Crate ffcine_core 

ffcine core types shared by every other ffcine crate: the Error type, colour spaces (ColorSpace), 3D LUTs (Lut), probed media info (MediaInfo) and 2D geometry (geom), camera input profiles and their LUTs (camera), and libav-free probes for camera files: ISO BMFF (bmff), WAV (wav) and picture profile detection (profile). No GPU or ffmpeg dependency (feature ffmpeg only adds Error::Ffmpeg).

Re-exports

pub use color::ColorSpace;

pub use lut::Lut;

pub use lut::LutType;

pub use probe::MediaInfo;

pub use probe::Rational;

pub use profile::Confidence;

pub use profile::PictureProfile;

pub use profile::ProfileGuess;

Modules

bmff

A small ISO BMFF (MP4 / QuickTime MOV / CR3) metadata reader, without libav: the header fields ffcine catalog needs (duration, tracks, codecs, frame rate, bit depth, chroma, colr tags, maker atoms, Apple mdta keys, the first DJI CAM meta sample). Seeks over mdat, so it reads a few hundred KB of a multi-GB clip, and runs anywhere std does (macOS offload).

camera

Camera input profiles: which colour-in a source needs, picked from the probe, and the camera LUTs (log -> Rec.709) that go with them.

color

Colour space description, independent of ffmpeg’s and libplacebo’s enums.

geom

Plane geometry for per-frame warps: 3x3 homogeneous matrices in pixel coordinates.

lut

3D LUTs (.cube), parsed on the CPU and uploaded by the backend.

probe

What a media file contains, as reported by a backend (ffcine_render::Backend).

profile

Picture profile detection: which camera curve a clip was shot in (DJI D-Log, D-Log M, D-Log2, HLG, PQ, Canon Log, Apple Log, or plain Rec.709), from container metadata. The grade picks the camera LUT from this, so each answer carries a Confidence and the evidence it came from.

wav

WAV (RIFF / RF64 / BWF) header reader, without libav: format, length and the bext (Broadcast Wave) origination stamp that recorders like the Zoom H2essential write.

Enums

Error

Errors from ffcine and its backends.

Type Aliases

Result