TIFF Tag Predictor

Description

A mathematical operator that is applied to the image data before an encoding scheme is applied.

The general idea is that subsequent pixels of an image resemble each other. Thus, substracting the information from a pixel that is already contained in previous one, is likely to reduce its information density considerably, and aid subsequent compression.

Theoretically, the predictor step is independent from the compression step, and thus can be combined with any compression scheme. It is, however, likely to be most useful when combined with either LZW or Deflate compression.

The specification defines these values:

1 = No prediction scheme used before coding.
2 = Horizontal differencing.

A specification supplement from Adobe "Adobe Photoshop TIFF Technical Note 3" defines this new value:

3 = Floating point horizontal differencing.

LibTiff defines these values:

PREDICTOR_NONE = 1;
PREDICTOR_HORIZONTAL = 2;
PREDICTOR_FLOATINGPOINT = 3;

The DNG specification adds these values:

34892 = Horizontal differencing X2. Same as value 2 (Horizontal differencing) except the pixel two to the left is used rather than the pixel one to the left.
34893 = Horizontal differencing X4. Same as value 2 (Horizontal differencing) except the pixel four to the left is used rather than the pixel one to the left.
34894 = Floating point horizontal differencing X2. Same as value 3 (Floating point horizontal differencing) except the pixel two to the left is used rather than the pixel one to the left.
34895 = Floating point horizontal differencing X4. Same as value 3 (Floating point horizontal differencing) except the pixel four to the left is used rather than the pixel one to the left.