Normally gif2png converts each file named on the command line, leaving the original in place. If a name does not have a .gif extension, the unmodified name will be tried first, followed by the name with .gif appended. For each file named foo.gif, a foo.png will be created.

When a multi-image GIF file named foo.gif is converted, gif2png creates multiple PNG files, each containing one frame; their names will be foo.png, foo.p01, foo.p02 etc.

If no source files are specified and stdin is a terminal, gif2png lists a usage summary and version information, then exits.

If no source files are specified, and stdin is a device or pipe, stdin is converted to noname.png. (The program can't be a normal stdin-to-stdout filter because of the possibility that the input GIF might have multiple images).

However, if filter mode is forced (with -f) stdin will be converted to stdout, with gif2png returning an error code if the GIF is multi-image.

The program will preserve the information contained in a GIF file as closely as possible, including GIF comment and application-data extension blocks. All graphics data (pixels, RGB color tables) will be converted without loss of information. Transparency is also preserved. There is one exception; GIF plain-text extensions are skipped.

The program automatically converts interlaced GIFs to interlaced PNGs. It detects images in which all colors are gray (equal R, G, and B values) and converts such images to PNG grayscale. Other images are converted to use the PNG palette type. Duplicate color entries are silently preserved. Unused color-table entries cause an error message.

The action of the program can be modified with the following command-line switches:

-b {#}RRGGBB

Background. Replace transparent pixels with given RGB value, six hexadecimal digits interpreted as two hexits each of red, green, and blue value. The value may optionally be led with a #, HTML-style.

-d

Delete source GIF files after successful conversion.

-f

Filter mode. Convert GIF on stdin to PNG on stdout, return error if the GIF is multi-image.

-m

Preserve file modification time. the PNG output gets the mod time of the input file, not the time it was converted.

-g

Write gamma=1/2.2 and sRGB chunks in the PNG.

-h

Generate PNG color-frequency histogram chunks into converted color files.

-i

Force conversion to interlaced PNG files.

-n

Force conversion to non-interlaced PNG files.

-p

Display progress of PNG writing.

-r

Try to recover data from corrupted GIF files.

-s

Do not translate the GIF Software chunk to a PNG annotation.

-v

Verbose mode; show summary line, -vv enables conversion-statistics and debugging messages.

-w

Web-probe switch; list GIFs that do not have multiple images to stdout. GIFs that fail this filter cause error messages to stderr.

-O

Optimize; remove unused color-table entries. Normally these trigger an error message and disable -d (but conversion is completed anyway). Also, use zlib compression level 9 (best compression) instead of the default level. The recovery algorithm enabled by -r is as follows: Unused color table entries will not trigger an error message as they normally do, but will still be preserved unless -O is also on, in which case they will be discarded. Missing color tables will be patched with a default that puts black at index 0, white at index 1, and supplies red, green, blue, yellow, purple and cyan as the remaining color values. Missing image pixels will be set to 0. Unrecognized or corrupted extensions will be discarded.