dtourolleandClaude Opus 5
e235e99cce
An earlier commit claimed to move `film_sim` from `[tone, colour]` to `[effect]` and did not. It edited `ops/film_sim.yaml`, where `attributes:` is read, validated against the vocabulary, and then dropped: a `rust:` node publishes its own descriptor, and the type still said tone and colour. The stock went on appearing in the Light group beside exposure and again in Colour beside white balance, exactly as before, and every test passed. Nothing caught it because nothing could. The declaration parsed, the parity tests compare ids rather than attributes, and an operation filed under the wrong groups renders perfectly. It surfaced only on screen, as a missing Effects tab — which is indistinguishable from a category that genuinely has nothing in it, and is precisely how `Optics` looked for as long as it was empty. So three changes rather than one: `FilmSim`'s descriptor declares `Attribute::Effect`, which is the move the earlier commit described. `attributes:` joins the keys a `rust:` node may not carry, beside `params`, `uniforms`, `wgsl`, `helpers`, `define` and `label`. The rule was already written — "its descriptor comes from the type" — and attributes were the one field that slipped past it. A key that is silently ignored is worse than one that is rejected, because it reads as though it worked; the eight hand-written declarations lose a line that never did anything. And a test asserts that every attribute the chain carries reaches the tab strip. That is the property that was actually broken, and its failure mode is invisible from every direction: the controls exist, they are in the shader, and there is no way to filter to them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
40 lines
1.9 KiB
YAML
40 lines
1.9 KiB
YAML
# A hand-written node, and the first thing in `Attribute::Optics`.
|
|
#
|
|
# Written, tested and unreferenced until now: `src/ops/vignetting.rs` has
|
|
# existed with a full descriptor and a working polynomial, and without an entry
|
|
# here it was never in `chain()` — so it reached no photograph and no panel.
|
|
# This file is the whole of what was missing, which is the point of `ops/`
|
|
# being the one place the pipeline's order is written down.
|
|
id: vignetting
|
|
order: 5
|
|
|
|
rust: Vignetting
|
|
|
|
why_rust: |
|
|
It carries a lens profile's `pa` coefficients, which are not parameters: they
|
|
come from the body and lens that took the photograph, not from the
|
|
photographer, and no `uniforms:` expression could produce them. The one
|
|
parameter that *is* theirs — the manual trim — is composed with `k1` in Rust,
|
|
because the profile and the trim have to reach the shader as a single
|
|
polynomial rather than as two the fragment would have to add up.
|
|
|
|
placement: |
|
|
First in the chain, ahead of white balance and exposure.
|
|
|
|
Vignetting is what the lens did to the light before the sensor measured it,
|
|
so undoing it belongs with reading the file rather than with editing the
|
|
picture — everything downstream is then working on the frame the lens would
|
|
have delivered had it been even.
|
|
|
|
The ordering is load-bearing rather than tidy. Correcting a corner means
|
|
*dividing* by an attenuation below one, which pushes those pixels up: a fast
|
|
prime wide open needs about two stops there. Run after the tonal stages, that
|
|
recovery happens once the highlights have already been rolled off and
|
|
clipped, so it lifts values that no longer have anywhere to go and the
|
|
corners posterise instead of brightening. Run here, the headroom to hold them
|
|
still exists (ARCH §5.2).
|
|
|
|
Before distortion and chromatic aberration in intent, though those are
|
|
`lens::Warp`s rather than nodes and compose ahead of the fetch, so no `order:`
|
|
relates the two.
|