PCX is an image file format developed by the ZSoft Corporation of Marietta, Georgia, USA. It was the native file format for PC Paintbrush (PCX = "PC Paintbrush Exchange") and became one of the first widely accepted DOS imaging standards, although its use has since been succeeded by more sophisticated image formats such as GIF, JPEG, and PNG.
The PCX is a device-independent raster image format; the file header stores information about the display hardware (screen resolution, color depth and palette information, bit planes and so on) separately from the actual image information, allowing the image to be properly transferred and displayed on computer systems with different hardware. PCX files commonly store palette-indexed images ranging from 2 or 4 colors to 16 and 256 colors, although the format has been extended to record true-color (24-bit) images as well.
|Planes||Bit Depth||Display Type|
|4||1||EGA or VGA|
|3||8||Extended VGA and above|
The PCX file header contains an identifier byte (value 10), a version number, image dimensions, a 16 color palette, number color planes and the bit depth of each plane. While the file header could describe a wide variety of image formats, many are not of practical use. Convention has limited the supported combinations of plane count and bit depth to match specific PC display hardware. Many image editing programs will not read PCX files which do not match one of these conventions. Table A shows the most commonly supported combinations.
The header also contains a value for compression method. All PCX files are written with the same compression scheme and this value is always 1. No other values have been defined and there are no uncompressed PCX files.
PCX version numbers range from 0 to 5, though the file format does not change between versions.
The header is always 128 bytes long though only 74 bytes are used. The rest of the 128 bytes is padded and the image data always begins 128 bytes after the start of the file.
|Row 0||R R R R R R R R R|
|G G G G G G G G|
|B B B B B B B B B|
|A A A A A A A A A|
|Row 1||R R R R R R R R R|
|G G G G G G G G|
|B B B B B B B B B|
|A A A A A A A A A|
|Row 2 etc.||....|
PCX image data is stored in rows or scan lines in top-down order. Where the image has multiple planes these are stored by plane within row. Such that all the red data for row 0 is followed by all the green data for row 0, then all the blue data, then alpha (or intensity) data. This pattern is repeated for each line as shown in Table B.
When an image is less than 8 bits per pixel, each line is padded to the next byte boundary. For example, if an image has 1 plane of 1 bit data (monochrome) with a width of 22 pixels, each row will be 3 bytes long, having 24 bits per row with 2 bits unused.
PCX image data is compressed using run-length encoding (RLE), a simple lossless compression algorithm which collapses a series of three or more consecutive bytes with identical colors into a two-byte pair. As the file is processed, the two most-significant bits of a byte are used to determine whether the given data represents a single pixel of a given palette index or color value, or an RLE pair representing a series of several pixels of a single value. This RLE scheme makes a trade off, it can have more single pixel data but the maximum run length is 63 (compared to the 128 possible with TGA RLE compression).
Due to the use of the two most-significant bits as flags, pixel values from 192 to 255 (with their most-significant bit already set) must be stored in an RLE byte pair even when they only occur one or two pixels in succession, whereas color indexes 0 to 193 can be stored directly or in RLE byte pairs (whichever is more space-efficient); therefore, the actual compression ratio could be optimized with proper sorting of palette entries (though this is not feasible where the file must share its color palette with other images). For example, a palette could be optimized with the most commonly-used colors occurring in palette positions 0 to 191 and the least common colors allocated to the remaining quarter of the palette.
Another inefficiency with the RLE algorithm is that it is possible to store chunks with a length of 0, which allows whitespace in the file (which could be used for steganography), however this does allow it to be decoded a cycle quicker on some processors (namely on the DOS machines it was originally intended for).
The PCX compression algorithm requires very little processor power or memory to apply — a significant concern with the computer systems of the time — but as computers and display hardware grow more sophisticated, the PCX algorithm becomes less space-efficient. Compression algorithms used by newer image formats are more efficient when compressing dithered images such as photographs or complex computer graphics.
If a PCX file has a 256 color palette, it is found 768 bytes from the end of the file. In this case the value in the byte preceding the palette should be 192. Unfortunately, not all programmers kept to this scheme precisely. PCX files have been created with 2, 4, 8, or 16 color palettes at the end of the file rather than in the header. The palette is stored as a sequence of RGB triples. It's usable length defined by the number of colors in the image.
Colors values in PCX palettes always use 8 bits irrespective of the bit depth of the image.
It is possible (though unlikely) for the value 192 to be found 769 bytes from the end of the file as part of the image data. Due to the PCX compression scheme the only way to find the actual length of the image data is to read and process it. This effort is made difficult because the format allows for the compressed data to run beyond the image dimensions, often padding it to the next 8 or 16 line boundary.
There is a multi-page version of PCX which is used by some computer fax and document management programs, using the file extension DCX.