Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.08590 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918380559138816 |
|---|---|
| author | Ling, Zeyu Shuai, Qing Zhang, Teng Li, Shiyang Han, Bo Zou, Changqing |
| author_facet | Ling, Zeyu Shuai, Qing Zhang, Teng Li, Shiyang Han, Bo Zou, Changqing |
| contents | Text-to-motion generation has advanced rapidly, yet two challenges persist. First, existing motion autoencoders compress each frame into a single monolithic latent vector, entangling trajectory and per-joint rotations in an unstructured representation that downstream generators struggle to model faithfully. Second, text-to-motion, pose-conditioned generation, and long-horizon sequential synthesis typically require separate models or task-specific mechanisms, with autoregressive approaches suffering from severe error accumulation over extended rollouts.
We present PRISM, addressing each challenge with a dedicated contribution. (1) A joint-factorized motion latent space: each body joint occupies its own token, forming a structured 2D grid (time
joints) compressed by a causal VAE with forward-kinematics supervision. This simple change to the latent space -- without modifying the generator -- substantially improves generation quality, revealing that latent space design has been an underestimated bottleneck. (2) Noise-free condition injection: each latent token carries its own timestep embedding, allowing conditioning frames to be injected as clean tokens (timestep0) while the remaining tokens are denoised. This unifies text-to-motion and pose-conditioned generation in a single model, and directly enables autoregressive segment chaining for streaming synthesis. Self-forcing training further suppresses drift in long rollouts. With these two components, we train a single motion generation foundation model that seamlessly handles text-to-motion, pose-conditioned generation, autoregressive sequential generation, and narrative motion composition, achieving state-of-the-art on HumanML3D, MotionHub, BABEL, and a 50-scenario user study. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_08590 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PRISM: Streaming Human Motion Generation with Per-Joint Latent Decomposition Ling, Zeyu Shuai, Qing Zhang, Teng Li, Shiyang Han, Bo Zou, Changqing Computer Vision and Pattern Recognition Text-to-motion generation has advanced rapidly, yet two challenges persist. First, existing motion autoencoders compress each frame into a single monolithic latent vector, entangling trajectory and per-joint rotations in an unstructured representation that downstream generators struggle to model faithfully. Second, text-to-motion, pose-conditioned generation, and long-horizon sequential synthesis typically require separate models or task-specific mechanisms, with autoregressive approaches suffering from severe error accumulation over extended rollouts. We present PRISM, addressing each challenge with a dedicated contribution. (1) A joint-factorized motion latent space: each body joint occupies its own token, forming a structured 2D grid (time joints) compressed by a causal VAE with forward-kinematics supervision. This simple change to the latent space -- without modifying the generator -- substantially improves generation quality, revealing that latent space design has been an underestimated bottleneck. (2) Noise-free condition injection: each latent token carries its own timestep embedding, allowing conditioning frames to be injected as clean tokens (timestep0) while the remaining tokens are denoised. This unifies text-to-motion and pose-conditioned generation in a single model, and directly enables autoregressive segment chaining for streaming synthesis. Self-forcing training further suppresses drift in long rollouts. With these two components, we train a single motion generation foundation model that seamlessly handles text-to-motion, pose-conditioned generation, autoregressive sequential generation, and narrative motion composition, achieving state-of-the-art on HumanML3D, MotionHub, BABEL, and a 50-scenario user study. |
| title | PRISM: Streaming Human Motion Generation with Per-Joint Latent Decomposition |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.08590 |