17 +00:00
pub fn queue_meshes(
mut commands: Commands,
draw_functions: Res<DrawFunctions>,
2021-06-21 23:28:52 +00:00
render_device: Res<RenderDevice>,
2021-06-02 02:59:17 +00:00
pbr_shaders: Res<PbrShaders>,
shadow_shaders: Res<ShadowShaders>,
2021-06-21 23:28:52 +00:00
mesh_meta: ResMut<MeshMeta>,
mut light_meta: ResMut<LightMeta>,
2021-06-02 02:59:17 +00:00
view_meta: Res<ViewMeta>,
mut extracted_meshes: ResMut<ExtractedMeshes>,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
gpu_images: Res<RenderAssets<Image>>,
render_materials: Res<RenderAssets<StandardMaterial>>,
mut views: Query<(
Entity,
&ExtractedView,
&ViewLights,
&mut RenderPhase<Transparent3dPhase>,
)>,
2021-06-02 02:59:17 +00:00
mut view_light_shadow_phases: Query<&mut RenderPhase<ShadowPhase>>,
) {
2021-06-21 23:28:52 +00:00
let mesh_meta = mesh_meta.into_inner();
2021-07-02 01:05:20 +00:00
if view_meta.uniforms.is_empty() {
return;
}
2021-06-21 23:28:52 +00:00
light_meta.shadow_view_bind_group.get_or_insert_with(|| {
render_device.create_bind_group(&BindGroupDescriptor {
entries: &[BindGroupEntry {
binding: 0,
resource: view_meta.uniforms.binding(),
}],
label: None,
layout: &shadow_shaders.view_layout,
})
});
2021-06-02 02:59:17 +00:00
if extracted_meshes.meshes.is_empty() {
return;
}
2021-06-21 23:28:52 +00:00
let transform_uniforms = &mesh_meta.transform_uniforms;
mesh_meta.mesh_transform_bind_group.next_frame();
mesh_meta.mesh_transform_bind_group_key =
Some(mesh_meta.mesh_transform_bind_group.get_or_insert_with(
transform_uniforms.uniform_buffer().unwrap().id(),
|| {
render_device.create_bind_group(&BindGroupDescriptor {
entries: &[BindGroupEntry {
binding: 0,
resource: transform_uniforms.binding(),
}],
label: None,
layout: &pbr_shaders.mesh_layout,
})
},
));
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
for (entity, view, view_lights, mut transparent_phase) in views.iter_mut() {
2021-06-21 23:28:52 +00:00
// TODO: cache this?
let view_bind_group = render_device.create_bind_group(&BindGroupDescriptor {
entries: &[
BindGroupEntry {
binding: 0,
resource: view_meta.uniforms.binding(),
},
BindGroupEntry {
binding: 1,
resource: light_meta.view_gpu_lights.binding(),
},
BindGroupEntry {
binding: 2,
resource: BindingResource::TextureView(
&view_lights.point_light_depth_texture_view,
),
2021-06-21 23:28:52 +00:00
},
BindGroupEntry {
binding: 3,
resource: BindingResource::Sampler(&shadow_shaders.point_light_sampler),
},
BindGroupEntry {
binding: 4,
resource: BindingResource::TextureView(
&view_lights.directional_light_depth_texture_view,
),
},
BindGroupEntry {
binding: 5,
resource: BindingResource::Sampler(&shadow_shaders.directional_light_sampler),
2021-06-21 23:28:52 +00:00
},
],
label: None,
layout: &pbr_shaders.view_layout,
});
2021-06-02 02:59:17 +00:00
commands.entity(entity).insert(MeshViewBindGroups {
2021-06-21 23:28:52 +00:00
view: view_bind_group,
2021-06-02 02:59:17 +00:00
});
let draw_pbr = draw_functions.read().get_id::<DrawPbr>().unwrap();
mesh_meta.mesh_draw_info.clear();
mesh_meta.material_bind_groups.next_frame();
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
let view_matrix = view.transform.compute_matrix();
let view_row_2 = view_matrix.row(2);
for (i, mesh) in extracted_meshes.meshes.iter_mut().enumerate() {
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
let gpu_material = &render_materials
.get(&mesh.material_handle)
.expect("Failed to get StandardMaterial PreparedAsset");
let material_bind_group_key =
mesh_meta
.material_bind_groups
.get_or_insert_with(gpu_material.buffer.id(), || {
let (base_color_texture_view, base_color_sampler) =
image_handle_to_view_sampler(
&pbr_shaders,
&gpu_images,
&gpu_material.base_color_texture,
);
let (emissive_texture_view, emissive_sampler) =
image_handle_to_view_sampler(
&pbr_shaders,
&gpu_images,
&gpu_material.emissive_texture,
);
let (metallic_roughness_texture_view, metallic_roughness_sampler) =
image_handle_to_view_sampler(
&pbr_shaders,
&gpu_images,
&gpu_material.metallic_roughness_texture,
);
let (occlusion_texture_view, occlusion_sampler) =
image_handle_to_view_sampler(
&pbr_shaders,
&gpu_images,
&gpu_material.occlusion_texture,
);
render_device.create_bind_group(&BindGroupDescriptor {
entries: &[
BindGroupEntry {
binding: 0,
resource: gpu_material.buffer.as_entire_binding(),
},
BindGroupEntry {
binding: 1,
resource: BindingResource::TextureView(base_color_texture_view),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 2,
resource: BindingResource::Sampler(base_color_sampler),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 3,
resource: BindingResource::TextureView(emissive_texture_view),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 4,
resource: BindingResource::Sampler(emissive_sampler),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 5,
resource: BindingResource::TextureView(
metallic_roughness_texture_view,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
),
},
BindGroupEntry {
binding: 6,
resource: BindingResource::Sampler(metallic_roughness_sampler),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 7,
resource: BindingResource::TextureView(occlusion_texture_view),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
BindGroupEntry {
binding: 8,
resource: BindingResource::Sampler(occlusion_sampler),
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
},
],
label: None,
layout: &pbr_shaders.material_layout,
})
});
mesh_meta.mesh_draw_info.push(MeshDrawInfo {
material_bind_group_key,
});
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
// NOTE: row 2 of the view matrix dotted with column 3 of the model matrix
// gives the z component of translation of the mesh in view space
let mesh_z = view_row_2.dot(mesh.transform.col(3));
// FIXME: Switch from usize to u64 for portability and use sort key encoding
// similar to https://realtimecollisiondetection.net/blog/?p=86 as appropriate
// FIXME: What is the best way to map from view space z to a number of bits of unsigned integer?
let sort_key = (((mesh_z * 1000.0) as usize) << 10)
| (material_bind_group_key.index() & ((1 << 10) - 1));
2021-06-02 02:59:17 +00:00
// TODO: currently there is only "transparent phase". this should pick transparent vs opaque according to the mesh material
transparent_phase.add(Drawable {
draw_function: draw_pbr,
draw_key: i,
bevy_pbr2: Add support for most of the StandardMaterial textures (#4) * bevy_pbr2: Add support for most of the StandardMaterial textures Normal maps are not included here as they require tangents in a vertex attribute. * bevy_pbr2: Ensure RenderCommandQueue is ready for PbrShaders init * texture_pipelined: Add a light to the scene so we can see stuff * WIP bevy_pbr2: back to front sorting hack * bevy_pbr2: Uniform control flow for texture sampling in pbr.frag From 'fintelia' on the Bevy Render Rework Round 2 discussion: "My understanding is that GPUs these days never use the "execute both branches and select the result" strategy. Rather, what they do is evaluate the branch condition on all threads of a warp, and jump over it if all of them evaluate to false. If even a single thread needs to execute the if statement body, however, then the remaining threads are paused until that is completed." * bevy_pbr2: Simplify texture and sampler names The StandardMaterial_ prefix is no longer needed * bevy_pbr2: Match default 'AmbientColor' of current bevy_pbr for now * bevy_pbr2: Convert from non-linear to linear sRGB for the color uniform * bevy_pbr2: Add pbr_pipelined example * Fix view vector in pbr frag to work in ortho * bevy_pbr2: Use a 90 degree y fov and light range projection for lights * bevy_pbr2: Add AmbientLight resource * bevy_pbr2: Convert PointLight color to linear sRGB for use in fragment shader * bevy_pbr2: pbr.frag: Rename PointLight.projection to view_projection The uniform contains the view_projection matrix so this was incorrect. * bevy_pbr2: PointLight is an OmniLight as it has a radius * bevy_pbr2: Factoring out duplicated code * bevy_pbr2: Implement RenderAsset for StandardMaterial * Remove unnecessary texture and sampler clones * fix comment formatting * remove redundant Buffer:from * Don't extract meshes when their material textures aren't ready * make missing textures in the queue step an error Co-authored-by: Aevyrie <aevyrie@gmail.com> Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2021-06-27 23:10:23 +00:00
sort_key,
2021-06-02 02:59:17 +00:00
});
}
// ultimately lights should check meshes for relevancy (ex: light views can "see" different meshes than the main view can)
let draw_shadow_mesh = draw_functions.read().get_id::<DrawShadowMesh>().unwrap();
for view_light_entity in view_lights.lights.iter().copied() {
let mut shadow_phase = view_light_shadow_phases.get_mut(view_light_entity).unwrap();
// TODO: this should only queue up meshes that are actually visible by each "light view"
for (i, mesh) in extracted_meshes.meshes.iter().enumerate() {
if mesh.casts_shadows {
shadow_phase.add(Drawable {
draw_function: draw_shadow_mesh,
draw_key: i,
sort_key: 0, // TODO: sort back-to-front
});
}
2021-06-02 02:59:17 +00:00
}
}
}
}
// TODO: this logic can be moved to prepare_meshes once wgpu::Queue is exposed directly
pub struct PbrNode;
impl Node for PbrNode {
fn run(
&self,
_graph: &mut RenderGraphContext,
2021-06-21 23:28:52 +00:00
render_context: &mut RenderContext,
2021-06-02 02:59:17 +00:00
world: &World,
) -> Result<(), NodeRunError> {
let mesh_meta = world.get_resource::<MeshMeta>().unwrap();
let light_meta = world.get_resource::<LightMeta>().unwrap();
mesh_meta
.transform_uniforms
2021-06-21 23:28:52 +00:00
.write_to_uniform_buffer(&mut render_context.command_encoder);
2021-06-02 02:59:17 +00:00
light_meta
.view_gpu_lights
2021-06-21 23:28:52 +00:00
.write_to_uniform_buffer(&mut render_context.command_encoder);
2021-06-02 02:59:17 +00:00
Ok(())
}
}
2021-06-21 23:28:52 +00:00
type DrawPbrParams<'s, 'w> = (
Res<'w, PbrShaders>,
Res<'w, MeshMeta>,
Res<'w, ExtractedMeshes>,
Res<'w, RenderAssets<Mesh>>,
2021-06-21 23:28:52 +00:00
Query<
'w,
's,
(
&'w ViewUniformOffset,
&'w ViewLights,
&'w MeshViewBindGroups,
),
>,
2021-06-02 02:59:17 +00:00
);
2021-06-21 23:28:52 +00:00
2021-06-02 02:59:17 +00:00
pub struct DrawPbr {
2021-06-21 23:28:52 +00:00
params: SystemState<DrawPbrParams<'static, 'static>>,
2021-06-02 02:59:17 +00:00
}
impl DrawPbr {
pub fn new(world: &mut World) -> Self {
Self {
params: SystemState::new(world),
}
}
}
impl Draw for DrawPbr {
2021-06-21 23:28:52 +00:00
fn draw<'w, 's>(
&'s mut self,
world: &'w World,
pass: &mut TrackedRenderPass<'w>,
2021-06-02 02:59:17 +00:00
view: Entity,
draw_key: usize,
_sort_key: usize,
2021-06-02 02:59:17 +00:00
) {
let (pbr_shaders, mesh_meta, extracted_meshes, meshes, views) = self.params.get(world);
2021-06-21 23:28:52 +00:00
let (view_uniforms, view_lights, mesh_view_bind_groups) = views.get(view).unwrap();
let extracted_mesh = &extracted_meshes.into_inner().meshes[draw_key];
let mesh_meta = mesh_meta.into_inner();
2021-06-21 23:28:52 +00:00
pass.set_render_pipeline(&pbr_shaders.into_inner().pipeline);
2021-06-02 02:59:17 +00:00
pass.set_bind_group(
0,
2021-06-21 23:28:52 +00:00
&mesh_view_bind_groups.view,
&[view_uniforms.offset, view_lights.gpu_light_binding_index],
2021-06-02 02:59:17 +00:00
);
let mesh_draw_info = &mesh_meta.mesh_draw_info[draw_key];
2021-06-02 02:59:17 +00:00
pass.set_bind_group(
1,
// &mesh_meta.material_bind_groups[sort_key & ((1 << 10) - 1)],
&mesh_meta.material_bind_groups[mesh_draw_info.material_bind_group_key],
&[],
);
pass.set_bind_group(
2,
mesh_meta
.mesh_transform_bind_group
.get_value(mesh_meta.mesh_transform_bind_group_key.unwrap())
.unwrap(),
2021-06-21 23:28:52 +00:00
&[extracted_mesh.transform_binding_offset],
2021-06-02 02:59:17 +00:00
);
let gpu_mesh = meshes.into_inner().get(&extracted_mesh.mesh).unwrap();
pass.set_vertex_buffer(0, gpu_mesh.vertex_buffer.slice(..));
if let Some(index_info) = &gpu_mesh.index_info {
2021-06-21 23:28:52 +00:00
pass.set_index_buffer(index_info.buffer.slice(..), 0, IndexFormat::Uint32);
2021-06-02 02:59:17 +00:00
pass.draw_indexed(0..index_info.count, 0, 0..1);
} else {
panic!("non-indexed drawing not supported yet")
}
}
}