NAME

picfile - raster graphic image format

DESCRIPTION

Files in this format store images represented as two-dimensional arrays of multiple-channel pixels. A picfile consists of an textual header followed by binary data encoding the pixels in row-major order. The header is a list of attribute/value pairs separated by newlines, terminated by an empty line. Each header line has the form name=value The name may not contain an ASCII NUL, newline, or =; the value may not contain NUL or newline. The last line of a header is empty.

The standard attributes are described below; all but TYPE and WINDOW are optional. TYPE must come first; otherwise order is irrelevant. As any unrecognized attribute is passed over uninterpreted by all standard software, applications are welcome to include arbitrary annotations, such as SHOESIZE=10½, if they wish.

TYPE=type

How the pixels are encoded. Standard types are

runcode

A run-length encoding. The data are a sequence of (nchan+1)-byte records each containing a count k and nchan bytes giving a pixel value to be repeated k+1 times. A run may not span scanlines.

dump

A two-dimensional array of nchan-byte records in row major order.

bitmap

One-bit pixels, packed into bytes high bit leftmost. Zero bits are white, one bits are black. Rows are padded with zeros to a multiple of 16 bits.

ccitt-g4

A black-and-white image under CCITT FAX Group 4 compression. This format is highly compressive on images of text and line art. Similarly, ccitt-g31 and ccitt-g32 for Group 3, 1-D and 2-D.

pico

A sequence of nchan two-dimensional arrays of single bytes.

ccir601

Pixels are in dump order, 2 bytes per pixel encoded according to the IEEE digital component video standard.

WINDOW=x0 y0 x1 y1

The x,y coordinates of the upper left corner and the point just diagonally outside the lower right corner, x increasing to the right, y down.

NCHAN=nchan

The number of channels, default 1.

CHAN=channels

The names of the channels. Channels should be nchan characters long. Certain substrings of channels are conventionally understood by most programs that read and write picture files: m is a monochrome image channel, rgb is a full-color image, a is an alpha channel, and z... is a floating point (four-byte, single precision) z value. Some very old monochrome pictures have CHAN=r. This usage is deprecated but still recognized by some programs.

RES=x y

The digitizing resolution horizontally and vertically, in pixels/inch.

CMAP=

(The value is empty.) A color map, a 256×3-byte translation table for color values, follows the header. In a full-color picture, each color-map row maps pixel values of the corresponding channel. In a monochrome picture, pixel values index the color map to yield red, green and blue, like this:

uchar cmap[256][3];

red = cmap[pixel][0];

green = cmap[pixel][1];

blue = cmap[pixel][2];

EXAMPLES

sed '/^$/q' image

Print a header. A sample header follows.

TYPE=dump

WINDOW=0 0 512 512

NCHAN=1

CHAN=m

RES=300 300

CMAP=

COMMAND= antiquantize 'halftone CLASSIC' 512.halftone LIBERTY.anticlassic

COMMAND=  halftone CLASSIC 512.liberty 512.halftone 1.75 512.halftone 

COMMAND=   transpose IN OUT 

COMMAND=    resample 512 IN OUT 

COMMAND=     transpose IN OUT 

COMMAND=      resample 512 IN OUT 

COMMAND=       clip 400 400 LIBERTY OUT

SEE ALSO

bitmap(6)
Tom Duff, ``Raster Graphics in Plan 9''