HDR content light level (CTA-861.3 / ISOBMFF clli): the maximum content light
level (MaxCLL) and maximum frame-average light level (MaxFALL), both in cd/m²
(nits). Written as the ContentLightLevelBox.
Zero-copy-or-minimal-copy raw YCbCr result. Single-item HEIC images retain
their decoder allocation and expose the visible crop through plane offsets
and strides. Grid images are naturally tightly packed.
Visible layout of one owned image plane. stride and offset are measured
in samples, not bytes. The backing allocation may include coded padding
before, after, or between visible rows.
Decode to display-ready packed 8-bit RGB using a default Decoder.
Monochrome images are expanded to gray RGB. The returned buffer always has
exactly three bytes per pixel.
Decode a display-ready image with one explicit settings value. This creates
a decoder and worker pool for the call; reuse Decoder::from_settings for
a sequence of images.
Decode a HEIF/HEIC file and return typed, owning, strided raw YCbCr
planes using a default Decoder. Eight-bit items use u8; 10/12-bit
items use u16. Single-item images retain coded stride and padding.
Decode just the frame shown at seconds, seeking from the nearest preceding
random-access point. Convenience wrapper over VideoDecoder::decode_frame_at
for one-shot use. The frame rate comes from the stream’s VUI timing; pass an
explicit rate with VideoDecoder::decode_frame_at_fps if the stream has none.