cmptiff Manual Page

SYNOPSIS

cmptiff [options] IMAGE1 IMAGE2

DESCRIPTION

cmptiff takes two TIFF images and outputs the difference, according to the options described below.

OPTIONS

-d abs_thresh

absolute threshold of acceptable error, specified as:

• [1-255] for 8-bit channels, or
• [1-65535] for 16-bit channels, or
• [0.0-1.0] for float images

-t perc_thresh

percentage threshold of acceptable error as fraction of 1: so the equivalent of -d 32 for an 8-bit image is -t .125.

-x o,e

relative acceptable error heuristic tests:

o = acceptable number of outliers, small integer number of samples allowed to be above 'threshold' without triggering an error

e = egregious error window, fractional threshold increment above threshold; any single sample difference above theshold+e classifies the images as significantly different, independent of o.

-m frac

fraction of diffs of non-zero samples which triggers an error, despite being below threshold; e.g. 0.4 means consider the images to be different if more than 40% of the samples differ in some way.

-e

require exact match, else non-zero exit status

-q

quiet mode, suppress details in 'essentially identical' case

-qq

quiet-on-success mode, no output at all for exact matches

-qqq

no output at all, just set exit status

EXAMPLES

• For float images, -d .125 is like -t .125

• For 8-bit images, -d 32 is like -t .125

• A fuzzy test of images from different platforms might be:

  -t .001 -m .3 -x 2,.1