MOV video sample descriptions

In a video track (handler vide), each entry of the stsd atom is an image description: the common sample description header, fixed video fields, an optional colour table and then extension atoms. Offsets below count from the start of the entry.

Image description

Offset	Size	Type	Name	Meaning
0	4	uint32	sample description size	Bytes in this entry, extensions included.
4	4	FourCC	data format	Compression type or uncompressed pixel format (tables below).
8	6	–	reserved	0.
14	2	uint16	data reference index	1-based index into `dref`.
16	2	int16	version	Version of the compressed data; 0 unless the compressor changed its format.
18	2	int16	revision level	0.
20	4	FourCC	vendor	Maker of the compressor, often `appl`.
24	4	uint32	temporal quality	Degree of temporal compression, 0 to 1023.
28	4	uint32	spatial quality	Degree of spatial compression, 0 to 1024.
32	2	uint16	width	Width of the source image in pixels.
34	2	uint16	height	Height in pixels.
36	4	16.16 fixed	horizontal resolution	Pixels per inch, usually 72.
40	4	16.16 fixed	vertical resolution	Pixels per inch.
44	4	uint32	data size	0.
48	2	uint16	frame count	Frames of compressed data per sample, usually 1.
50	32	Pascal string	compressor name	A length byte and up to 31 characters, such as `jpeg`; the rest is padding.
82	2	int16	depth	Pixel depth: 1, 2, 4, 8, 16, 24 or 32 for colour (32 only with alpha); 34, 36 and 40 for 2-, 4- and 8-bit greyscale.
84	2	int16	color table ID	−1: the default colour table for the depth (a standard Macintosh table below 16 bits; none at 16, 24 and 32). 0: a colour table, in the `ctab` data layout, follows this field.
86	varies	atoms	extensions	Extension atoms, after the colour table if there is one.

Some descriptions end with an optional 4-byte terminator of zeros, counted in the size. Anything other than 0 or 4 bytes left over after the last extension means the description is malformed.

Compression types

Apple’s list, which it says is not complete. More codes, including ProRes and the DV variants, are on Data types and codes.

Code	Format
`cvid`	Cinepak
`jpeg`	JPEG
`smc`	Graphics
`rle`	Animation
`rpza`	Apple Video (“road pizza”)
`kpcd`	Kodak Photo CD
`png`	Portable Network Graphics
`mjpa`	Motion-JPEG format A
`mjpb`	Motion-JPEG format B
`SVQ1`	Sorenson Video 1
`SVQ3`	Sorenson Video 3
`mp4v`	MPEG-4 Part 2 video (needs `esds`)
`avc1`	H.264 (needs `avcC`)
`dvc`	NTSC DV-25
`dvcp`	PAL DV-25
`gif`	CompuServe GIF
`h263`	H.263
`tiff`	TIFF
`raw`	Uncompressed RGB
`2vuY`	Uncompressed Y′CbCr 4:2:2, 8 bits per component
`yuv2`	Uncompressed Y′CbCr 4:2:2, 8 bits per component
`v308`	Uncompressed Y′CbCr 4:4:4, 8 bits per component
`v408`	Uncompressed Y′CbCr 4:4:4:4, 8 bits per component
`v216`	Uncompressed Y′CbCr 4:2:2, 10, 12, 14 or 16 bits per component
`v410`	Uncompressed Y′CbCr 4:4:4, 10 bits per component
`v210`	Uncompressed Y′CbCr 4:2:2, 10 bits per component

Apple writes the 8-bit 4:2:2 code as 2vuY; files in practice use 2vuy (lower-case y), which is how FFmpeg and Apple’s own Core Video name it.

Extensions

Atom	Meaning
`gama`	16.16 fixed gamma at which the image was captured. Superseded by `colr`; ignore it when `colr` is present.
`fiel`	Field handling, two bytes (below). Required for every uncompressed Y′CbCr format.
`mjqt`	Default quantisation table for a Motion-JPEG stream.
`mjht`	Default Huffman table for a Motion-JPEG stream.
`esds`	MPEG-4 elementary stream descriptor; required for `mp4v`.
`avcC`	AVC decoder configuration record; required for `avc1`.
`pasp`	Pixel aspect ratio; required when pixels are not square.
`colr`	Colour parameters; required for uncompressed Y′CbCr.
`clap`	Clean aperture; required for uncompressed Y′CbCr.

fiel

Offset	Size	Name	Meaning
8	1	field count	1 progressive, 2 interlaced.
9	1	field ordering	When there are two fields, T being the field with the topmost line and B the other (table below).

Value	Displayed first	Stored first
0	Only one field.
1	T	T
6	B	B
9	B	T
14	T	B

With two fields, each sample holds two separately compressed images, one per field.

esds and avcC

esds: version (uint8, 0), flags (3 bytes, 0), then an MPEG-4 elementary stream descriptor as defined in ISO/IEC 14496-1, with the limits of ISO/IEC 14496-14. Present only with mp4v.

avcC: an AVCDecoderConfigurationRecord as defined in ISO/IEC 14496-15, the same as in MP4. Present only with avc1. Apple’s documentation stops here; HEVC files use hvc1 or hev1 with hvcC, also from ISO/IEC 14496-15.

pasp

Offset	Size	Name	Meaning
8	4	hSpacing	uint32 horizontal spacing of pixels.
12	4	vSpacing	uint32 vertical spacing of pixels.

Only the ratio matters. Common values:

Picture	hSpacing	vSpacing
4:3 square pixels (composite NTSC or PAL)	1	1
4:3 non-square 525 (NTSC)	10	11
4:3 non-square 625 (PAL)	59	54
16:9 analogue (composite NTSC or PAL)	4	3
16:9 digital 525 (NTSC)	40	33
16:9 digital 625 (PAL)	118	81
1920 × 1035 HDTV (SMPTE 260M-1992)	113	118
1920 × 1035 HDTV (SMPTE RP 187-1995)	1018	1062
1920 × 1080 or 1280 × 720 HDTV	1	1

colr

Offset	Size	Name	Meaning
8	4	colour parameter type	`nclc` for video; `prof` for print, followed by an ICC profile instead of the fields below.
12	2	primaries index	Chromaticities of the primaries and white point.
14	2	transfer function index	Transfer function between linear RGB and Y′CbCr.
16	2	matrix index	Matrix between R′G′B′ and Y′CbCr.

Primaries	Meaning
0	Reserved
1	ITU-R BT.709: white 0.3127, 0.3290 (D65); red 0.640, 0.330; green 0.300, 0.600; blue 0.150, 0.060
2	Unknown
3, 4	Reserved
5	ITU-R BT.601 625-line: white D65; red 0.64, 0.33; green 0.29, 0.60; blue 0.15, 0.06
6	ITU-R BT.601 525-line: white D65; red 0.630, 0.340; green 0.310, 0.595; blue 0.155, 0.070
7, 8	Reserved
9	ITU-R BT.2020: white D65; red 0.708, 0.292; green 0.170, 0.797; blue 0.131, 0.046
10	Reserved
11	SMPTE RP 431-2 (DCI P3): white 0.314, 0.351; red 0.680, 0.320; green 0.265, 0.690; blue 0.150, 0.060
12	SMPTE EG 432-1 (P3 D65, Display P3): white D65; primaries as index 11
13 to 65535	Reserved

Transfer function	Meaning
0	Reserved
1	ITU-R BT.709-2, SMPTE 274M-1995, 296M-1997, 293M-1996, 170M-1994
2	Unknown
3 to 6	Reserved (MPEG-2’s code 6 is the same curve as 1; writers map 6 to 1)
7	SMPTE 240M-1995 and 274M-1995
8 to 16	Reserved
17	SMPTE ST 428-1 (peak white normally 48 cd/m²)
18 to 65535	Reserved

Matrix	Meaning
0	Reserved
1	ITU-R BT.709-2 (1125/60/2:1 only), SMPTE 274M-1995, 296M-1997
2	Unknown
3 to 5	Reserved
6	ITU-R BT.601-4 and BT.470-4 System B and G, SMPTE 170M-1994, 293M-1996
7	SMPTE 240M-1995, 274M-1995
8	Reserved
9	ITU-R BT.2020 non-constant luminance
10 to 65535	Reserved

Apple defines two ways of storing Y′CbCr. Scheme A, wide range: Y′ unsigned over the full range (0 to 255 at 8 bits) and Cb, Cr two’s complement (−127 to +127). Scheme B, video range, as in ITU-R BT.601: Y′ 16 to 235 and Cb, Cr offset binary 16 to 240 at 8 bits; 64 to 940 and 64 to 960 at 10 bits (Apple’s text gives 65 as the lower chroma limit, which differs from BT.601’s 64). In scheme B the values 0 and 255 (8-bit), or 0 to 3 and 1020 to 1023 (10-bit), are reserved for synchronisation and never stored. Values just outside the nominal range carry filter overshoot and must be accepted.

clap

The clean aperture: the part of the stored picture meant to be shown, as a size and an offset of its centre from the picture centre. Each value is a fraction of two int32, a numerator then a denominator (1 for whole numbers).

Offset	Size	Name	Meaning
8	4	apertureWidth_N	Width in pixels, or its numerator.
12	4	apertureWidth_D	Denominator, or 1.
16	4	apertureHeight_N	Height in lines, or its numerator.
20	4	apertureHeight_D	Denominator, or 1.
24	4	horizOff_N	Horizontal offset of the aperture centre minus (width − 1) ÷ 2, usually 0; or its numerator.
28	4	horizOff_D	Denominator, or 1.
32	4	vertOff_N	Vertical offset of the aperture centre minus (height − 1) ÷ 2, usually 0; or its numerator.
36	4	vertOff_D	Denominator, or 1.

Width and height must be positive; offsets may be negative.

.MOV File Format – Video sample descriptions