mirror of
https://github.com/storytold/filmcraft.git
synced 2026-10-24 10:10:05 -06:00
aiongg
0a881266c8
* Tests: one integration-test binary per crate Each crate with several tests/*.rs files keeps them as modules of tests/it/, so cargo links one test executable per crate instead of one per file. Files that change process-wide state or time a workload stay binaries of their own. * Profiles: line tables for workspace crates, no debug info for dependencies * Windows ARM64: nightly and on release pushes instead of every push to main * xtask ci: clippy and tests in the dev profile The release profile's thin LTO and single codegen unit made linking the test binaries take most of the gate. The codec and audio crates stay at opt-level 3 in the dev profile, so their tests run at release speed. * xtask ci: nextest, doctests, no examples, one wasm check Tests run with cargo nextest when it is installed (cargo test otherwise) and build with --lib --bins --tests, so examples aren't linked; doctests run separately. The wasm check keeps going past the first failing crate. VP8 fixture generation locks across processes, since nextest runs each test in its own. * VP9 synthetic fixtures: write aside and rename A test in another process could read a fixture ffmpeg was still writing. * Fuzz tests: a third of each sweep by default, all of it with FILMCRAFT_FUZZ_ITERS Every seed, mutation kind and thread count still runs. Header bytes each get one overwrite value and two bit flips in rotation, so every value and bit index still occurs across the region. FILMCRAFT_FUZZ_ITERS=N runs the whole sweep plus N random rounds per seed, as before. * Platform fallback tests: one failure point per hybrid path The failure-anywhere test injected a failure at four points around every restart point; the hybrid's paths differ only in the first GOP, around the second and last restart points, at the end and never. The replay test needs the first two GOPs. * Export tests: shorter ranges with the same checks The bitrate test measures 0.5–1.5 s, where each preset reaches its own target (on 0–3 s the simple demo left two presets at the same rate). The every-preset test exports three frames per preset. * Audio tests: effects in parallel, a table-driven FFT for the Opus quality metric The extreme-settings test keeps both settings and the 40 s tail for every effect, one thread each. The Opus quality metric's FFT builds its twiddles once and allocates nothing per transform (same algorithm and summation order). * CI: dev-profile gate with nextest xtask ci no longer builds with --release, so the release-profile overrides go; nextest is installed for the test step. * CI: apt packages through the cached install-apt-packages action * Review fixes: fuzz tests with a panic hook in their own binaries, loud fixture lock The fuzz tests of av1, h264, hevc, mpeg2v, vp8 and vp9 count panics through a process-wide hook, so a panic in another test of the same process would fail them; they are binaries of their own again. The VP8 fixture lock fails the test instead of skipping it silently, and the full-sweep doc comments say how many truncation points the full run samples. * Fuzz tests: rotate the default subset per run (FILMCRAFT_FUZZ_SEED replays one) * platform tests: decode-switch tests in their own binaries, one NVENC lock
filmcraft-apv
Clean-room Advanced Professional Video (APV) decoder and encoder in pure Rust, implemented strictly from the public IETF RFC 9924 specification (Advanced Professional Video, February 2026, ISSN 2070-1721).
Layer L0: depends only on filmcraft-bitstream, thiserror, and (optional, feature threads,
default on) rayon. Builds for wasm32-unknown-unknown; unsafe_code = "forbid".
API
use filmcraft_apv::*;
// Decode one APV Access Unit (or raw bitstream AU prefixed with a 32-bit au_size).
let frame: Frame = decode_frame(&au)?;
let f8 = decode_frame_with(&au, &DecodeOptions { bit_depth: Some(8), threads: true })?;
decode_frame_into(&au, &DecodeOptions::default(), &mut frame_buf)?;
let hdr: FrameHeader = probe(&au)?;
// Split a multi-AU raw .apv bitstream (RFC 9924 Appendix A).
let aus: Vec<&[u8]> = split_raw_bitstream(&raw_apv_bytes)?;
// Encode progressive intra frames.
let mut enc = Encoder::new(Profile::P422_10, 1920, 1080)?;
let au: Vec<u8> = enc.encode(&frame)?; // Access Unit starting with 'aPv1'
let raw_au: Vec<u8> = enc.encode_raw_au(&frame)?; // Prefixed with 32-bit au_size (ISOBMFF apv1 / .apv)
let apvc: Vec<u8> = enc.decoder_config_record(); // APVDecoderConfigurationRecord payload
Specification Coverage (IETF RFC 9924)
Decoder
- Access Units & PBUs (§5.3.1–5.3.3, Appendix A): Accepts both unprefixed Access Units
(
'aPv1'signature) and length-prefixedraw_bitstream_access_unit()packets. Parses primary (pbu_type == 1), non-primary (2), preview (25), auxiliary alpha (27), AU info (65), metadata (66), and filler (67) PBUs. PBUs or frame headers with non-zeroreserved_zero_*fields are ignored per §5.3.3 and §5.3.5. - Profiles & Sampling (§9.3): All 7 standard profiles (
422-10idc 33,422-12idc 44,444-10idc 55,444-12idc 66,4444-10idc 77,4444-12idc 88,400-10idc 99) at coded bit depths 10–16, with optional output depth rescaling (8..=16). - Quantization Matrices & Tiles (§5.3.7–5.3.8, §5.3.12–5.3.14): Default flat (
16) and custom per-component 8×8 quantization matrices (use_q_matrix), optionaltile_size_in_fhverification, and independent per-tile, per-component byte-aligned entropy partitions. Tiles decode in parallel viarayonwhenthreadsis enabled. - Entropy Decoding (§5.3.15–5.3.16, §7.1): Adaptive Golomb-Rice / Exp-Golomb VLC decoding for
abs_dc_coeff_diff,coeff_zero_run, andabs_ac_coeff_minus1with state tracking (PrevDC,PrevDcDiff,Prev1stAcLevel,PrevLevel,PrevRun) across the 8×8 blocks of each macroblock. - Scaling & Inverse Transform (§6.3): Exact integer dequantization (
levelScale = [40, 45, 51, 57, 64, 71],bdShift = BitDepth - 2) and 2-D separable 8×8 integer inverse transform using the RFC 9924 Figure 25transMatrix. Note: in RFC 9924 §6.3.2.2,transMatrix[m][n]has rowmas the frequency index (row 0 is[64; 8]) and columnnas the spatial sample index, so the 1-D inverse transform evaluatessum_j transMatrix[j][i] * x[j]. - Metadata (§8): Parses
metadata_mdcv(payloadType == 5, Mastering Display Color Volume) andmetadata_cll(payloadType == 6, Content Light-Level Information). - Robustness (§10): All size fields, tile coordinates, VLC codeword lengths, and coefficient
ranges (
-32768..=32767) are checked before allocation or shift. Verified bytests/it/fuzz.rsacross thousands of mutated and random inputs.
Encoder
- Progressive intra encoding for all 7 profiles (
422-10,422-12,444-10,444-12,4444-10,4444-12,400-10). - Supports fixed
qpor binary-search frame-budget rate control over forward-transformed tiles. - Emits conforming tile grids (
tile_width_in_mbs >= 16,tile_height_in_mbs >= 8,TileCols, TileRows <= 20), optional custom quantization matrices, optionalMDCV/CLLmetadata PBUs, andAPVDecoderConfigurationRecord(apvC) metadata.
Accuracy (vs FFmpeg as an External Oracle)
tests/it/oracle.rs encodes synthetic test patterns across all 7 profiles, odd dimensions, multiple
tiles, custom asymmetric quantization matrices, and HDR metadata PBUs, then decodes the resulting
.apv streams with ffmpeg -v error -xerror -f apv and compares against filmcraft_apv::decode_frame.
Because APV specifies an exact integer inverse transform, filmcraft-apv is 100% bit-exact
(max_diff = 0) with FFmpeg's APV decoder across every profile and test case.