Taken from ISO/IEC 10918-1:1994.
Most hexadecimal digits are written in uppercase; b is written in lowercase to reduce confusion of B with 8.
| sequence | length | abbr | description | reference to standard |
|---|---|---|---|---|
| FF00 | 0 | not a marker; actual 0xFF value in entropy-coded data segment | §§ D.1.6, F.1.2.3 | |
| FF01 | 0 | TEM | temporary value used during arithmetic coding | § D.1.6 |
| FF02-FFbF | ? | RES | reserved | |
| FFC0 | b2-b4 | SOF0 | Start-of-Frame, non-differential, Huffman, baseline DCT | § B.2.2 |
| FFC1 | b2-b4 | SOF1 | Start-of-Frame, non-differential, Huffman, extended sequential DCT | § B.2.2 |
| FFC2 | b2-b4 | SOF2 | Start-of-Frame, non-differential, Huffman, progressive DCT | § B.2.2 |
| FFC3 | b2-b4 | SOF3 | Start-of-Frame, non-differential, Huffman, lossless (sequential) | § B.2.2 |
| FFC4 | b2-b4 | DHT | Define Huffman Tables | § B.2.4.2 |
| FFC5 | b2-b4 | SOF5 | Start-of-Frame, differential, Huffman, sequential DCT | § J |
| FFC6 | b2-b4 | SOF6 | Start-of-Frame, differential, Huffman, progressive DCT | § J |
| FFC7 | b2-b4 | SOF7 | Start-of-Frame, differential, Huffman, lossless (sequential) | § J |
| FFC8 | ? | JPG | reserved for JPEG extensions | |
| FFC9 | b2-b4 | SOF9 | Start-of-Frame, non-differential, arithmetic, extended sequential DCT | § B.2.2 |
| FFCA | b2-b4 | SOF10 | Start-of-Frame, non-differential, arithmetic, progressive DCT | § B.2.2 |
| FFCb | b2-b4 | SOF11 | Start-of-Frame, non-differential, arithmetic, lossless (sequential) | § B.2.2 |
| FFCC | b2-b4 | DAC | Define Arithmetic Coding conditionings | § B.2.4.3 |
| FFCD | b2-b4 | SOF13 | Start-of-Frame, differential, arithmetic, sequential DCT | § J |
| FFCE | b2-b4 | SOF14 | Start-of-Frame, differential, arithmetic, progressive DCT | § J |
| FFCF | b2-b4 | SOF15 | Start-of-Frame, differential, arithmetic, lossless (sequential) | § J |
| FFD0-FFD7 | 0 | RSTm | ReSTart with modulo 8 count m (low nibble of second byte) | § B.2.1 |
| FFD8 | 0 | SOI | Start of Image | § B.2.1 |
| FFD9 | 0 | EOI | End of Image | § B.2.1 |
| FFDA | b2-b4 | SOS | Start of Scan (image data) | § B.2.3 |
| FFDb | b2-b4 | DQT | Define Quantization Tables | § B.2.4.1 |
| FFDC | b2-b4 | DNL | Define Number of Lines | § B.2.5 |
| FFDD | b2-b4 | DRI | Define Restart Interval | § B.2.4.4 |
| FFDE | b2-b4 | DHP | Define Hierarchical Progression | § B.3.2 |
| FFDF | b2-b4 | EXP | EXPand reference components | § B.3.3 |
| FFE0-FFEF | b2-b4 | APPn | application segment n (low nibble of second byte) | § B.2.4.6 |
| FFF0 | b2-b4 | VER | VERsion | ext § B.3 |
| FFF1 | b2-b4 | DTI | Define Tiled Image | ext § B.5.2 |
| FFF2 | b2-b4 | DTT | Define Tile | ext § B.5.3 |
| FFF3 | b2-b4 | SRF | Selectively Refined Frame | ext § B.4.2 |
| FFF4 | b2-b4 | SRS | Selectively Refined Scan | ext § B.4.3 |
| FFF5 | b2-b4 | DCR | Define Component Registration | ext § B.6 |
| FFF6 | b2-b4 | DQS | Define Quantizer Scale selection | ext § B.7 |
| FFF7-FFFD | ? | JPGn | JPEG extension n (low nibble of second byte) | |
| FFFE | b2-b4 | COM | COMment | § B.2.4.5 |
| FFFF | not a marker; sequences of 0xFF can be reduced down to one 0xFF | § B.1.1.2 |
b2-b4: length is specified as a big-endian 16-bit unsigned integer in the two bytes directly following the marker sequence; this length includes the length field but not the marker sequence
references to standard are
- to ISO/IEC 10918-1:1994 by default
- to ISO/IEC 10918-3:1997 if preceded by
ext