pamsistoaglyph (1) - Linux Manuals

pamsistoaglyph: convert a single-image stereogram to a red/cyan


pamsistoaglyph - convert a single-image stereogram to a red/cyan anaglyphic image


pamsistoaglyph [--invert] [--sep=number] [--minsep=number] [--gray=number] [in_netpbmfile

All options can be abbreviated to their shortest unique prefix. You may use either white space or an equals sign between an option name and its value.


This program is part of Netpbm(1)

pamsistoaglyph reads a Netpbm image as input and produces a Netpbm image as output.

pamsistoaglyph takes a single-image stereogram (SIS) such as those produced by pamstereogram(1)

and converts it to a red/cyan anaglyphic image such as those produced by ppm3d(1) Many people have trouble tricking their eyes into focusing beyond the image in front of them and are therefore unable to perceive the 3-D shape hidden within a single-image stereogram. Anaglyphic stereograms are easier to perceive in 3-D but require a pair of red/cyan glasses such as those often used to watch 3-D movies. The goal of pamsistoaglyph is to help people who have trouble viewing single-image stereograms see the intriguing 3-D effect.

pamsistoaglyph can convert single-image random-dot stereograms (SIRDS), wallpaper stereograms, and even dual-image stereograms to anaglyphic images.


For most images, no command-line options need to be specified. The following options are available, however, for unusual circumstances:

Swap the left- and right-eye
 images. pamsistoaglyph assumes that its input
 represents a wall-eyed stereogram and generates the anaglyphic
 image accordingly. If the generated image appears to recede into
 the page where it should pop out of the page (and vice versa),
 this typically implies that the input image represents a
 cross-eyed stereogram. Use --invert to correct
 the image depth.

Specify the distance in pixels between the left- and right-eye
 images. Essentially, this corresponds to the distance between
 repetitions of the background pattern.  The --sep
 option should rarely be necessary
 as pamsistoaglyph is fairly good at determining
 automatically the eye-separation distance.

This option is similar to --sep but
 constrains pamsistoaglyph only to
 minimum eye-separation distance. Any distance larger
 than number is acceptable.  The --minsep
 option should rarely be necessary
 as pamsistoaglyph is fairly good at determining
 automatically the eye-separation distance.  The default value for
 the minimum eye-separation distance is 10% of the image width;
 this value seems to work well in practice.

Limit the number of gray levels to use when searching for the
 optimal eye-separation
 distance.  Because pamsistoaglyph looks for
 repeated patterns, it is vulnerable to being confused by slight
 variations in color.  By reducing the input image to grayscale and
 capping the number of gray levels,
 pamsistoaglyph ameliorates the effects of
 unintentional color variations (such as those caused by conversion
 from a low-quality JPEG image, for example). The default of 63
 seems to work well so the --gray option should
 rarely be necessary.


The registration algorithm used by pamsistoaglyph was developed specifically for this program. As far as the author knows, there are no existing algorithms for converting stereograms to anaglyphs. The algorithm works as follows:

Convert the image to grayscale to increase the ability to identify

Count the number of pixels that match N pixels ahead in the
 image for all N in [1, width/2].

Maintain a running mean (μ) and standard deviation (σ) of
 the number of matched pixels.

Store the N corresponding to each spike in the number of
 matched pixels. A spike is defined as a tally that exceeds the
 mean plus one, two, or three standard deviations. Only the first
 spike of a given standard-deviation multiplier is stored.

If a tally greater than μ+3σ was encountered, return the
 corresponding N. If not, then if a tally greater than
 μ+2σ was encountered, return the
 corresponding N. If not, then if a tally greater than
 μ+σ was encountered, return the
 corresponding N. If not, then return the N that
 produces the minimum average distance between matched pixels
 (i.e., #matches divided by #pixels). If no
 such N exceeds the minimum allowable eye-separation value,
 return zero to indicate failure.

If the algorithm returned zero, rerun the algorithm independently
 on each row of the input image and return the median of
 all N that exceed the minimum allowable eye-separation
 value. If no such N exists, abort with an error


Scott Pakin wrote pamsistoaglyph in April 2009. It first appeared in Netpbm in Release 10.47 (June 2009).


Copyright (C) 2009 Scott Pakin, scott+pbm [at]