PIXHALVE(1)

NAME

pixhalve - reduce the resolution of a color pix file by one half

On this page

SYNOPSIS
DESCRIPTION
OPTIONS
EXAMPLES
SEE ALSO
AUTHOR
COPYRIGHT
BUG REPORTS

SYNOPSIS

pixhalve [-a] [-s squaresize] [-w file_width] [-n file_height] [file.pix] > outfile.pix

pixhalve reads a color pix(5) image and writes a new pix(5) image at one half the resolution in each direction, so the output has one quarter as many pixels. Because the tool is used mainly to prepare imagery for NTSC television, it converts RGB to YUV and applies a 3x3 pyramid filter to the luminance (Y) channel and a 5x5 pyramid filter to the chrominance (U and V) channels before downsampling.

The input is read from file.pix if given, otherwise from standard input; the result is always written to standard output. The input width must be known so scanlines can be split correctly: it defaults to 512 pixels and may be set with -w, with -s (square images), or detected automatically from a named file with -a. The image height does not need to be specified, as input is processed until end of file.

OPTIONS

-a: Autosize the input by reading the dimensions from the named file. This requires a file argument and has no effect on piped input.
-s squaresize: Set both the input width and (nominally) height to squaresize pixels, for square images. Disables autosizing.
-w file_width: Set the input scanline width to file_width pixels (default 512). Disables autosizing.
-n file_height: Accepted for compatibility and to disable autosizing. The height value itself is ignored, since input is consumed until end of file.

EXAMPLES

Example 1. Downsample a rendered frame for NTSC preview

rt -o frame.pix -w 1024 -n 1024 model.g component &&
pixhalve -s 1024 frame.pix | pix-fb -w 512 -n 512

Renders a 1024x1024 image with rt(1), halves it to 512x512 with the NTSC-tuned YUV filtering, and displays the result on a framebuffer with pix-fb(1).

Example 2. Halve a file twice for a quarter-size thumbnail

pixhalve -a big.pix | pixhalve -w 256 > quarter.pix

Autosizes the first pass from the file header, producing a half-size intermediate, then halves again so the final image is one quarter of the original width (e.g. 1024 to 256).