Fujifilm compressed RAF

Imago decodes uncompressed, lossless compressed and lossy compressed RAF sensor data in pure Rust. The shared RAW decoder serves import, desktop previews, editing, CLI rendering, web and export. These files no longer enter the embedded-JPEG-only fallback. After restarting with the updated build, existing imported preview-only photos need Photo → Reload from Disk to refresh their status; edits are retained.

Fujifilm now uses a guarded colour/tone estimate from each file’s own embedded JPEG when the reference is usable, plus relative as-shot WB and bundled pooled profiles for X-H2S and X-T4. Output pixels still come from the sensor. A rejected profile fit retries the per-file colour fit; if both fail, the neutral fallback remains. Measured sensor calibration and a professional photo organizer fidelity remain separate gaps. See camera colour verification.

Format and implementation

crates/raw/src/vendor/raf.rs reads the RAF header, raw TIFF IFD, crop, CFA, black level and white balance. The big-endian compression field at byte 108 distinguishes uncompressed (0), lossless compressed (2) and lossy compressed (3); testing only the strip's first two bytes would misclassify an uncompressed 16-bit sample of 0x5349.

crates/raw/src/vendor/rafc.rs handles the compression stream:

• A 16-byte IS header carries compression mode, sensor type (Bayer or X-Trans), bit depth, dimensions, padded width, 768-pixel stripe width, stripe count and six-row group count. It must agree with the TIFF dimensions and CFA.
• A big-endian size table, padded to 16 bytes, locates independent entropy streams. Lossy files add a table of one quantizer byte per six-row group for each stripe, each table padded to 16 bytes.
• Each six-row group is collated into three red, six green and three blue vectors. Bayer vectors have 384 samples; X-Trans vectors have 512 slots, including interpolation slots that consume no bits. The last stripe's padding is decoded for predictor state but excluded from the output.
• Colour-vector pairs alternate R/G and G/B. Even columns lead odd columns by seven positions. Two preceding lines of each colour and the preceding line's edge values provide prediction context.
• Signed prediction errors use adaptive Rice coding: a unary prefix followed by a context-dependent number of bits, with a bounded escape for large errors. Three pair phases have separate even/odd statistics. The 81 signed gradient contexts fold into 41 states.
• Lossy quantizers change the error range and reconstruction step. Low-gradient sites retain finer precision (0, 1 or 2), with five folded contexts per precision and statistics that survive main-quantizer changes. The main context statistics reset when its quantizer changes. Their initial sum is max(2, (range + 32) / 64); using 16 as the minimum breaks more heavily quantized files.

Stripes decode in parallel using the existing Rayon dependency. Offsets, sizes, dimensions and entropy reads are checked; truncated data, impossible codes and inconsistent headers return errors. Unknown container compression modes, disagreements between container and stream modes, and compressed X-Trans layouts other than the supported CFA return errors. Header-only probing validates the tables and stripe ranges without decompressing the image.

Verification

Complete sensor arrays were compared sample by sample with an external decoder used only as a black-box oracle (rawpy 0.27.1). No decoder source, the vendor data, camera matrices or profiles were used. This tool is not a product or test dependency.

All 24 files matched the reference sensor arrays exactly. This covers 13 camera bodies and both older and current compression variants. Synthetic tests also cover 12-bit streams; a real compressed 12-bit camera file has not been verified. Support is selected from the file's layout, without a camera-name allowlist. Other bodies using this layout should decode, but this is not a claim that every Fujifilm model has been tested. Older FinePix layouts without the supported raw TIFF IFD remain unsupported.

The 17 CC0 files live in the gitignored corpus. Their published SHA-256 identities are pinned in raf-corpus.sha256; their source licence records were checked as CC0 1.0. cargo xtask corpus --download fetches them. cargo test -p imago-raw corpus_fujifilm_compressed_samples -- --nocapture verifies each available file's checksum before checking its reference sensor sum and position-weighted sum. The test skips absent corpus files; the full-array comparisons were performed locally. No supplied media or large binary fixtures are committed.

Unconditional procedural tests cover Bayer/X-Trans, 12/14/16 bits, partial final stripes, quantizer changes, residual signs and escapes, malformed headers and truncated/random entropy streams. Regressions also check overflowing uncompressed row strides, unsupported modes and CFA layouts, and container/stream mode mismatches.

Local release decode measurements: about 0.16–0.24 seconds for 24–40 MP X-Trans files, 0.29 seconds for 50 MP Bayer, and 0.5–0.6 seconds for 100 MP Bayer. These measure sensor decoding, not the full develop/export pipeline.

App verification imported all four supplied compressed files, confirmed previewOnly: null, applied +2 EV exposure edits and inspected headless UI screenshots. CLI JPEG exports at a 1600-pixel long edge succeeded for all four files (0.58–0.76 seconds each, including process startup).

Sources

• Fabian's original prose description of lossless RAF compression: independent stripes, colour vectors, iteration order, predictors' neighbourhoods, Rice coding and gradient adaptation. Its linked implementations were not consulted.
• ExifTool Fujifilm tag-name documentation: RAF header and raw-IFD metadata fields.
• Fujifilm's compression-mode explanation.
• raw.pixls.us: CC0 camera samples, published checksums and licence metadata.
• rawpy's documented raw_image API: the complete sensor array includes margins, allowing comparisons before demosaic or colour processing.

Exact byte layouts, colour-vector masks, prediction formulas, adaptation parameters and lossy precision selection were established through sample analysis and black-box comparisons.