OmniCam takes a single RGB panorama and a text instruction as input.
Generating camera trajectories from a single panorama and text using geometry-grounded pose tokens
OmniCam autoregressively outputs pose sequences with point-cloud conditioning, hybrid rotation/translation tokenization, and explicit 3D target anchoring, reducing trajectory errors by 28–47% on the new OmniCaT benchmark.
Chinese Tech
Zhenyang Liu · Chenjie Cao · Yisu Zhang · Xuhui Zuo · Xiangyang Xue · Yanwei Fu · +2 more
Fudan University · Shanghai Innovation Institute · Tencent Hunyuan · Zhejiang University
Research Digest··3 min read
The authors introduce OmniCam, an autoregressive model that generates camera pose sequences from a single panoramic image and a textual trajectory description.
Why this paper
From Tencent Hunyuan and 3 others
In one line
OmniCam generates camera trajectories from a single panorama and text, reducing trajectory errors by 28-47% and collisions by 65.8%.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§