CurveCodec 2 compresses skeletal animation to 0.22-0.37x ACL bytes, skeleton-agnostic
HKU-CGVU · hf · 2026-10-06
Skeletal animation stores every joint transform at every frame, though most of it is implied by the body. The HKU team measures where redundancy actually lies: the largest saving comes from predicting each quantized curve from its own past, the second from closed-loop per-joint sample selection. Notably, a nearest-neighbour oracle over millions of samples is no better than linear interpolation on the gaps—what a network should learn is the residual distribution.
CurveCodec 2 codes each sub-track as a curve in the log map, quantized in closed loop and thinned to rate-distortion-selected keys, with residuals entropy-coded under a small learned model with bit-exact integer inference across platforms. On 4,472 held-out clips from 33 datasets:
- 0.37x ACL's bytes at 0.01 cm under the worst-case per-joint contract
- 0.22x at 0.1 cm under the mean-error contract
- Decodes on one CPU core and transfers without retraining to a species absent from training
More from Research
- Contravariance theory: for hard enough tasks, brain-DNN convergent evolution is inevitable — aran_nayebi · 2026-10-06
- CoNLL 2023 paper: instruction-tuned GPT models beat children on Theory of Mind tests — dioscuri · 2026-10-06
- How Shazam worked pre-AI: spectrogram peaks hashed into a hashtable lookup — deedydas · 2026-10-06
- AI formal verification hits reality: the halting problem blocks existing codebases — danbri · 2026-10-06
- GeoCR: one generalist model unifies cloud removal across 10 datasets, no fine-tuning needed — KAIST-VIClab · 2026-10-06
- Pi²: turning Wikipedia tables into long-context reasoning training data (COLM 2026) — tuvllms · 2026-10-06