Merge origin/main into #480 and apply the review:
- the Samsung-with-a-Pentax-maker-note -> PEF rule now also needs raw evidence; its test fixture
gets a CFA photometric, and the same RGB file without it is not a raw;
- for the Pentax family the Exif CFAPattern (0xa302) counts as raw evidence, as main's PEF reader
takes its layout from it (uncompressed / PackBits PEFs);
- Samsung's private raw compressions 32769..=32773 count for SRW (NX1 / NX500 use 32773);
- the engine fallback in files.rs is dropped (main's version kept): a corrupt raw without a
preview stays an import error; the engine regression test for camera-authored TIFFs is kept.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merge origin/main into #555. The decode_hasselblad doc comment records how the coding rule was
established (first hypothesis recalled by the AI author from training data that includes other
converters, corrected for byte order and confirmed by a black-box search over 17,280 layout
hypotheses on the CC0 files; the code is original). The LR-IMP-FORMATS row on main's text now
lists the decoded FFF and lossless-JPEG 3FR files instead of preview-only. RENDER_CACHE_VERSION
37 -> 38.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's reconstruction (#607) replaces this branch's rewrite of highlight::reconstruct. Kept from the
branch: develop_binned_masked (clipping decided per sample before averaging) and the denoised twin
using the same mask. New: highlight::reconstruct_masked, #607's reconstruction with the clipped
channels told by the mask (block means as lower bounds), shared with reconstruct_with through a
generic clip test (no slowdown on full-size images), and CameraColourModel::rebuild_highlights_masked
so binned previews keep the colour model's clipped white. RENDER_CACHE_VERSION 38.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merge origin/main into #560 (fit_preview / camera-look refactors of #528, #583, #568, #616).
`starting_colour` now tries the per-file camera-JPEG fit (with the camera profile when there is
one) first and uses the spectral-sensitivity matrices only when that fit fails, before the neutral
fallback, so no photo with an accepted JPEG fit changes. Tests pin the order through an injectable
fit; docs, parity, ROADMAP and what's new describe it. RENDER_CACHE_VERSION 37 -> 38.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main (#575, #580) rewrote default_crop. The image crop now keeps main's
order for every Sony body (Sony's crop tags, then DefaultCrop, then the
recorded-size crop), so framing does not change. Only the ILCE-7RM4A
distortion warp takes its centre and normalization from DefaultCrop when
the file has one, which keeps the geometry validated against Sony's
exports (centre x 4783.5 on the A7R IVA sample). The DNG round-trip test
now checks both: Sony's crop on the image, the warp centred on the
default crop. RENDER_CACHE_VERSION 38.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's SR2SubIFD white balance (#575) already reads the A900's applied
levels (2688 / 1024 / 1024 / 1516, the same 2.625 / 1.480 gains), so the
hard-coded maker-note 0x0020 block at offset 188 is dropped. The centred
crop now goes through #580's recorded_size_crop (a `centred` flag, only
for the packed 12-bit layout); every other body keeps main's framing.
Adds the CC0 A900 sample to the corpus with checks for its white balance,
black level, bit depth and crop. RENDER_CACHE_VERSION 38.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Testing the first version on ILCE-7RM4 photos found a partly clipped sky between
bare branches leaning cyan to teal under #523's edit, where Lightroom shows white.
Where only green clipped, red and blue say such a sky is a pale cyan-blue; the
rebuild kept that colour until halfway to full clipping, and darkening and a camera
profile showed it. Colour borrowed from another surface (bright foliage below the
sky) could also push green up, with a small step where green first clips.
- highlight::reconstruct_with: the colour of a pale partly clipped pixel
(saturation relative to the colour model's white below 0.2, easing off by 0.4)
fades to the clipped neutral from the first clip on instead of from halfway;
brightness and strongly coloured pixels keep the old schedule.
- Borrowed colour counts less and less as both the pixel's own channels and the
rim's colour contradict it (0.15 to 0.4 stop); either alone can be wrong.
- Right past the rim every estimate continues the rim's colour (it used to hold
the clipped channels at their clip, for the neutral estimate only), so the first
clip has no step whatever lent the estimate. The rim is built whenever there is
clipping.
- Nothing jumps where a later channel clips: in the rim distance and the
contradiction, a clipped channel counts as the lower bound it is, and a channel
goes over to that from 0.75 of its clip; a channel about to clip is a floor
capped like the clip floor it becomes.
- Tests: a pale sky above yellow-green foliage with thin dark branches in it (no
step where green clips, no green from the foliage, the same beside the branches,
neutral well past the clip); the fringe test compares the sky beside the branch
with the sky away from it, on its way to neutral; a band whose red clips second
next to a rim of another colour renders without a step, in six layouts.
- docs/parity.md: the Highlights note. Render cache version 28.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Uncompressed 16-bit words are read in the other byte order only when the
file's order leaves more than 1% of sampled words at or above
1 << max(bits, 14) and the other order leaves none. Measured on the CC0
DSC-R1 (raw.pixls.us 3221). Crop and white balance (SR2SubIFD) are not
addressed. RENDER_CACHE_VERSION 31 (adjust at merge).
Single-strip ORFs at 12.8 bits per pixel are 16-byte blocks: ten little-endian
12-bit samples in bytes 0..14 plus a zero pad byte. Main routed them through the
XZ-2 12-bit word path (stripes, magenta cast). New branch before it; the exactly
12.0 bpp XZ-2 branch is unchanged. Layout measured on CC0 files (log and scripts in
data/testing/retry-olympus-e300 and tools/retry-olympus-e300).
RENDER_CACHE_VERSION bumped to 30 (open PRs use up to 29; adjust at merge).
Canon's SensorInfo left border is not where light starts: on every corpus
body 10-36 columns left of it already hold image (EOS 6D: masked columns
0-70, border 84), and the black level was averaged over all of them. The
6D read 2591 instead of 2047 (4 % of the range), so CR2s rendered dark with
crushed shadows; 40D +141, 5DS R +131, 5D Mark II +73. The masked run is
now measured from the data: columns whose mean stays within 1/256 of the
range of the leftmost columns', minus a two-column margin.
Against Lightroom Classic (EOS 6D still life, 2000 px, ΔE00): base 7.63 ->
5.23 (p95 24.7 -> 11.3); mean over 86 edit variants abs 9.13 -> 6.19,
effect 3.09 -> 2.75.
RENDER_CACHE_VERSION 23.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sensor-clipped pixels are rebuilt at raw decode. The tone map rolls them off to
white at Neutral, so any colour they carry shows only when Highlights, a negative
Exposure or Whites darkens them. They took colour from the nearest unclipped pixel
(dark foliage, lens fringes along branches, pixels demosaiced from clipped samples),
left two clipped channels at their white-balance-scaled clip levels (red or pink),
and were neutral only in camera RGB, not after the colour model (a look fitted to
the camera's JPEG, a camera profile).
- highlight::reconstruct_with: chromaticity only from bright unclipped content away
from edges and from the clip, through a normalised push-pull pyramid; a neutral
estimate blended in where reliable content is scarce at every scale up to a
quarter of the long edge, applied only as far as the pixel lies past the nearest
near-clip unclipped pixels (so a surface still runs into clipping without a
step); clipped pairs keep the estimate's ratio at their clip floors; partly
clipped pixels fade into the clipped neutral on their way from the first channel
clipping to the last; fully clipped pixels become the colour model's neutral.
Pixels without a clipped channel are left exactly as they were.
- engine: the decode colour step as CameraColourModel, with clipped_white, which
keeps the camera's neutral for nearly singular or implausible models.
- Tests in raw (continuity along a ramp through every clip, also as lines 1 to 16
rows thick with no reliable colour, on canvases 32 to 1024 rows), engine
(pipeline renders at Highlights -100, #523's edit and Exposure -2 through the
first clip, over the same sweep; a CC0 ILCE-7RM4 corpus file added to
RAW_SAMPLES) and the GPU equivalence suite.
- Render cache version 27.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
High Efficiency NEFs (HE / HE★, Z 8, Z 9, Z 6III, Z f) keep compression
34713 but carry a JPEG XS-based codestream (SOC + CAP, FF10 FF50) and no
maker-note 0x0096 table. They fell through to "Nikon compressed NEF
without a linearization table", so the Preview only notice gave Z f users
no hint why their raw wasn't decoded or that Lossless compressed NEFs work.
Detect the codestream before asking for the table and report "Nikon High
Efficiency NEF" with the workaround. Add a synthetic regression test, CC0
Z f HE + Lossless samples to the corpus (raw.pixls.us) with a corpus test,
and list HE NEF among the format gaps in parity.md and ROADMAP.md.
No decoder: the format has no public specification and the existing
decoders (LibRaw, rawler) are LGPL and derived from Nikon's binaries.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011bbmJTxYJyYqKfNH8uXRC6
Highlight reconstruction rebuilds clipped channels from the chromaticity of
unclipped pixels nearby. Three changes to that colour:
- an unclipped pixel counts by its brightness ((m / 0.8 clip)^4, at most 1),
filled coarse to fine by push-pull, so dark wires across a bright cloud no
longer tint it (the iPhone frame of #548 turned cyan at Highlights -100);
- pixels within 6 px of clipping (partly demosaiced from clipped samples, 25 %
greener than the sky on that frame) give no colour unless nothing else does;
- the rebuilt colour is drawn toward neutral at its brightest channel as more
channels clip ((n - 1 + s) / 2, s rising as the next channel nears its clip).
Binned previews keep each channel the mean of its block and mark blocks with a
clipped sample in a mask (develop_binned_masked, reconstruct_masked), instead
of taking the block maximum: one clipped specular sample no longer becomes a
green speckle at 500 px (HUAWEI PRA-LX1, WAS-LX1A). The denoised twin uses the
same mask. Camera-colour fitting keeps the old binning.
RENDER_CACHE_VERSION 26 (adjust at merge).
Rows below the split row use the existing lossy Huffman path; from the split
row on every pixel is a (bits - 4)-bit word: a complete prefix code for the
difference category followed by the top bits of the T.81 extra bits, dequantised
to the centre of the kept bin, with the usual two-left predictor and an
unclamped running value. The rule was derived by black-box analysis of the 11
CC0 samples (all decode to the same samples as the analysis decoder).
The decoder checks that the fixed-rate section ends within the last byte of the
strip (whole rows plus 0..7 padding bits) and otherwise keeps refusing the file
(preview only), so a file that does not follow the rule is never decoded wrongly.
RENDER_CACHE_VERSION 28 (main is 22, open PRs use 23-27): adjust at merge.
C5060WZ, C7070WZ, SP-510UZ, SP-550UZ, SP-565UZ and SP-570UZ files say
uncompressed 12-bit with many strips whose byte counts sum to exactly
width x height x 1.5. main refused them because the 12-bit branch wanted a
single strip, so they fell to "Olympus compressed ORF" and opened
preview-only. The samples are packed MSB-first, the strips hold two fields
(even output rows, then odd ones) with one gap at the seam, and the last row
of the second field sits right after the last strip. Accepted only when all of
that holds exactly. CFA comes from the Exif tag, white balance from maker-note
entries 0x1017/0x1018, black 0. RENDER_CACHE_VERSION 27 (adjust at merge).
A DNG can state a white level the sensor never reaches (an iPhone 15 Pro Max Bayer DNG states 4095 but clips at 3567), so clipped highlights were not recognised and white balance tinted them magenta. On decode, samples that form a plateau clearly below the white level, with nothing above it, are raised to the white level. RENDER_CACHE_VERSION 25 (needs adjusting at merge).
Part of #535 (items 4 and 5).
- Pre-2017 bodies without Sony crop tags shot in the camera's 16:9 mode now get
the 16:9 crop of their camera JPEG: width from FullImageSize, anchored left as
before (drops the ILCE-7SM2's 32 padding columns); height from FullImageSize
(ILCE-7SM2) or from an Exif image size of the same width (DSLR-A580, whose
FullImageSize still says 3:2); centred vertically at an even offset, as
measured by registering each camera JPEG on its raw. The aspect now matches
the JPEG, so the camera-look fit runs: ILCE-7SM2 dE 15.2 -> 4.8,
DSLR-A580 dE 26.7 -> 3.9 (vs main 88ea1a8).
- The DSC-R1's SR2 (maker-note FileFormat 1 0 0 0) stores encrypted samples and
decoded to noise; it is now Unsupported in full and header mode and imports
preview-only with that reason.
- Only these 3 of 140 Sony decodes change; render cache version 26 (23-25 are
taken by #568, #542, #575).
- Tests: crop helper (3:2 unchanged, both 16:9 sources, even offsets, hostile
sizes), synthetic Sony file with FileFormat SR2 vs ARW, corpus test of the
two crops and the SR2 refusal; three samples added to the corpus list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Part of #535 (item 1). Stacked on #542 (SR2 black levels); supersedes its
positional keystream tables.
- White balance is WB_RGGBLevels (0x7313), the gains the camera applied:
the raw IFD's plain copy when present, else the copy in the encrypted
SR2SubIFD, which every ARW from the DSLR-A200 on carries (identical to
the plain one on all 57 CC0 files that have both). The maker note's
Tag2010 gains stay only as a last resort.
- 14 bodies had no white balance and rendered green whenever the camera
look was rejected (A450 ... A700, NEX-3/5/C3, SLT-A33/A35/A55,
ILCE-6001, UMC-R10C). On others Tag2010 missed the applied gains by up
to 100 % (ILCE-3500 at 5600 K: R -31 / B +101 % against the camera
JPEG's neutral pixels, now +4 / +0 %; ILCE-7S on Shade -9 / +15 ->
-1 / +1 %; SLT-A57 -25 / +26 -> -6 / -3 %).
- New vendor/sr2.rs decrypts the whole SR2SubIFD, found black-box:
exiftool -v5's decrypted dump as known plaintext gives the keystream
(key 11 22 33 44, the same at every position in 126 CC0 ARWs from 97
bodies, 15 layouts); an exhaustive search over lag pairs found
k[n] = k[n-63] ^ k[n-127] on 32-bit words, so the first 127 words
generate the rest, matching all 126 files (5.2 MB of keystream).
- The SR2 black level (0x7310) is now read through the same decryptor,
replacing #542's head and per-position keystream tables: one SR2
decryption path. Identical results on all 131 Sony files (every
value #542 decoded; the 51 it could not are the files' plain 0x7310).
Unit tests pin the keystream against answers recorded without the
generator: every byte #542 and #148 had recovered, the SHA-256 of the
whole 62112-byte keystream recovered from the CC0 DSLR-A700 sample, and
the encrypted WB_RGGBLevels of four CC0 files (one per layout) with the
levels ExifTool decrypts.
- Sweep (4f350c4 -> this): gains equal ExifTool's decrypted
WB_RGGBLevels on all 100 decoded CC0 CFA files; on 83 files with
>= 200 JPEG-neutral pixels, within 10 % of the camera's applied gains
71 -> 83, over 10 % 5 -> 0. Accepted fits unchanged (max 0.15 dE);
forced fallback median dE 18.50 -> 18.06, neutral-area cast on the 22
changed files 8.8 -> 1.2.
- Corpus test for five CC0 samples (A500, A33, A700, ILCE-3500 5600 K,
ILCE-7S Shade; three added to RAW_SAMPLES); the pre-2017 test's gains
move to WB_RGGBLevels (RX100 III, RX100, ILCE-7RM2: within 2.6 % of
the old Tag2010 values in Auto). Synthetic tests for decryption,
directory parsing, source precedence, black level through the full
DNGPrivateData chain, and hostile input; #542's layout tests (incl.
the A700 trap) now run through the decryptor.
- Render cache version 25.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Part of #535 (items 2 and 3).
- SR2SubIFD black level: decrypt the directory's first entry (always the
black level, 0x7310) to find where its value sits, and decrypt it with the
keystream known for that position (1638, 2166, 2418, 2510). The fixed read at
2510 gave 354-365 on the DSLR-A450/A500/A550 and 975 on the DSLR-A700, where
ExifTool and the data floor say 512 (A700 render: dE 20.1 -> 8.8,
dL* -9.8 -> 0.0). Keystreams recovered black-box (plain 0x7310 and exiftool -v3
as known plaintext), consistent over 115 ARWs from 97 bodies, 15 layouts.
- Downsized lossless M/S ARWs (linear YCbCr): black 1024 and white 16895, i.e.
the recorded 512 plus 512, measured against same-scene full-size codings on
12 files from 4 bodies (offset 1015-1052, scale 0.98-1.02). The 512 pedestal
lifted shadows by ~3 % of white; M/S renders now match their full-size twins
(7RM5 M dE 6.7 -> 3.1).
- Only these 16 of 140 Sony files change; render cache version 24.
- Tests: synthetic SR2 blocks for every known layout (incl. the A700 trap),
YCbCr black/white; corpus test of black levels against the data floor;
A500 and A700 samples added to the corpus list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The YCbCr M coding's pedestal is corrected separately (#535, item 3), so
this test no longer depends on which change lands first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Three CC0 raw.pixls.us files of one ILCE-7CR (61 MP) scene: lossless
compressed L and M (subsampled YCbCr) and compressed ARW2. A raw corpus
test pins model, colour-filter layout, black level, white level, as-shot
gains and default crop, and checks that the lossless and compressed
mosaics agree inside the crop (green block correlation 0.9976). An engine
corpus test checks each gets a fitted camera look, not the neutral
fallback. All nine public codings decode and fit (mean dE76 1.59 against
their camera JPEGs).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sony's Dynamic Range Optimizer (factory default Auto) brightens the darker
regions of the camera JPEG, not the raw. The tone curve fitted to that JPEG
carried the brightening into the raw's starting look: on 8 ILCE-7CR photos
with DRO on, LightCraft rendered +6.4 L* from Lightroom's default (median
dE76 7.5), against -0.2 L* (3.6) for the 16 with DRO off.
- lightcraft_raw::embedded_preview_dynamic_range_optimized reads Sony
maker-note tag 0xb025 (every documented value: 0 off; 1-3, 8-12 and
16-23, i.e. up to Lv8, on). It was present on all 264 Sony raws checked
(98 bodies); 0xb04f only says on/off and is missing on older bodies.
- With DRO on, the look keeps the colour fitted to the JPEG and its tone
curve is lowered, knot by knot, to Sony's curve without DRO wherever that
is darker (assets/camera-tone/sony-dro-off.json, pooled from 12 CC0
raw.pixls.us ARWs of 10 interchangeable-lens bodies; sources and
derivation in the file and assets/ATTRIBUTION.md). The second-last knot
is lowered only as far as keeps the shoulder past the last knot at or
below the fitted curve, so the result is never brighter anywhere.
Compacts that map raw values darker keep their own curve.
- DRO off, unknown and other makers: unchanged (byte-identical renders on
71 + 35 photos). Render cache version 23.
DRO-on photos now land at +0.8 L* from Lightroom (median dE76 3.3), band by
band like the DRO-off ones. Docs: camera-preview-colour.md, parity row.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Compression 32767 with BitsPerSample 12 and one strip of exactly width x height x 1.5 bytes is plain packed 12-bit data: two pixels per three bytes, least-significant-bit first (measured on the DSLR-A900 sample by same-colour neighbour smoothness, 12x smoother than the other orders). RGGB per the file's pattern tag (checked against the camera JPEG), black 128, crop = FullImageSize centred in the frame, white balance from the plain maker-note block. Any other strip size stays Unsupported. RENDER_CACHE_VERSION 22 -> 24 (adjust at merge).
A raw without colour matrices of its own now gets DNG ColorMatrix1/2 and
ForwardMatrix1/2 for its model when the model is one of the 52 cameras
in rawtoaces-data (Academy Software Foundation, Apache-2.0): 22 Canon,
11 Nikon, 11 Sony, 3 Fujifilm and 5 others, plus the other names its
alias list gives (Rebel T3i and Kiss X5 for the EOS 600D; bodies it
lists as sharing a sensor). tools/spectral_camera_table.py fits, under
Standard Illuminant A and D65, a white-preserving 3x3 to the measured
spectral sensitivities over rawtoaces-data's 190 training reflectances
and writes crates/raw/src/spectral_table.rs. No number comes from Adobe
or from a raw decoder. lightcraft_raw::spectral looks a camera up by
Exif make and model; the DNG colour model does the rest.
camera_preview::starting_colour takes the first available of: the
file's own matrices, a camera profile for the model (fitted to the
JPEG for tone), the spectral matrices, the file-local JPEG fit, the
neutral fallback. A profiled model whose profile fit fails takes its
spectral matrices. White balance stays relative to the as-shot look for
every raw without matrices of its own, so Photo::relative_wb and the
controls are unchanged; calibrate still pools covered models.
Tests: one per precedence step (synthetic), the table's lookup, aliases
and DNG shape, and a corpus test of the order on ILCE-7M4 (profile),
EOS 5D Mark III and D5100 (spectral), EOS 7D (own fit) and a DNG. The
ILCE-7RM2 and GFX100S corpus tests now expect the spectral matrices.
Attribution in assets/ATTRIBUTION.md and NOTICE. RENDER_CACHE_VERSION 23.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
