Research map / Video / Video generation
Unified Camera Positional Encoding for Controlled Video Generation
Cheng Zhang, Boying Li, Meng Wei, Yan-Pei Cao, Camilo Gambardella, Dinh Phung, Jianfei Cai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use 1/8 compression instead of 1/2 (ablation) | UCPE adapter · Synthesized camera-diverse dataset (w/ abs orientation) | FVD: 534.44 -> 495.14 | Improved |
| use 1/8 compression instead of 1/12 (ablation) | UCPE adapter · Synthesized camera-diverse dataset (w/ abs orientation) | RotErr (deg): 5.13 -> 4.12 | Improved |
| use 1/8 compression instead of 1/4 (ablation) | UCPE adapter · Synthesized camera-diverse dataset (w/ abs orientation) | RotErr (deg): 3.43 -> 4.12 | No clear effect |
| use parallel adapter instead of pre-attention insertion (ablation) | UCPE · Synthesized camera-diverse dataset (w/ abs orientation) | FoV error (deg): 8.47 -> 8.22 | Improved |
| use parallel adapter instead of post-attention insertion (ablation) | UCPE · Synthesized camera-diverse dataset (w/ abs orientation) | RotErr-based CamMC: 20.00 -> 17.59 | Improved |
| use relative ray encoding instead of PRoPE (ablation) | UCPE · Synthesized camera-diverse dataset (w/ abs orientation) | RotErr (deg): 5.35 -> 4.12 | Improved |
| use relative ray encoding instead of GTA (ablation) | UCPE · Synthesized camera-diverse dataset (w/ abs orientation) | FoV error (deg): 8.80 -> 8.22 | Improved |
| use relative ray encoding instead of PRoPE/GTA (ablation) | UCPE · Synthesized camera-diverse dataset (w/ abs orientation) | pitch/roll error: 4.18 -> 4.35 (pitch, PRoPE vs UCPE) | No clear effect |
| add Lat-Up absolute orientation encoding | UCPE · Synthesized camera-diverse dataset | FoV error (deg): 9.62 -> 8.22 | Improved |
| add Lat-Up absolute orientation encoding | UCPE · Synthesized camera-diverse dataset | pitch error (deg): 8.03 -> 4.35 | Improved |
| use UCPE instead of ReCamMaster | Wan2.1 DiT · Synthesized camera-diverse dataset (w/o abs orientation) | RotErr (deg): 10.89 -> 4.29 | Improved |
| use UCPE instead of AC3D | camera-controlled T2V · RealEstate10K | RotErr (deg): 0.62 -> 0.56 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReCamMaster — ReCamMaster: Camera-Controlled Generative Rendering from A Single Video (ICCV 2025) · better on 6 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Cameras as Relative Positional Encoding (NeurIPS 2025 · similarity 0.92)
- CRePE: Curved Ray Expectation Positional Encoding for Unified-Camera-Controlled Video Generation (NeurIPS 2026 · similarity 0.90)
- URoPE: Universal Relative Position Embedding across Geometric Spaces (ECCV 2026 · similarity 0.90)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.88)
- VD3D: Taming Large Video Diffusion Transformers for 3D Camera Control (ICLR 2025 · similarity 0.88)
- DPPE: Rethinking Camera-Based Positional Encoding for Scaling Multi-View Transformers (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
