Commit Graph
4 Commits
Author SHA1 Message Date
Andrew FrahnandClaude Opus 5.5 fb097d61be Rename crates to neutral dac-* codename and binaries to app/app-cli
Mechanical rename right after the fork so upstream merges map cleanly.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-10 09:43:50 +11:00
Brandon ThomasandClaude Opus 5.5 c812352940 M5: highlight reconstruction starts at half resolution (2-2.5x faster, identical output)
A clipped pixel is never valid at full resolution, so its diffused chromaticity was always the
bilinear sample of the half-resolution pyramid level; the full-resolution chroma field (288 MB
at 24 MP) and its fill pass were wasted work. The pyramid now starts at half resolution straight
from the image. Bit-identical to the previous implementation (kept as a test reference over odd
sizes and clipping densities). 24 MP: 319 -> 145 ms isolated; real CR2/ARW loads 0.8-1.1 s ->
0.3-0.5 s under load. Runs on every full-resolution raw load (1:1 loupe, export).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-01 22:35:18 -04:00
Brandon ThomasandClaude Opus 5.5 ec6637f238 M5.4b: faster raw path — bin previews/thumbnails straight from the mosaic
- lightcraft-raw: RawImage::develop_binned(k) — k×k superpixel binning of the CFA
  (Bayer k even, X-Trans k multiple of 3), black/white normalised row by row, default
  crop only, clipping preserved for highlight reconstruction. 2560 px previews of a
  24 MP raw bin 2×, thumbnails 4–8×; exports / 1:1 still demosaic (AHD) at full size.
- highlight::reconstruct and its pyramid fill are rayon-parallel (bit-exact).
- X-Trans directional demosaic: precomputed per-phase neighbour offsets and no
  reflection in the interior, no per-pixel allocation (bit-exact, tested against the
  reference formulation): 2.9 s → 0.75 s CPU at 24 MP.
- files::load_bytes runs on a rayon worker so its short parallel loops don't each
  wait for the pool to wake (dominant cost on a loaded machine); LIGHTCRAFT_PROFILE
  prints per-stage source timings; files::bin_factor picks the block size.
- render_bench ONLY=source: per-file decode / normalise / demosaic / binning / load.
- Fix: u64 overflow on hostile IFD sizes in DNG/ARW/NEF raw-IFD selection (proptest).

load_bytes(2560) CPU time, ARW a7 III 24 MP: 2347 → 240 ms; X-T20: 3120 → 1220 ms.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-01 09:01:53 -04:00
Brandon ThomasandClaude Opus 5.5 05450c791c M3.2-M3.6: lightcraft-raw DNG, LJ92, demosaic, highlights, DNG colour
- LJ92 (T.81 process 14) decoder + Annex K encoder: predictors 1-7,
  1-4 components, 2-16 bit, point transform, restart intervals
- DNG: uncompressed (packed 1-16 bit, 16/24/32-bit float), LJ92 tiles/
  strips, Deflate with predictors 2/3/34892-34895; CFA + LinearRaw;
  linearization, BlackLevelRepeatDim/Delta, WhiteLevel, ActiveArea,
  DefaultCrop; colour tags; OpcodeList1-3 parse/write/apply
- DNG writer (uncompressed / LJ92 tiles) for export and round trips
- demosaic: bilinear, PPG, AHD (Hirakawa-Parks), X-Trans directional
- highlight clip/reconstruct; DNG colour model with dual illuminant,
  forward matrices, as-shot neutral -> xy, Rec.2020 transform
- embedded preview extraction; malformed-input proptests

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-30 11:33:19 -04:00