Cross-checking synthetic files with LibRaw's dcraw_emu showed it reads
uncompressed RAF samples as little-endian words and unpacked data under
the Nikon compression tag as big-endian words shifted left by four; the
readers and test writers now do the same (CR2, lossless NEF, ARW2 and
PEF already decoded identically, ~86 dB). The default crop and the
final orientation step no longer copy a full frame when they change
nothing, lowering the 24 MP peak by two frame copies.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
ManagedDecoderProvider decodes JPEG/PNG/TIFF natively, DNG and the
supported RAWs through RawDecoder, and falls back to the largest
suitable embedded JPEG (oriented like the main image) for other RAWs,
CR3, HEIF and unsupported bodies. Requests use JPEG DCT scaling, CFA
binning when MaxLongEdge allows, bilinear demosaic for thumbnails and
analysis, and ForcePreview. DecoderFactory builds the client (WIC then
managed) and server chains and the per-format tier table; Probe gives
the per-file tier without decoding.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Each container reader builds a RawFrame for the shared pipeline: CR2
sliced lossless JPEG with SensorInfo crop, masked-column black and
ColorData WB; NEF uncompressed and Nikon Huffman (lossy type 1,
lossless, lossy type 2 incl. split); ARW uncompressed and ARW2 cRAW
with tone curve; PEF uncompressed and Pentax Huffman; RAF uncompressed
Bayer and X-Trans with FujiIFD. Colour comes from a dcraw adobe_coeff
matrix table; bodies without an entry are reported unsupported so the
caller uses the embedded preview. Native RAWs detect the sensor's
saturation level for clipping only, never rescaling per frame.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Reads CFA and LinearRaw DNGs (compression 1/7/8/34892), linearises with
LinearizationTable, BlackLevel (repeat + deltas) and WhiteLevel over the
ActiveArea, applies OpcodeList1-3 (GainMap, WarpRectilinear,
FixVignetteRadial, FixBadPixels*, TrimBounds, MapTable/MapPolynomial,
Delta/ScalePer*), solves white balance per DNG 1.6 chapter 6 (CCT
interpolation, ForwardMatrix, Bradford to D65), demosaics with AHD
(Hirakawa & Parks 2005), bilinear, a neighbourhood interpolation for
non-Bayer CFAs or binning, and reconstructs clipped highlights.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
EncodeRow8/16 reuse one pooled scratch row for RGB-to-gray output
instead of allocating per row; drop unused JPEG marker constants; the
24 MP timing test skips itself when the JIT optimiser is disabled.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Single-strip uncompressed files (Photoshop's default) exceed
TiffReader's bounded read; such strips are now read in pieces, the
uncompressed path skips a copy, and hostile strip/tile counts are
rejected up front.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
16-bit LinearRaw DNG 1.4 with tiles (lossless JPEG or uncompressed),
ColorMatrix1 = XYZ(D65) to linear sRGB, ForwardMatrix1 to XYZ(D50),
D65 illuminant, neutral AsShotNeutral, WhiteLevel 65535 and optional
Exif. LibRaw renders test patches back to the exact source values.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
TiffImageReader decodes strips and tiles (chunky or planar) with none,
LZW (old-style bit order tolerated), Deflate, PackBits and JPEG
compression (JPEGTables, YCbCr or RGB, lossless SOF3 tiles reshaped as
DNG writes them), predictors 2/3 and the DNG x2/x4 variants, 1-16 bit
integer and 16/32-bit float samples; tiles decode in parallel. Output
is a raw TiffRaster or a linear ImageBuffer. JPEG strips match libtiff
bit for bit.
TiffWriter writes 8/16-bit gray/RGB/RGBA, uncompressed or Deflate with
horizontal predictor, and embeds capture metadata from an Exif block.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Every colour type and bit depth (1/2/4/8/16), PLTE + tRNS (palette alpha
and colour keys), Adam7, all five filters, zlib via ZLibStream; CRC
mismatches and truncated data are tolerated with warnings. PngEncoder
gains an additive ImageBuffer overload (8/16-bit sRGB with sRGB chunk)
and the Palette colour type for decoding.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Port libjpeg's resync_to_restart decision rules (skip bogus markers and
stale RSTs, account for lost intervals), reject oversubscribed Huffman
tables before filling the lookahead table (found by fuzzing), and keep
triangle chroma upsampling at 1/8 scale for better thumbnails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC
Baseline, extended 12-bit, progressive and lossless (SOF3) Huffman
decoding; 1/3/4 components with Adobe CMYK/YCCK; all integral sampling
factors with libjpeg fancy upsampling; restart resynchronisation;
DCT-domain scaling. Full-size blocks use a port of libjpeg's ISLOW IDCT,
so output is bit-identical to djpeg on the cjpeg fixtures.
PixelConvert maps interleaved 8/16-bit samples to linear ImageBuffers and
back, gray/RGB, Exif orientation and clamping.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012iLMBNwBE2GnphAsu4cjaC