Let a detail pass carry a list, not only a kernel

Every neighbourhood pass so far has been a convolution, whose whole
description fits in the uniform block because its structure fixes how many
numbers it needs. Spot removal is not that shape: sixty-four repairs and
one repair are the same shader with a different buffer behind it.

So a pass may declare `storage`, which arrives at binding 3 as
`array<vec4<f32>>` with `arrayLength` in scope. The alternative — packing
the list into uniforms — needs a fixed maximum paid for on every frame, a
composer that can emit vec4 fields because a uniform array's stride is 16
whatever it holds, and it gives the next operation that wants a table
nothing to build on.

The property worth having is what stays out of the generated source: the
count is in the buffer, so placing the tenth spot uploads 512 bytes and
reuses the compiled pipeline, exactly as moving a slider does for the
fused pass. `changing_the_list_does_not_recompile` is that, asserted.

One bind group entry rather than two more layouts, and one placeholder
buffer allocated in `new` rather than sixteen bytes per pass per frame —
a zero-length storage buffer cannot be bound, and per-frame allocation is
what this module's documentation exists to refuse.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-08-26 20:00:24 +02:00
co-authored by Claude Opus 5
parent 44444f7768
commit 6997c0f7ac
8 changed files with 287 additions and 5 deletions
+36
View File
@@ -359,6 +359,29 @@ pub struct DetailPass {
/// Uniform values this pass's body reads.
pub uniforms: Vec<Uniform>,
/// TRACES: FR-DEV-8
/// Per-instance data, for a pass whose work is a *list* rather than a
/// kernel.
///
/// Reaches the body as `instances: array<vec4<f32>>`, with
/// `instance_count` in scope as a `u32`. Empty for every pass that is a
/// convolution, which is every pass that existed before spot removal.
///
/// # Why not the uniform block
///
/// Because the uniform block is fixed by the pass's *structure*, and this
/// is not: sixty-four repairs and one repair are the same shader with a
/// different buffer behind it. Packing the list into uniforms would need a
/// fixed maximum, paid for on every frame whether the photograph carries
/// one spot or none, and it would need the composer to emit `vec4` fields —
/// a WGSL uniform array has a stride of 16 whatever it holds.
///
/// The property that matters more: with the list in storage the generated
/// source does not mention how many there are, so placing the tenth spot
/// uploads 512 bytes and reuses the compiled pipeline, exactly as moving a
/// slider does for the fused pass.
pub storage: Vec<[f32; 4]>,
}
/// TRACES: FR-DEV-3 | FR-DEV-8
@@ -396,6 +419,9 @@ pub struct ComposedDetailPass {
pub source: String,
/// Uniform values in the order the generated struct declares them.
pub uniforms: Vec<f32>,
/// TRACES: FR-DEV-8
/// The instance list, if this pass declared one. See [`DetailPass::storage`].
pub storage: Vec<[f32; 4]>,
/// See [`DetailPass::radius`].
pub radius: u32,
/// Whether this pass writes the display/export texture rather than another
@@ -509,6 +535,7 @@ pub fn compose_detail(
radius: 0,
wgsl: String::new(),
uniforms: Vec::new(),
storage: Vec::new(),
},
0,
scale,
@@ -651,6 +678,10 @@ struct Params {{
@group(0) @binding(0) var source: texture_2d<f32>;
@group(0) @binding(1) var<uniform> u: Params;
@group(0) @binding(2) var output: texture_storage_2d<{store_format}, write>;
// A pass whose work is a list rather than a kernel reads it here; every other
// pass leaves this bound to a single empty element and never looks at it. See
// `DetailPass::storage` for why the list is not in the uniform block.
@group(0) @binding(3) var<storage, read> instances: array<vec4<f32>>;
// A neighbour, clamped to the edge of the image.
//
@@ -681,6 +712,10 @@ fn main(@builtin(global_invocation_id) gid: vec3<u32>) {{
// What this render is, relative to the export it has to match.
let render_dims = u.detail_base.xy;
let render_scale = u.detail_base.z;
// How many entries `instances` actually holds, read from the buffer itself
// rather than from a uniform so the two cannot disagree. A pass that
// declared no list is bound to a one-element placeholder and never asks.
let instance_count = arrayLength(&instances);
var c = tap(coord, vec2<i32>(0));
// One scalar per pixel that survives the hand-off from one pass to the
@@ -708,6 +743,7 @@ fn main(@builtin(global_invocation_id) gid: vec3<u32>) {{
label,
source,
uniforms: uniform_values,
storage: pass.storage.clone(),
radius: pass.radius,
writes_output,
structure_hash,
+2
View File
@@ -141,6 +141,8 @@ impl DetailStage for BoxBlur {
.map(|(axis, _)| DetailPass {
label: if axis == 0 { "horizontal" } else { "vertical" },
radius: r,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: vec![
Uniform {
name: "radius",
@@ -358,6 +358,8 @@ impl DetailStage for CaptureSharpen {
.map(|label| DetailPass {
label,
radius: extent,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: vec![
Uniform {
// −100…100 as a gain around zero. A hundred percent is
@@ -410,6 +412,8 @@ fn nothing_to_sharpen() -> DetailPass {
label: "unresolved",
// Reads only the pixel it writes, so a tile needs no halo at all.
radius: 0,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: Vec::new(),
wgsl: "// The chosen radius is finer than one pixel of this render, so the detail
// it would act on is not in this texture — it was lost to the downscale
@@ -475,12 +475,16 @@ impl<B: Band> DetailStage for LocalContrast<B> {
DetailPass {
label: "base",
radius,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: shape,
wgsl: BASE_X.to_string(),
},
DetailPass {
label: "combine",
radius,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: combine,
wgsl: combine_body(B::RECIPE.midtone_taper),
},
@@ -436,6 +436,8 @@ impl DetailStage for NoiseReduction {
passes.push(DetailPass {
label: "luminance",
radius: luma,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: vec![
Uniform {
name: "radius",
@@ -474,6 +476,8 @@ impl DetailStage for NoiseReduction {
"chroma-vertical"
},
radius: chroma,
// A convolution, not a list: nothing to bind at binding 3.
storage: Vec::new(),
uniforms: vec![
Uniform {
name: "radius",