A PGX file contains of a single text header line of the form "PG <byte order> [+|-]<bit-depth> <cols> <rows>immediately by the raw binary data. There may be 1, 2 or 4 bytes per sample corresponding to bit depths from 1 to 32. Byte order is "LM" (little endian) or "ML" (big endian). The signed or unsigned pixel representation is indicated by '+' or '-' character before the bit depth. If the '+' is omitted, unsigned is implied.

The input file must be unencapsulated.

PGX supports big or little endian byte order in the case of > 8 bit data. The output byte order is always in the same as the input, since the raw bytes are copied from input to output without any conversion. Use dccp(1) first to change the encoding if necessary.

PGX also supports signed data, so the Pixel Representation of the input is used to specify whether the output is signed or unsigned.