media creative tools
A Video Generator Becomes a One-Pass 4D Observer
Gen4P reuses clean video latents and an on-demand decoder to estimate geometry, motion, depth and camera pose without iterative denoising.
Summary
Gen4P reuses clean video latents and an on-demand decoder to estimate geometry, motion, depth and camera pose without iterative denoising.
Video generators learn more than how frames look. Gen4P feeds clean video latents through a pretrained generator's encoder once, then uses lightweight cross-attention queries to request 4D attributes at chosen places and times. A frozen video-depth adapter supplies extra geometric prior without new trainable parameters. The authors report state-of-the-art or competitive results for 3D point tracking, video depth and camera-pose estimation, plus qualitative motion tracking on in-the-wild clips. Those are research evaluations, not a general claim that generated-video models understand every physical scene.
Why it matters
Gen4P reuses clean video latents and an on-demand decoder to estimate geometry, motion, depth and camera pose without iterative denoising.
Limits and context
- Those are research evaluations, not a general claim that generated-video models understand every physical scene.
Key claims
Gen4P reuses clean video latents and an on-demand decoder to estimate geometry, motion, depth and camera pose without iterative denoising.
Qualification: Those are research evaluations, not a general claim that generated-video models understand every physical scene.
Evidence: source-2026-10-16-003
Sources
- arXiv preprint 2610.22371, original v1arXiv / submitting research authors · primary research
Corrections
No corrections have been recorded for this story.