The code values of every field whose value is a code or a set of flags, from the TIFF 6.0 specification unless a source is named. The fields themselves are listed on Tags. Values of 32768 and above are private allocations. A reader that meets a value it does not support should give up on that image gracefully rather than guess.
Compression (259)
| Value | Name | Defined by | Meaning |
|---|---|---|---|
| 1 | None | TIFF 6.0 baseline | Uncompressed, packed as tightly as possible; rows padded to a byte (or to a SHORT or LONG when every sample is 16 or 32 bits). |
| 2 | CCITT 1D (Modified Huffman) | TIFF 6.0 baseline | Group 3 one-dimensional run-length coding, bilevel only, each row starting on a byte, no EOL codes. |
| 3 | T4 (Group 3 fax) | TIFF 6.0 extension | CCITT T.4 coding, one- or two-dimensional as T4Options says. |
| 4 | T6 (Group 4 fax) | TIFF 6.0 extension | CCITT T.6 two-dimensional coding, ended by EOFB. |
| 5 | LZW | TIFF 6.0 extension | Lempel-Ziv-Welch with 9- to 12-bit codes, each strip coded separately. |
| 6 | JPEG (old style) | TIFF 6.0 extension | The Section 22 design using JPEGProc and the table offset fields. TIFF Technical Note 2 discourages writing it; the number stays reserved. |
| 7 | JPEG | TIFF Technical Note 2 | Each strip or tile is a complete JPEG stream; shared tables may go in JPEGTables. |
| 8 | Deflate (Adobe) | Adobe Photoshop TIFF notes | A zlib stream per strip or tile. |
| 9 | JBIG (T.85) | TIFF-FX, RFC 3949 | ITU-T T.85 JBIG coding for fax. |
| 10 | JBIG colour (T.43) | TIFF-FX, RFC 3949 | ITU-T T.43 colour and grey by layered JBIG. |
| 32766 | NeXT | private (NeXT) | NeXT 2-bit run-length coding. |
| 32771 | CCITT RLE, word-aligned | private | As 2, but each row starts on a 16-bit word. |
| 32773 | PackBits | TIFF 6.0 baseline | Apple Macintosh byte-oriented run-length coding. |
| 32809 | ThunderScan | private (Thunderware) | ThunderScan 4-bit run-length coding. |
| 32895 | IT8 CT with padding | TIFF/IT (ISO 12639) | Continuous-tone picture data. |
| 32896 | IT8 linework RLE | TIFF/IT (ISO 12639) | Run-length coded line work. |
| 32897 | IT8 monochrome picture | TIFF/IT (ISO 12639) | Monochrome picture data. |
| 32898 | IT8 binary line art | TIFF/IT (ISO 12639) | Binary line art. |
| 32908 | Pixar film | private (Pixar) | Companded 10-bit LZW. |
| 32909 | Pixar log | private (Pixar) | Companded 11-bit samples, Deflate coded. |
| 32946 | Deflate (old) | private | Earlier number for Deflate; readers treat it as 8. |
| 32947 | Kodak DCS | private (Kodak) | Kodak DCS camera encoding. |
| 34661 | JBIG | private | ISO JBIG (ISO/IEC 11544). |
| 34676 | SGI LogL | private (SGI) | Run-length coded log luminance (used with Photometric 32844 or 32845). |
| 34677 | SGI LogLuv 24-bit | private (SGI) | 24-bit packed log luminance and chroma. |
| 34712 | JPEG 2000 | private (LEAD Technologies) | A JPEG 2000 codestream. |
| 34887 | LERC | private (Esri) | Esri limited-error raster compression. |
| 34925 | LZMA2 | private | LZMA2 coding. |
| 50000 | Zstandard | LibTIFF, not in Adobe’s registry | A Zstandard frame per strip or tile. |
| 50001 | WebP | LibTIFF, not in Adobe’s registry | A WebP image per strip or tile. |
| 50002 | JPEG XL | LibTIFF, not in Adobe’s registry | A JPEG XL codestream per strip or tile. |
| 52546 | JPEG XL (DNG 1.7) | DNG 1.7 | JPEG XL as defined for DNG. |
Values 9 and above apart from 32773 are not part of TIFF 6.0. The private codecs other than those of TIFF-FX, TIFF/IT and DNG have no public specification this site could find; their names and numbers come from LibTIFF’s header.
PhotometricInterpretation (262)
| Value | Name | Defined by | Meaning |
|---|---|---|---|
| 0 | WhiteIsZero | TIFF 6.0 baseline | Bilevel or grey: 0 is white, 2BPS − 1 is black. Normal for Compression 2. |
| 1 | BlackIsZero | TIFF 6.0 baseline | Bilevel or grey: 0 is black, 2BPS − 1 is white. With CCITT coding the runs are shown reversed. |
| 2 | RGB | TIFF 6.0 baseline | Red, green, blue in that order; 0 is minimum intensity. |
| 3 | Palette colour | TIFF 6.0 baseline | Each pixel is an index into ColorMap; SamplesPerPixel must be 1. |
| 4 | Transparency mask | TIFF 6.0 baseline | A 1-bit mask for another image in the file: 1 inside the region, 0 outside. Goes with NewSubfileType bit 2. |
| 5 | Separated | TIFF 6.0 extension | Ink coverage per component, usually CMYK; larger values mean more ink. |
| 6 | YCbCr | TIFF 6.0 extension | Luma and two chroma components, as CCIR 601 by default. |
| 8 | CIE L*a*b* | TIFF 6.0 extension | 1976 CIELAB: L* unsigned 0 to 100, a* and b* signed 8-bit −127 to 127. |
| 9 | ICC L*a*b* | Adobe TIFF Technical Note 4 | CIELAB encoded as in ICC profiles, with a* and b* unsigned. |
| 10 | ITU L*a*b* | TIFF-FX, RFC 3949 | CIELAB encoded as ITU-T T.42 defines, for fax. |
| 32803 | Colour filter array | TIFF/EP and DNG | Raw sensor data behind a colour filter mosaic. |
| 32844 | LogL | private (SGI) | Log-encoded luminance. |
| 32845 | LogLuv | private (SGI) | Log-encoded luminance with u′ v′ chroma. |
| 34892 | Linear raw | DNG | Demosaiced linear camera data. |
There is no value 7. PhotometricInterpretation has no default, and a reader must not assume one.
NewSubfileType (254), flag bits
| Bit | Mask | Meaning when set |
|---|---|---|
| 0 | 1 | A reduced-resolution version of another image in this file. |
| 1 | 2 | One page of a multi-page image (see PageNumber). |
| 2 | 4 | A transparency mask for another image in this file; PhotometricInterpretation must be 4. |
The other 29 bits are unused and should be 0. The flags are independent; 0 means a single full-resolution image.
SubfileType (255), deprecated
| Value | Meaning |
|---|---|
| 1 | Full-resolution image. |
| 2 | Reduced-resolution image. |
| 3 | One page of a multi-page image. |
Orientation (274)
| Value | Row 0 is the visual… | Column 0 is the visual… | To show it upright |
|---|---|---|---|
| 1 | top | left | nothing |
| 2 | top | right | mirror left to right |
| 3 | bottom | right | rotate 180° |
| 4 | bottom | left | mirror top to bottom |
| 5 | left | top | transpose (swap rows and columns) |
| 6 | right | top | rotate 90° clockwise |
| 7 | right | bottom | transverse (transpose and rotate 180°) |
| 8 | left | bottom | rotate 90° anticlockwise |
The first three columns are the specification’s definitions; the last is what they imply. Baseline readers need support only 1.
Other short code lists
| Field | Value | Meaning |
|---|---|---|
| Threshholding (263) | 1 | No dithering or halftoning applied. |
| 2 | Ordered dither or halftone; CellWidth and CellLength give the matrix size. | |
| 3 | Randomised process such as error diffusion. | |
| FillOrder (266) | 1 | Lower columns, or earlier code bits, in the higher-order bits of each byte. |
| 2 | Lower columns in the lower-order bits. Only for 1-bit data, uncompressed or CCITT; baseline readers need not support it. | |
| PlanarConfiguration (284) | 1 | Chunky: each pixel’s components stored together, e.g. RGBRGB. |
| 2 | Planar: each component in its own plane, with its own strips or tiles. | |
| GrayResponseUnit (290) | 1 | GrayResponseCurve values are tenths of a unit. |
| 2 | Hundredths. | |
| 3 | Thousandths. | |
| 4 | Ten-thousandths. | |
| 5 | Hundred-thousandths. | |
| ResolutionUnit (296) | 1 | No absolute unit; the resolutions give only the aspect ratio. |
| 2 | Inch. | |
| 3 | Centimetre. | |
| Predictor (317) | 1 | No prediction. |
| 2 | Horizontal differencing. | |
| 3 | Floating-point predictor (Adobe Photoshop TIFF Technical Note 3; not in TIFF 6.0). | |
| InkSet (332) | 1 | CMYK, in the order cyan, magenta, yellow, black. InkNames should be absent. |
| 2 | Not CMYK; InkNames names the inks. | |
| ExtraSamples (338), each value | 0 | Unspecified data. |
| 1 | Associated alpha: opacity, with the colour components premultiplied by it. | |
| 2 | Unassociated alpha: a transparency matte independent of the colour. | |
| SampleFormat (339), each value | 1 | Unsigned integer. |
| 2 | Two’s complement signed integer. | |
| 3 | IEEE floating point. | |
| 4 | Undefined: treat as unsigned integer. | |
| 5 | Complex signed integer (LibTIFF; not in TIFF 6.0). | |
| 6 | Complex IEEE floating point (LibTIFF; not in TIFF 6.0). | |
| JPEGProc (512) | 1 | Baseline sequential process. |
| 14 | Lossless process with Huffman coding. | |
| YCbCrPositioning (531) | 1 | Centred: the first chroma sample sits at (subsampling ÷ 2 − 0.5) luma samples in each direction. Needed for PostScript Level 2 and QuickTime. |
| 2 | Co-sited: the first chroma sample sits on the first luma sample. Needed for CCIR 601 video. | |
| Indexed (346) | 0 | Not indexed. |
| 1 | Indexed through a colour map. | |
| OPIProxy (351) | 0 | No higher-resolution version exists. |
| 1 | This is a proxy for a higher-resolution image. |
T4Options (292) and T6Options (293), flag bits
| Field | Bit | Meaning when set |
|---|---|---|
| T4Options | 0 | Two-dimensional coding (otherwise one-dimensional). Each strip must then start with a one-dimensionally coded row. |
| 1 | Uncompressed mode is used. | |
| 2 | Fill bits are added before each EOL so that it ends on a byte boundary. | |
| T6Options | 0 | Unused; always 0. |
| 1 | Uncompressed mode is allowed in the coding. |
All other bits must be 0. Both fields default to 0.
YCbCrSubSampling (530) and YCbCrCoefficients (529)
| Value | First SHORT (horizontal) | Second SHORT (vertical) |
|---|---|---|
| 1 | Chroma as wide as luma | Chroma as tall as luma |
| 2 | Half the width | Half the height |
| 4 | A quarter of the width | A quarter of the height |
The vertical factor must not exceed the horizontal one. ImageWidth, ImageLength, TileWidth and TileLength must be multiples of the factors, and RowsPerStrip a multiple of the vertical one.
| Standard | LumaRed | LumaGreen | LumaBlue |
|---|---|---|---|
| CCIR Recommendation 601-1 (the default) | 299/1000 | 587/1000 | 114/1000 |
| CCIR Recommendation 709 | 2125/10000 | 7154/10000 | 721/10000 |
Useful ReferenceBlackWhite values for YCbCr are 0, 255, 128, 255, 128, 255 (no headroom) and 15, 235, 128, 240, 128, 240 (CCIR 601.1); for 8-bit RGB, 0, 255 three times or 16, 235 three times.
JPEGLosslessPredictors (517)
A is the sample to the left, B the one above, C the one above and to the left. Division by 2 is an arithmetic right shift.
| Value | Prediction |
|---|---|
| 1 | A |
| 2 | B |
| 3 | C |
| 4 | A + B − C |
| 5 | A + ((B − C) ÷ 2) |
| 6 | B + ((A − C) ÷ 2) |
| 7 | (A + B) ÷ 2 |