2023-05-09 19:43:09 +02:00

314 lines
11 KiB
Rust

use super::super::{AaMode, Bound, Consts, GlobalsLayouts, Vertex as VertexTrait};
use bytemuck::{Pod, Zeroable};
use std::mem;
use vek::*;
#[repr(C)]
#[derive(Copy, Clone, Debug, Zeroable, Pod)]
pub struct Vertex {
pos_norm: u32,
atlas_pos: u32,
}
impl Vertex {
/// NOTE: meta is true when the terrain vertex is touching water.
pub fn new(atlas_pos: Vec2<u16>, pos: Vec3<f32>, norm: Vec3<f32>, meta: bool) -> Self {
const EXTRA_NEG_Z: f32 = 32768.0;
#[allow(clippy::bool_to_int_with_if)]
let norm_bits = if norm.x != 0.0 {
if norm.x < 0.0 { 0 } else { 1 }
} else if norm.y != 0.0 {
if norm.y < 0.0 { 2 } else { 3 }
} else if norm.z < 0.0 {
4
} else {
5
};
Self {
pos_norm: ((pos.x as u32) & 0x003F) << 0
| ((pos.y as u32) & 0x003F) << 6
| (((pos + EXTRA_NEG_Z).z.clamp(0.0, (1 << 16) as f32) as u32) & 0xFFFF) << 12
| u32::from(meta) << 28
| (norm_bits & 0x7) << 29,
atlas_pos: ((atlas_pos.x as u32) & 0xFFFF) << 0 | ((atlas_pos.y as u32) & 0xFFFF) << 16,
}
}
pub fn new_figure(atlas_pos: Vec2<u16>, pos: Vec3<f32>, norm: Vec3<f32>, bone_idx: u8) -> Self {
let norm_bits = u32::from(norm.x.min(norm.y).min(norm.z) >= 0.0);
let axis_bits = if norm.x != 0.0 {
0
} else if norm.y != 0.0 {
1
} else {
2
};
Self {
pos_norm: pos
.map2(Vec3::new(0, 9, 18), |e, shift| {
(((e * 2.0 + 256.0) as u32) & 0x1FF) << shift
})
.reduce_bitor()
| (((bone_idx & 0xF) as u32) << 27)
| (norm_bits << 31),
atlas_pos: ((atlas_pos.x as u32) & 0x7FFF) << 2
| ((atlas_pos.y as u32) & 0x7FFF) << 17
| axis_bits & 3,
}
}
pub fn make_col_light(
// 0 to 31
light: u8,
// 0 to 31
glow: u8,
col: Rgb<u8>,
ao: bool,
) -> [u8; 4] {
//[col.r, col.g, col.b, light]
// It would be nice for this to be cleaner, but we want to squeeze 5 fields into
// 4. We can do this because both `light` and `glow` go from 0 to 31,
// meaning that they can both fit into 5 bits. If we steal a bit from
// red and blue each (not green, human eyes are more sensitive to
// changes in green) then we get just enough to expand the nibbles of
// the alpha field enough to fit both `light` and `glow`.
//
// However, we now have a problem. In the shader code with use hardware
// filtering to get at the `light` and `glow` attributes (but not
// colour, that remains constant across a block). How do we resolve this
// if we're twiddling bits? The answer is to very carefully manipulate
// the bit pattern such that the fields we want to filter (`light` and
// `glow`) always sit as the higher bits of the fields. Then, we can do
// some modulation magic to extract them from the filtering unharmed and use
// unfiltered texture access (i.e: `texelFetch`) to access the colours, plus a
// little bit-fiddling.
//
// TODO: This isn't currently working (no idea why). See `srgb.glsl` for current
// impl that intead does manual bit-twiddling and filtering.
[
(light.min(31) << 3) | ((col.r >> 1) & 0b111),
(glow.min(31) << 3) | ((col.b >> 1) & 0b111),
(col.r & 0b11110000) | (col.b >> 4),
(col.g & 0xFE) | ao as u8,
]
}
pub fn make_col_light_figure(
// 0 to 31
light: u8,
glowy: bool,
shiny: bool,
col: Rgb<u8>,
) -> [u8; 4] {
let attr = 0 | ((glowy as u8) << 0) | ((shiny as u8) << 1);
[
(light.min(31) << 3) | ((col.r >> 1) & 0b111),
(attr.min(31) << 3) | ((col.b >> 1) & 0b111),
(col.r & 0b11110000) | (col.b >> 4),
col.g, // Green is lucky, it remains unscathed
]
}
/// Set the bone_idx for an existing figure vertex.
pub fn set_bone_idx(&mut self, bone_idx: u8) {
self.pos_norm = (self.pos_norm & !(0xF << 27)) | ((bone_idx as u32 & 0xF) << 27);
}
pub fn desc<'a>() -> wgpu::VertexBufferLayout<'a> {
const ATTRIBUTES: [wgpu::VertexAttribute; 2] =
wgpu::vertex_attr_array![0 => Uint32,1 => Uint32];
wgpu::VertexBufferLayout {
array_stride: Self::STRIDE,
step_mode: wgpu::InputStepMode::Vertex,
attributes: &ATTRIBUTES,
}
}
}
impl VertexTrait for Vertex {
// Note: I think it's u32 due to figures??
// potentiall optimize by splitting
const QUADS_INDEX: Option<wgpu::IndexFormat> = Some(wgpu::IndexFormat::Uint32);
const STRIDE: wgpu::BufferAddress = mem::size_of::<Self>() as wgpu::BufferAddress;
}
#[repr(C)]
#[derive(Copy, Clone, Debug, Zeroable, Pod)]
// TODO: new function and private fields??
pub struct Locals {
model_mat0: [f32; 4],
model_mat1: [f32; 4],
model_mat2: [f32; 4],
model_mat3: [f32; 4],
atlas_offs: [i32; 4],
load_time: f32,
_dummy: [f32; 3],
}
impl Locals {
pub fn new(model_offs: Vec3<f32>, ori: Quaternion<f32>, atlas_offs: Vec2<u32>, load_time: f32) -> Self {
let mat = Mat4::from(ori).translated_3d(model_offs);
let mat_arr = mat.into_col_arrays();
Self {
model_mat0: mat_arr[0],
model_mat1: mat_arr[1],
model_mat2: mat_arr[2],
model_mat3: mat_arr[3],
load_time,
atlas_offs: Vec4::new(atlas_offs.x as i32, atlas_offs.y as i32, 0, 0).into_array(),
_dummy: [0.0; 3],
}
}
}
impl Default for Locals {
fn default() -> Self {
Self {
model_mat0: [1.0, 0.0, 0.0, 0.0],
model_mat1: [0.0, 1.0, 0.0, 0.0],
model_mat2: [0.0, 0.0, 1.0, 0.0],
model_mat3: [0.0, 0.0, 0.0, 1.0],
load_time: 0.0,
atlas_offs: [0; 4],
_dummy: [0.0; 3],
}
}
}
pub type BoundLocals = Bound<Consts<Locals>>;
pub struct TerrainLayout {
pub locals: wgpu::BindGroupLayout,
}
impl TerrainLayout {
pub fn new(device: &wgpu::Device) -> Self {
Self {
locals: device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
label: None,
entries: &[
// locals
wgpu::BindGroupLayoutEntry {
binding: 0,
visibility: wgpu::ShaderStage::VERTEX | wgpu::ShaderStage::FRAGMENT,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
],
}),
}
}
pub fn bind_locals(&self, device: &wgpu::Device, locals: Consts<Locals>) -> BoundLocals {
let bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
label: None,
layout: &self.locals,
entries: &[wgpu::BindGroupEntry {
binding: 0,
resource: locals.buf().as_entire_binding(),
}],
});
BoundLocals {
bind_group,
with: locals,
}
}
}
pub struct TerrainPipeline {
pub pipeline: wgpu::RenderPipeline,
}
impl TerrainPipeline {
pub fn new(
device: &wgpu::Device,
vs_module: &wgpu::ShaderModule,
fs_module: &wgpu::ShaderModule,
global_layout: &GlobalsLayouts,
layout: &TerrainLayout,
aa_mode: AaMode,
) -> Self {
common_base::span!(_guard, "TerrainPipeline::new");
let render_pipeline_layout =
device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
label: Some("Terrain pipeline layout"),
push_constant_ranges: &[],
bind_group_layouts: &[
&global_layout.globals,
&global_layout.shadow_textures,
&global_layout.col_light,
&layout.locals,
],
});
let samples = aa_mode.samples();
let render_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
label: Some("Terrain pipeline"),
layout: Some(&render_pipeline_layout),
vertex: wgpu::VertexState {
module: vs_module,
entry_point: "main",
buffers: &[Vertex::desc()],
},
primitive: wgpu::PrimitiveState {
topology: wgpu::PrimitiveTopology::TriangleList,
strip_index_format: None,
front_face: wgpu::FrontFace::Ccw,
cull_mode: Some(wgpu::Face::Back),
clamp_depth: false,
polygon_mode: wgpu::PolygonMode::Fill,
conservative: false,
},
depth_stencil: Some(wgpu::DepthStencilState {
format: wgpu::TextureFormat::Depth32Float,
depth_write_enabled: true,
depth_compare: wgpu::CompareFunction::GreaterEqual,
stencil: wgpu::StencilState {
front: wgpu::StencilFaceState::IGNORE,
back: wgpu::StencilFaceState::IGNORE,
read_mask: !0,
write_mask: !0,
},
bias: wgpu::DepthBiasState {
constant: 0,
slope_scale: 0.0,
clamp: 0.0,
},
}),
multisample: wgpu::MultisampleState {
count: samples,
mask: !0,
alpha_to_coverage_enabled: false,
},
fragment: Some(wgpu::FragmentState {
module: fs_module,
entry_point: "main",
targets: &[
wgpu::ColorTargetState {
format: wgpu::TextureFormat::Rgba16Float,
blend: None,
write_mask: wgpu::ColorWrite::ALL,
},
wgpu::ColorTargetState {
format: wgpu::TextureFormat::Rgba8Uint,
blend: None,
write_mask: wgpu::ColorWrite::ALL,
},
],
}),
});
Self {
pipeline: render_pipeline,
}
}
}