bevy/crates/bevy_pbr/src/pbr_material.rs

532 lines
20 KiB
Rust
Raw Normal View History

use crate::{AlphaMode, MaterialPipeline, SpecializedMaterial, PBR_SHADER_HANDLE};
use bevy_asset::{AssetServer, Handle};
use bevy_ecs::system::{lifetimeless::SRes, SystemParamItem};
use bevy_math::Vec4;
use bevy_reflect::TypeUuid;
use bevy_render::{
color::Color,
Mesh vertex buffer layouts (#3959) This PR makes a number of changes to how meshes and vertex attributes are handled, which the goal of enabling easy and flexible custom vertex attributes: * Reworks the `Mesh` type to use the newly added `VertexAttribute` internally * `VertexAttribute` defines the name, a unique `VertexAttributeId`, and a `VertexFormat` * `VertexAttributeId` is used to produce consistent sort orders for vertex buffer generation, replacing the more expensive and often surprising "name based sorting" * Meshes can be used to generate a `MeshVertexBufferLayout`, which defines the layout of the gpu buffer produced by the mesh. `MeshVertexBufferLayouts` can then be used to generate actual `VertexBufferLayouts` according to the requirements of a specific pipeline. This decoupling of "mesh layout" vs "pipeline vertex buffer layout" is what enables custom attributes. We don't need to standardize _mesh layouts_ or contort meshes to meet the needs of a specific pipeline. As long as the mesh has what the pipeline needs, it will work transparently. * Mesh-based pipelines now specialize on `&MeshVertexBufferLayout` via the new `SpecializedMeshPipeline` trait (which behaves like `SpecializedPipeline`, but adds `&MeshVertexBufferLayout`). The integrity of the pipeline cache is maintained because the `MeshVertexBufferLayout` is treated as part of the key (which is fully abstracted from implementers of the trait ... no need to add any additional info to the specialization key). * Hashing `MeshVertexBufferLayout` is too expensive to do for every entity, every frame. To make this scalable, I added a generalized "pre-hashing" solution to `bevy_utils`: `Hashed<T>` keys and `PreHashMap<K, V>` (which uses `Hashed<T>` internally) . Why didn't I just do the quick and dirty in-place "pre-compute hash and use that u64 as a key in a hashmap" that we've done in the past? Because its wrong! Hashes by themselves aren't enough because two different values can produce the same hash. Re-hashing a hash is even worse! I decided to build a generalized solution because this pattern has come up in the past and we've chosen to do the wrong thing. Now we can do the right thing! This did unfortunately require pulling in `hashbrown` and using that in `bevy_utils`, because avoiding re-hashes requires the `raw_entry_mut` api, which isn't stabilized yet (and may never be ... `entry_ref` has favor now, but also isn't available yet). If std's HashMap ever provides the tools we need, we can move back to that. Note that adding `hashbrown` doesn't increase our dependency count because it was already in our tree. I will probably break these changes out into their own PR. * Specializing on `MeshVertexBufferLayout` has one non-obvious behavior: it can produce identical pipelines for two different MeshVertexBufferLayouts. To optimize the number of active pipelines / reduce re-binds while drawing, I de-duplicate pipelines post-specialization using the final `VertexBufferLayout` as the key. For example, consider a pipeline that needs the layout `(position, normal)` and is specialized using two meshes: `(position, normal, uv)` and `(position, normal, other_vec2)`. If both of these meshes result in `(position, normal)` specializations, we can use the same pipeline! Now we do. Cool! To briefly illustrate, this is what the relevant section of `MeshPipeline`'s specialization code looks like now: ```rust impl SpecializedMeshPipeline for MeshPipeline { type Key = MeshPipelineKey; fn specialize( &self, key: Self::Key, layout: &MeshVertexBufferLayout, ) -> RenderPipelineDescriptor { let mut vertex_attributes = vec![ Mesh::ATTRIBUTE_POSITION.at_shader_location(0), Mesh::ATTRIBUTE_NORMAL.at_shader_location(1), Mesh::ATTRIBUTE_UV_0.at_shader_location(2), ]; let mut shader_defs = Vec::new(); if layout.contains(Mesh::ATTRIBUTE_TANGENT) { shader_defs.push(String::from("VERTEX_TANGENTS")); vertex_attributes.push(Mesh::ATTRIBUTE_TANGENT.at_shader_location(3)); } let vertex_buffer_layout = layout .get_layout(&vertex_attributes) .expect("Mesh is missing a vertex attribute"); ``` Notice that this is _much_ simpler than it was before. And now any mesh with any layout can be used with this pipeline, provided it has vertex postions, normals, and uvs. We even got to remove `HAS_TANGENTS` from MeshPipelineKey and `has_tangents` from `GpuMesh`, because that information is redundant with `MeshVertexBufferLayout`. This is still a draft because I still need to: * Add more docs * Experiment with adding error handling to mesh pipeline specialization (which would print errors at runtime when a mesh is missing a vertex attribute required by a pipeline). If it doesn't tank perf, we'll keep it. * Consider breaking out the PreHash / hashbrown changes into a separate PR. * Add an example illustrating this change * Verify that the "mesh-specialized pipeline de-duplication code" works properly Please dont yell at me for not doing these things yet :) Just trying to get this in peoples' hands asap. Alternative to #3120 Fixes #3030 Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-02-23 23:21:13 +00:00
mesh::MeshVertexBufferLayout,
prelude::Shader,
render_asset::{PrepareAssetError, RenderAsset, RenderAssets},
render_resource::{
std140::{AsStd140, Std140},
*,
},
renderer::RenderDevice,
texture::Image,
};
/// A material with "standard" properties used in PBR lighting
/// Standard property values with pictures here
/// <https://google.github.io/filament/Material%20Properties.pdf>.
///
/// May be created directly from a [`Color`] or an [`Image`].
#[derive(Debug, Clone, TypeUuid)]
#[uuid = "7494888b-c082-457b-aacf-517228cc0c22"]
pub struct StandardMaterial {
/// Doubles as diffuse albedo for non-metallic, specular for metallic and a mix for everything
/// in between. If used together with a base_color_texture, this is factored into the final
/// base color as `base_color * base_color_texture_value`
pub base_color: Color,
pub base_color_texture: Option<Handle<Image>>,
// Use a color for user friendliness even though we technically don't use the alpha channel
// Might be used in the future for exposure correction in HDR
pub emissive: Color,
pub emissive_texture: Option<Handle<Image>>,
/// Linear perceptual roughness, clamped to [0.089, 1.0] in the shader
/// Defaults to minimum of 0.089
/// If used together with a roughness/metallic texture, this is factored into the final base
/// color as `roughness * roughness_texture_value`
pub perceptual_roughness: f32,
/// From [0.0, 1.0], dielectric to pure metallic
/// If used together with a roughness/metallic texture, this is factored into the final base
/// color as `metallic * metallic_texture_value`
pub metallic: f32,
pub metallic_roughness_texture: Option<Handle<Image>>,
/// Specular intensity for non-metals on a linear scale of [0.0, 1.0]
/// defaults to 0.5 which is mapped to 4% reflectance in the shader
pub reflectance: f32,
pub normal_map_texture: Option<Handle<Image>>,
/// Normal map textures authored for DirectX have their y-component flipped. Set this to flip
/// it to right-handed conventions.
pub flip_normal_map_y: bool,
pub occlusion_texture: Option<Handle<Image>>,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
/// Support two-sided lighting by automatically flipping the normals for "back" faces
/// within the PBR lighting shader.
/// Defaults to false.
/// This does not automatically configure backface culling, which can be done via
/// `cull_mode`.
pub double_sided: bool,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
/// Whether to cull the "front", "back" or neither side of a mesh
/// defaults to `Face::Back`
pub cull_mode: Option<Face>,
pub unlit: bool,
pub alpha_mode: AlphaMode,
}
impl Default for StandardMaterial {
fn default() -> Self {
StandardMaterial {
base_color: Color::rgb(1.0, 1.0, 1.0),
base_color_texture: None,
emissive: Color::BLACK,
emissive_texture: None,
// This is the minimum the roughness is clamped to in shader code
// See <https://google.github.io/filament/Filament.html#materialsystem/parameterization/>
// It's the minimum floating point value that won't be rounded down to 0 in the
// calculations used. Although technically for 32-bit floats, 0.045 could be
// used.
perceptual_roughness: 0.089,
// Few materials are purely dielectric or metallic
// This is just a default for mostly-dielectric
metallic: 0.01,
metallic_roughness_texture: None,
// Minimum real-world reflectance is 2%, most materials between 2-5%
// Expressed in a linear scale and equivalent to 4% reflectance see
// <https://google.github.io/filament/Material%20Properties.pdf>
reflectance: 0.5,
occlusion_texture: None,
normal_map_texture: None,
flip_normal_map_y: false,
double_sided: false,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
cull_mode: Some(Face::Back),
unlit: false,
alpha_mode: AlphaMode::Opaque,
}
}
}
impl From<Color> for StandardMaterial {
fn from(color: Color) -> Self {
StandardMaterial {
base_color: color,
alpha_mode: if color.a() < 1.0 {
AlphaMode::Blend
} else {
AlphaMode::Opaque
},
..Default::default()
}
}
}
impl From<Handle<Image>> for StandardMaterial {
fn from(texture: Handle<Image>) -> Self {
StandardMaterial {
base_color_texture: Some(texture),
..Default::default()
}
}
}
// NOTE: These must match the bit flags in bevy_pbr/src/render/pbr.wgsl!
bitflags::bitflags! {
#[repr(transparent)]
pub struct StandardMaterialFlags: u32 {
const BASE_COLOR_TEXTURE = (1 << 0);
const EMISSIVE_TEXTURE = (1 << 1);
const METALLIC_ROUGHNESS_TEXTURE = (1 << 2);
const OCCLUSION_TEXTURE = (1 << 3);
const DOUBLE_SIDED = (1 << 4);
const UNLIT = (1 << 5);
const ALPHA_MODE_OPAQUE = (1 << 6);
const ALPHA_MODE_MASK = (1 << 7);
const ALPHA_MODE_BLEND = (1 << 8);
const TWO_COMPONENT_NORMAL_MAP = (1 << 9);
const FLIP_NORMAL_MAP_Y = (1 << 10);
const NONE = 0;
const UNINITIALIZED = 0xFFFF;
}
}
/// The GPU representation of the uniform data of a [`StandardMaterial`].
#[derive(Clone, Default, AsStd140)]
pub struct StandardMaterialUniformData {
/// Doubles as diffuse albedo for non-metallic, specular for metallic and a mix for everything
/// in between.
pub base_color: Vec4,
// Use a color for user friendliness even though we technically don't use the alpha channel
// Might be used in the future for exposure correction in HDR
pub emissive: Vec4,
/// Linear perceptual roughness, clamped to [0.089, 1.0] in the shader
/// Defaults to minimum of 0.089
pub roughness: f32,
/// From [0.0, 1.0], dielectric to pure metallic
pub metallic: f32,
/// Specular intensity for non-metals on a linear scale of [0.0, 1.0]
/// defaults to 0.5 which is mapped to 4% reflectance in the shader
pub reflectance: f32,
pub flags: u32,
/// When the alpha mode mask flag is set, any base color alpha above this cutoff means fully opaque,
/// and any below means fully transparent.
pub alpha_cutoff: f32,
}
/// The GPU representation of a [`StandardMaterial`].
#[derive(Debug, Clone)]
pub struct GpuStandardMaterial {
/// A buffer containing the [`StandardMaterialUniformData`] of the material.
pub buffer: Buffer,
/// The bind group specifying how the [`StandardMaterialUniformData`] and
/// all the textures of the material are bound.
pub bind_group: BindGroup,
pub has_normal_map: bool,
pub flags: StandardMaterialFlags,
pub base_color_texture: Option<Handle<Image>>,
pub alpha_mode: AlphaMode,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
pub cull_mode: Option<Face>,
}
impl RenderAsset for StandardMaterial {
type ExtractedAsset = StandardMaterial;
type PreparedAsset = GpuStandardMaterial;
type Param = (
SRes<RenderDevice>,
SRes<MaterialPipeline<StandardMaterial>>,
SRes<RenderAssets<Image>>,
);
fn extract_asset(&self) -> Self::ExtractedAsset {
self.clone()
}
fn prepare_asset(
material: Self::ExtractedAsset,
(render_device, pbr_pipeline, gpu_images): &mut SystemParamItem<Self::Param>,
) -> Result<Self::PreparedAsset, PrepareAssetError<Self::ExtractedAsset>> {
let (base_color_texture_view, base_color_sampler) = if let Some(result) = pbr_pipeline
.mesh_pipeline
.get_image_texture(gpu_images, &material.base_color_texture)
{
result
} else {
return Err(PrepareAssetError::RetryNextUpdate(material));
};
let (emissive_texture_view, emissive_sampler) = if let Some(result) = pbr_pipeline
.mesh_pipeline
.get_image_texture(gpu_images, &material.emissive_texture)
{
result
} else {
return Err(PrepareAssetError::RetryNextUpdate(material));
};
let (metallic_roughness_texture_view, metallic_roughness_sampler) = if let Some(result) =
pbr_pipeline
.mesh_pipeline
.get_image_texture(gpu_images, &material.metallic_roughness_texture)
{
result
} else {
return Err(PrepareAssetError::RetryNextUpdate(material));
};
let (normal_map_texture_view, normal_map_sampler) = if let Some(result) = pbr_pipeline
.mesh_pipeline
.get_image_texture(gpu_images, &material.normal_map_texture)
{
result
} else {
return Err(PrepareAssetError::RetryNextUpdate(material));
};
let (occlusion_texture_view, occlusion_sampler) = if let Some(result) = pbr_pipeline
.mesh_pipeline
.get_image_texture(gpu_images, &material.occlusion_texture)
{
result
} else {
return Err(PrepareAssetError::RetryNextUpdate(material));
};
let mut flags = StandardMaterialFlags::NONE;
if material.base_color_texture.is_some() {
flags |= StandardMaterialFlags::BASE_COLOR_TEXTURE;
}
if material.emissive_texture.is_some() {
flags |= StandardMaterialFlags::EMISSIVE_TEXTURE;
}
if material.metallic_roughness_texture.is_some() {
flags |= StandardMaterialFlags::METALLIC_ROUGHNESS_TEXTURE;
}
if material.occlusion_texture.is_some() {
flags |= StandardMaterialFlags::OCCLUSION_TEXTURE;
}
if material.double_sided {
flags |= StandardMaterialFlags::DOUBLE_SIDED;
}
if material.unlit {
flags |= StandardMaterialFlags::UNLIT;
}
let has_normal_map = material.normal_map_texture.is_some();
if has_normal_map {
match gpu_images
.get(material.normal_map_texture.as_ref().unwrap())
.unwrap()
.texture_format
{
// All 2-component unorm formats
TextureFormat::Rg8Unorm
| TextureFormat::Rg16Unorm
| TextureFormat::Bc5RgUnorm
| TextureFormat::EacRg11Unorm => {
flags |= StandardMaterialFlags::TWO_COMPONENT_NORMAL_MAP
}
_ => {}
}
if material.flip_normal_map_y {
flags |= StandardMaterialFlags::FLIP_NORMAL_MAP_Y;
}
}
// NOTE: 0.5 is from the glTF default - do we want this?
let mut alpha_cutoff = 0.5;
match material.alpha_mode {
AlphaMode::Opaque => flags |= StandardMaterialFlags::ALPHA_MODE_OPAQUE,
AlphaMode::Mask(c) => {
alpha_cutoff = c;
flags |= StandardMaterialFlags::ALPHA_MODE_MASK;
}
AlphaMode::Blend => flags |= StandardMaterialFlags::ALPHA_MODE_BLEND,
};
let value = StandardMaterialUniformData {
base_color: material.base_color.as_linear_rgba_f32().into(),
emissive: material.emissive.into(),
roughness: material.perceptual_roughness,
metallic: material.metallic,
reflectance: material.reflectance,
flags: flags.bits(),
alpha_cutoff,
};
let value_std140 = value.as_std140();
let buffer = render_device.create_buffer_with_data(&BufferInitDescriptor {
label: Some("pbr_standard_material_uniform_buffer"),
usage: BufferUsages::UNIFORM | BufferUsages::COPY_DST,
contents: value_std140.as_bytes(),
});
let bind_group = render_device.create_bind_group(&BindGroupDescriptor {
entries: &[
BindGroupEntry {
binding: 0,
resource: buffer.as_entire_binding(),
},
BindGroupEntry {
binding: 1,
resource: BindingResource::TextureView(base_color_texture_view),
},
BindGroupEntry {
binding: 2,
resource: BindingResource::Sampler(base_color_sampler),
},
BindGroupEntry {
binding: 3,
resource: BindingResource::TextureView(emissive_texture_view),
},
BindGroupEntry {
binding: 4,
resource: BindingResource::Sampler(emissive_sampler),
},
BindGroupEntry {
binding: 5,
resource: BindingResource::TextureView(metallic_roughness_texture_view),
},
BindGroupEntry {
binding: 6,
resource: BindingResource::Sampler(metallic_roughness_sampler),
},
BindGroupEntry {
binding: 7,
resource: BindingResource::TextureView(occlusion_texture_view),
},
BindGroupEntry {
binding: 8,
resource: BindingResource::Sampler(occlusion_sampler),
},
BindGroupEntry {
binding: 9,
resource: BindingResource::TextureView(normal_map_texture_view),
},
BindGroupEntry {
binding: 10,
resource: BindingResource::Sampler(normal_map_sampler),
},
],
label: Some("pbr_standard_material_bind_group"),
layout: &pbr_pipeline.material_layout,
});
Ok(GpuStandardMaterial {
buffer,
bind_group,
flags,
has_normal_map,
base_color_texture: material.base_color_texture,
alpha_mode: material.alpha_mode,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
cull_mode: material.cull_mode,
})
}
}
#[derive(Clone, PartialEq, Eq, Hash)]
pub struct StandardMaterialKey {
normal_map: bool,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
cull_mode: Option<Face>,
}
impl SpecializedMaterial for StandardMaterial {
type Key = StandardMaterialKey;
fn key(render_asset: &<Self as RenderAsset>::PreparedAsset) -> Self::Key {
StandardMaterialKey {
normal_map: render_asset.has_normal_map,
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
cull_mode: render_asset.cull_mode,
}
}
Mesh vertex buffer layouts (#3959) This PR makes a number of changes to how meshes and vertex attributes are handled, which the goal of enabling easy and flexible custom vertex attributes: * Reworks the `Mesh` type to use the newly added `VertexAttribute` internally * `VertexAttribute` defines the name, a unique `VertexAttributeId`, and a `VertexFormat` * `VertexAttributeId` is used to produce consistent sort orders for vertex buffer generation, replacing the more expensive and often surprising "name based sorting" * Meshes can be used to generate a `MeshVertexBufferLayout`, which defines the layout of the gpu buffer produced by the mesh. `MeshVertexBufferLayouts` can then be used to generate actual `VertexBufferLayouts` according to the requirements of a specific pipeline. This decoupling of "mesh layout" vs "pipeline vertex buffer layout" is what enables custom attributes. We don't need to standardize _mesh layouts_ or contort meshes to meet the needs of a specific pipeline. As long as the mesh has what the pipeline needs, it will work transparently. * Mesh-based pipelines now specialize on `&MeshVertexBufferLayout` via the new `SpecializedMeshPipeline` trait (which behaves like `SpecializedPipeline`, but adds `&MeshVertexBufferLayout`). The integrity of the pipeline cache is maintained because the `MeshVertexBufferLayout` is treated as part of the key (which is fully abstracted from implementers of the trait ... no need to add any additional info to the specialization key). * Hashing `MeshVertexBufferLayout` is too expensive to do for every entity, every frame. To make this scalable, I added a generalized "pre-hashing" solution to `bevy_utils`: `Hashed<T>` keys and `PreHashMap<K, V>` (which uses `Hashed<T>` internally) . Why didn't I just do the quick and dirty in-place "pre-compute hash and use that u64 as a key in a hashmap" that we've done in the past? Because its wrong! Hashes by themselves aren't enough because two different values can produce the same hash. Re-hashing a hash is even worse! I decided to build a generalized solution because this pattern has come up in the past and we've chosen to do the wrong thing. Now we can do the right thing! This did unfortunately require pulling in `hashbrown` and using that in `bevy_utils`, because avoiding re-hashes requires the `raw_entry_mut` api, which isn't stabilized yet (and may never be ... `entry_ref` has favor now, but also isn't available yet). If std's HashMap ever provides the tools we need, we can move back to that. Note that adding `hashbrown` doesn't increase our dependency count because it was already in our tree. I will probably break these changes out into their own PR. * Specializing on `MeshVertexBufferLayout` has one non-obvious behavior: it can produce identical pipelines for two different MeshVertexBufferLayouts. To optimize the number of active pipelines / reduce re-binds while drawing, I de-duplicate pipelines post-specialization using the final `VertexBufferLayout` as the key. For example, consider a pipeline that needs the layout `(position, normal)` and is specialized using two meshes: `(position, normal, uv)` and `(position, normal, other_vec2)`. If both of these meshes result in `(position, normal)` specializations, we can use the same pipeline! Now we do. Cool! To briefly illustrate, this is what the relevant section of `MeshPipeline`'s specialization code looks like now: ```rust impl SpecializedMeshPipeline for MeshPipeline { type Key = MeshPipelineKey; fn specialize( &self, key: Self::Key, layout: &MeshVertexBufferLayout, ) -> RenderPipelineDescriptor { let mut vertex_attributes = vec![ Mesh::ATTRIBUTE_POSITION.at_shader_location(0), Mesh::ATTRIBUTE_NORMAL.at_shader_location(1), Mesh::ATTRIBUTE_UV_0.at_shader_location(2), ]; let mut shader_defs = Vec::new(); if layout.contains(Mesh::ATTRIBUTE_TANGENT) { shader_defs.push(String::from("VERTEX_TANGENTS")); vertex_attributes.push(Mesh::ATTRIBUTE_TANGENT.at_shader_location(3)); } let vertex_buffer_layout = layout .get_layout(&vertex_attributes) .expect("Mesh is missing a vertex attribute"); ``` Notice that this is _much_ simpler than it was before. And now any mesh with any layout can be used with this pipeline, provided it has vertex postions, normals, and uvs. We even got to remove `HAS_TANGENTS` from MeshPipelineKey and `has_tangents` from `GpuMesh`, because that information is redundant with `MeshVertexBufferLayout`. This is still a draft because I still need to: * Add more docs * Experiment with adding error handling to mesh pipeline specialization (which would print errors at runtime when a mesh is missing a vertex attribute required by a pipeline). If it doesn't tank perf, we'll keep it. * Consider breaking out the PreHash / hashbrown changes into a separate PR. * Add an example illustrating this change * Verify that the "mesh-specialized pipeline de-duplication code" works properly Please dont yell at me for not doing these things yet :) Just trying to get this in peoples' hands asap. Alternative to #3120 Fixes #3030 Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-02-23 23:21:13 +00:00
fn specialize(
Use storage buffers for clustered forward point lights (#3989) # Objective - Make use of storage buffers, where they are available, for clustered forward bindings to support far more point lights in a scene - Fixes #3605 - Based on top of #4079 This branch on an M1 Max can keep 60fps with about 2150 point lights of radius 1m in the Sponza scene where I've been testing. The bottleneck is mostly assigning lights to clusters which grows faster than linearly (I think 1000 lights was about 1.5ms and 5000 was 7.5ms). I have seen papers and presentations leveraging compute shaders that can get this up to over 1 million. That said, I think any further optimisations should probably be done in a separate PR. ## Solution - Add `RenderDevice` to the `Material` and `SpecializedMaterial` trait `::key()` functions to allow setting flags on the keys depending on feature/limit availability - Make `GpuPointLights` and `ViewClusterBuffers` into enums containing `UniformVec` and `StorageBuffer` variants. Implement the necessary API on them to make usage the same for both cases, and the only difference is at initialisation time. - Appropriate shader defs in the shader code to handle the two cases ## Context on some decisions / open questions - I'm using `max_storage_buffers_per_shader_stage >= 3` as a check to see if storage buffers are supported. I was thinking about diving into 'binding resource management' but it feels like we don't have enough use cases to understand the problem yet, and it is mostly a separate concern to this PR, so I think it should be handled separately. - Should `ViewClusterBuffers` and `ViewClusterBindings` be merged, duplicating the count variables into the enum variants? Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-07 16:16:35 +00:00
_pipeline: &MaterialPipeline<Self>,
Mesh vertex buffer layouts (#3959) This PR makes a number of changes to how meshes and vertex attributes are handled, which the goal of enabling easy and flexible custom vertex attributes: * Reworks the `Mesh` type to use the newly added `VertexAttribute` internally * `VertexAttribute` defines the name, a unique `VertexAttributeId`, and a `VertexFormat` * `VertexAttributeId` is used to produce consistent sort orders for vertex buffer generation, replacing the more expensive and often surprising "name based sorting" * Meshes can be used to generate a `MeshVertexBufferLayout`, which defines the layout of the gpu buffer produced by the mesh. `MeshVertexBufferLayouts` can then be used to generate actual `VertexBufferLayouts` according to the requirements of a specific pipeline. This decoupling of "mesh layout" vs "pipeline vertex buffer layout" is what enables custom attributes. We don't need to standardize _mesh layouts_ or contort meshes to meet the needs of a specific pipeline. As long as the mesh has what the pipeline needs, it will work transparently. * Mesh-based pipelines now specialize on `&MeshVertexBufferLayout` via the new `SpecializedMeshPipeline` trait (which behaves like `SpecializedPipeline`, but adds `&MeshVertexBufferLayout`). The integrity of the pipeline cache is maintained because the `MeshVertexBufferLayout` is treated as part of the key (which is fully abstracted from implementers of the trait ... no need to add any additional info to the specialization key). * Hashing `MeshVertexBufferLayout` is too expensive to do for every entity, every frame. To make this scalable, I added a generalized "pre-hashing" solution to `bevy_utils`: `Hashed<T>` keys and `PreHashMap<K, V>` (which uses `Hashed<T>` internally) . Why didn't I just do the quick and dirty in-place "pre-compute hash and use that u64 as a key in a hashmap" that we've done in the past? Because its wrong! Hashes by themselves aren't enough because two different values can produce the same hash. Re-hashing a hash is even worse! I decided to build a generalized solution because this pattern has come up in the past and we've chosen to do the wrong thing. Now we can do the right thing! This did unfortunately require pulling in `hashbrown` and using that in `bevy_utils`, because avoiding re-hashes requires the `raw_entry_mut` api, which isn't stabilized yet (and may never be ... `entry_ref` has favor now, but also isn't available yet). If std's HashMap ever provides the tools we need, we can move back to that. Note that adding `hashbrown` doesn't increase our dependency count because it was already in our tree. I will probably break these changes out into their own PR. * Specializing on `MeshVertexBufferLayout` has one non-obvious behavior: it can produce identical pipelines for two different MeshVertexBufferLayouts. To optimize the number of active pipelines / reduce re-binds while drawing, I de-duplicate pipelines post-specialization using the final `VertexBufferLayout` as the key. For example, consider a pipeline that needs the layout `(position, normal)` and is specialized using two meshes: `(position, normal, uv)` and `(position, normal, other_vec2)`. If both of these meshes result in `(position, normal)` specializations, we can use the same pipeline! Now we do. Cool! To briefly illustrate, this is what the relevant section of `MeshPipeline`'s specialization code looks like now: ```rust impl SpecializedMeshPipeline for MeshPipeline { type Key = MeshPipelineKey; fn specialize( &self, key: Self::Key, layout: &MeshVertexBufferLayout, ) -> RenderPipelineDescriptor { let mut vertex_attributes = vec![ Mesh::ATTRIBUTE_POSITION.at_shader_location(0), Mesh::ATTRIBUTE_NORMAL.at_shader_location(1), Mesh::ATTRIBUTE_UV_0.at_shader_location(2), ]; let mut shader_defs = Vec::new(); if layout.contains(Mesh::ATTRIBUTE_TANGENT) { shader_defs.push(String::from("VERTEX_TANGENTS")); vertex_attributes.push(Mesh::ATTRIBUTE_TANGENT.at_shader_location(3)); } let vertex_buffer_layout = layout .get_layout(&vertex_attributes) .expect("Mesh is missing a vertex attribute"); ``` Notice that this is _much_ simpler than it was before. And now any mesh with any layout can be used with this pipeline, provided it has vertex postions, normals, and uvs. We even got to remove `HAS_TANGENTS` from MeshPipelineKey and `has_tangents` from `GpuMesh`, because that information is redundant with `MeshVertexBufferLayout`. This is still a draft because I still need to: * Add more docs * Experiment with adding error handling to mesh pipeline specialization (which would print errors at runtime when a mesh is missing a vertex attribute required by a pipeline). If it doesn't tank perf, we'll keep it. * Consider breaking out the PreHash / hashbrown changes into a separate PR. * Add an example illustrating this change * Verify that the "mesh-specialized pipeline de-duplication code" works properly Please dont yell at me for not doing these things yet :) Just trying to get this in peoples' hands asap. Alternative to #3120 Fixes #3030 Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-02-23 23:21:13 +00:00
descriptor: &mut RenderPipelineDescriptor,
key: Self::Key,
_layout: &MeshVertexBufferLayout,
) -> Result<(), SpecializedMeshPipelineError> {
if key.normal_map {
descriptor
.fragment
.as_mut()
.unwrap()
.shader_defs
.push(String::from("STANDARDMATERIAL_NORMAL_MAP"));
}
StandardMaterial: expose a cull_mode option (#3982) This makes it possible for materials to configure front or back face culling, or disable culling. Initially I looked at specializing the Mesh which currently controls this state but conceptually it seems more appropriate to control this at the material level, not the mesh level. _Just for reference this also seems to be consistent with Unity where materials/shaders can configure the culling mode between front/back/off - as opposed to configuring any culling state when importing a mesh._ After some archaeology, trying to understand how this might relate to the existing 'double_sided' option, it was determined that double_sided is a more high level lighting option originally from Filament that will cause the normals for back faces to be flipped. For sake of avoiding complexity, but keeping control this currently keeps the options orthogonal, and adds some clarifying documentation for `double_sided`. This won't affect any existing apps since there hasn't been a way to disable backface culling up until now, so the option was essentially redundant. double_sided support could potentially be updated to imply disabling of backface culling. For reference https://github.com/bevyengine/bevy/pull/3734/commits also looks at exposing cull mode control. I think the main difference here is that this patch handles RenderPipelineDescriptor specialization directly within the StandardMaterial implementation instead of communicating info back to the Mesh via the `queue_material_meshes` system. With the way material.rs builds up the final RenderPipelineDescriptor first by calling specialize for the MeshPipeline followed by specialize for the material then it seems like we have a natural place to override anything in the descriptor that's first configured for the mesh state.
2022-03-05 03:37:23 +00:00
descriptor.primitive.cull_mode = key.cull_mode;
if let Some(label) = &mut descriptor.label {
*label = format!("pbr_{}", *label).into();
}
Mesh vertex buffer layouts (#3959) This PR makes a number of changes to how meshes and vertex attributes are handled, which the goal of enabling easy and flexible custom vertex attributes: * Reworks the `Mesh` type to use the newly added `VertexAttribute` internally * `VertexAttribute` defines the name, a unique `VertexAttributeId`, and a `VertexFormat` * `VertexAttributeId` is used to produce consistent sort orders for vertex buffer generation, replacing the more expensive and often surprising "name based sorting" * Meshes can be used to generate a `MeshVertexBufferLayout`, which defines the layout of the gpu buffer produced by the mesh. `MeshVertexBufferLayouts` can then be used to generate actual `VertexBufferLayouts` according to the requirements of a specific pipeline. This decoupling of "mesh layout" vs "pipeline vertex buffer layout" is what enables custom attributes. We don't need to standardize _mesh layouts_ or contort meshes to meet the needs of a specific pipeline. As long as the mesh has what the pipeline needs, it will work transparently. * Mesh-based pipelines now specialize on `&MeshVertexBufferLayout` via the new `SpecializedMeshPipeline` trait (which behaves like `SpecializedPipeline`, but adds `&MeshVertexBufferLayout`). The integrity of the pipeline cache is maintained because the `MeshVertexBufferLayout` is treated as part of the key (which is fully abstracted from implementers of the trait ... no need to add any additional info to the specialization key). * Hashing `MeshVertexBufferLayout` is too expensive to do for every entity, every frame. To make this scalable, I added a generalized "pre-hashing" solution to `bevy_utils`: `Hashed<T>` keys and `PreHashMap<K, V>` (which uses `Hashed<T>` internally) . Why didn't I just do the quick and dirty in-place "pre-compute hash and use that u64 as a key in a hashmap" that we've done in the past? Because its wrong! Hashes by themselves aren't enough because two different values can produce the same hash. Re-hashing a hash is even worse! I decided to build a generalized solution because this pattern has come up in the past and we've chosen to do the wrong thing. Now we can do the right thing! This did unfortunately require pulling in `hashbrown` and using that in `bevy_utils`, because avoiding re-hashes requires the `raw_entry_mut` api, which isn't stabilized yet (and may never be ... `entry_ref` has favor now, but also isn't available yet). If std's HashMap ever provides the tools we need, we can move back to that. Note that adding `hashbrown` doesn't increase our dependency count because it was already in our tree. I will probably break these changes out into their own PR. * Specializing on `MeshVertexBufferLayout` has one non-obvious behavior: it can produce identical pipelines for two different MeshVertexBufferLayouts. To optimize the number of active pipelines / reduce re-binds while drawing, I de-duplicate pipelines post-specialization using the final `VertexBufferLayout` as the key. For example, consider a pipeline that needs the layout `(position, normal)` and is specialized using two meshes: `(position, normal, uv)` and `(position, normal, other_vec2)`. If both of these meshes result in `(position, normal)` specializations, we can use the same pipeline! Now we do. Cool! To briefly illustrate, this is what the relevant section of `MeshPipeline`'s specialization code looks like now: ```rust impl SpecializedMeshPipeline for MeshPipeline { type Key = MeshPipelineKey; fn specialize( &self, key: Self::Key, layout: &MeshVertexBufferLayout, ) -> RenderPipelineDescriptor { let mut vertex_attributes = vec![ Mesh::ATTRIBUTE_POSITION.at_shader_location(0), Mesh::ATTRIBUTE_NORMAL.at_shader_location(1), Mesh::ATTRIBUTE_UV_0.at_shader_location(2), ]; let mut shader_defs = Vec::new(); if layout.contains(Mesh::ATTRIBUTE_TANGENT) { shader_defs.push(String::from("VERTEX_TANGENTS")); vertex_attributes.push(Mesh::ATTRIBUTE_TANGENT.at_shader_location(3)); } let vertex_buffer_layout = layout .get_layout(&vertex_attributes) .expect("Mesh is missing a vertex attribute"); ``` Notice that this is _much_ simpler than it was before. And now any mesh with any layout can be used with this pipeline, provided it has vertex postions, normals, and uvs. We even got to remove `HAS_TANGENTS` from MeshPipelineKey and `has_tangents` from `GpuMesh`, because that information is redundant with `MeshVertexBufferLayout`. This is still a draft because I still need to: * Add more docs * Experiment with adding error handling to mesh pipeline specialization (which would print errors at runtime when a mesh is missing a vertex attribute required by a pipeline). If it doesn't tank perf, we'll keep it. * Consider breaking out the PreHash / hashbrown changes into a separate PR. * Add an example illustrating this change * Verify that the "mesh-specialized pipeline de-duplication code" works properly Please dont yell at me for not doing these things yet :) Just trying to get this in peoples' hands asap. Alternative to #3120 Fixes #3030 Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-02-23 23:21:13 +00:00
Ok(())
}
fn fragment_shader(_asset_server: &AssetServer) -> Option<Handle<Shader>> {
Some(PBR_SHADER_HANDLE.typed())
}
#[inline]
fn bind_group(render_asset: &<Self as RenderAsset>::PreparedAsset) -> &BindGroup {
&render_asset.bind_group
}
fn bind_group_layout(
render_device: &RenderDevice,
) -> bevy_render::render_resource::BindGroupLayout {
render_device.create_bind_group_layout(&BindGroupLayoutDescriptor {
entries: &[
BindGroupLayoutEntry {
binding: 0,
visibility: ShaderStages::FRAGMENT,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: BufferSize::new(
StandardMaterialUniformData::std140_size_static() as u64,
),
},
count: None,
},
// Base Color Texture
BindGroupLayoutEntry {
binding: 1,
visibility: ShaderStages::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: ShaderStages::FRAGMENT,
ty: BindingType::Sampler(SamplerBindingType::Filtering),
count: None,
},
// Emissive Texture
BindGroupLayoutEntry {
binding: 3,
visibility: ShaderStages::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: ShaderStages::FRAGMENT,
ty: BindingType::Sampler(SamplerBindingType::Filtering),
count: None,
},
// Metallic Roughness Texture
BindGroupLayoutEntry {
binding: 5,
visibility: ShaderStages::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: ShaderStages::FRAGMENT,
ty: BindingType::Sampler(SamplerBindingType::Filtering),
count: None,
},
// Occlusion Texture
BindGroupLayoutEntry {
binding: 7,
visibility: ShaderStages::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: ShaderStages::FRAGMENT,
ty: BindingType::Sampler(SamplerBindingType::Filtering),
count: None,
},
// Normal Map Texture
BindGroupLayoutEntry {
binding: 9,
visibility: ShaderStages::FRAGMENT,
ty: BindingType::Texture {
multisampled: false,
sample_type: TextureSampleType::Float { filterable: true },
view_dimension: TextureViewDimension::D2,
},
count: None,
},
// Normal Map Texture Sampler
BindGroupLayoutEntry {
binding: 10,
visibility: ShaderStages::FRAGMENT,
ty: BindingType::Sampler(SamplerBindingType::Filtering),
count: None,
},
],
label: Some("pbr_material_layout"),
})
}
#[inline]
fn alpha_mode(render_asset: &<Self as RenderAsset>::PreparedAsset) -> AlphaMode {
render_asset.alpha_mode
}
}