PackBits

PackBits is a simple, lossless run-length encoding scheme used for compressing 8-bit data streams, notably in image formats such as TIFF and various Apple-era formats. It encodes data as a sequence of a single control byte followed by either literal data or a replicated byte, depending on the control value.

The encoding rules are defined by a signed interpretation of the control byte. If the control byte is in the range 0 to 127, the next control value plus one bytes are copied literally to the output. If the control byte is in the range 129 to 255, the next byte is repeated (257 minus the control value) times in the output. A control byte of 128 is a no-op and produces no output. Therefore, literal runs can be 1 to 128 bytes long, while repeated runs are 2 to 128 bytes.

PackBits is designed for simplicity and speed rather than maximum compression. It performs well on data with long runs of identical bytes but less so on highly varied data, where overhead from control bytes can reduce efficiency. The algorithm is lossless and preserves all original data when decoded correctly.

Historically, PackBits has been associated with Apple and Macintosh software and became one of the standard compression options in TIFF 6.0. It remains a widely supported but not universally preferred method for certain image pipelines, particularly where compatibility with older software is important. Implementations typically read the stream until the container format signals end-of-data, handling the 128 no-op tokens accordingly.