IT lexicon › Programming › Animation compression

Animation compression

Programming På svenska → Updated: 2026-10-25

Techniques for shrinking the data footprint of skeletal animation — curve fitting, quantisation and dropped keyframes — without visibly changing the motion.

A raw animation clip stores position, rotation and scale for every bone in every frame. With hundreds of bones and thousands of clips, that quickly becomes the heaviest thing on disk and in memory. Compression attacks it from several angles: drop keyframes that can be interpolated back anyway, quantise rotations to fewer bits (ideally as quaternions) and fit smooth curves instead of storing every sample.

The trade-off is always error budget against size: over-quantise and fingertips jitter, where a small rotation error is amplified at the far end of the bone chain. Modern libraries like ACL (Animation Compression Library) pick precision per bone and reach tenfold smaller data with error below what the eye can see. Feet and hands, which touch ground and weapons, get tighter tolerances than the hip.

← Back to the lexicon