Research map / Video / Video generation
EgoControl: Controllable Egocentric Video Generation via 3D Full-Body Poses
Enrico Pallotta, Sina Mokhtarzadeh Azar, Lars Doorenbos, Serdar Ozsoy, Umar Iqbal, Juergen Gall
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 |
|---|---|---|---|
| add full-body pose conditioning on top of head-only control (ablation) | EgoControl · Nymeria | mIoU: 33.70 -> 52.13 | Improved |
| add full-body pose conditioning on top of head-only control (ablation) | EgoControl · Nymeria | TransError: 5.16 -> 4.90 | Improved |
| add head-pose conditioning to finetuned video prediction (ablation) | Cosmos finetuned · Nymeria | FVD: 40.70 -> 22.68 | Improved |
| finetune Cosmos on Nymeria (ablation) | Cosmos-predict2 2B · Nymeria | FVD: 71.00 -> 40.70 | Improved |
| add cross-attention to AdaLN pose control (ablation) | EgoControl · Nymeria ablation | TransError: 6.07 -> 5.59 | Improved |
| add AdaLN to cross-attention pose control (ablation) | EgoControl · Nymeria ablation | mIoU: 37.84 -> 37.40 | No clear effect |
| add cross-attention to AdaLN-only control (ablation) | EgoControl · Nymeria ablation | mIoU: 33.44 -> 37.40 | Improved |
| use differential instead of cumulative head representation (ablation) | EgoControl head-only · Nymeria | TransError: 12.31 -> 6.75 | Improved |
| use pelvis-relative joints plus head and root deltas instead of per-joint deltas (ablation) | EgoControl · Nymeria | mIoU: 31.85 -> 37.40 | Improved |
| compare to PEVA at single-frame horizon | EgoControl vs PEVA XXL · Nymeria 2 s ahead | LPIPS: 29.8 -> 24.3 | Improved |
| compare to camera-control model CAMI2V | EgoControl vs CAMI2V · Nymeria first 16 frames | mIoU: 24.72 -> 56.85 | Improved |
Closest papers by meaning
- Whole-Body Conditioned Egocentric Video Prediction (NeurIPS 2025 · similarity 0.91)
- EgoVid-5M: A Large-Scale Video-Action Dataset for Egocentric Videos Generation (NeurIPS 2025 · similarity 0.89)
- EgoGen: An Egocentric Synthetic Data Generator (CVPR 2024 · similarity 0.88)
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.88)
- FRAME: Floor-aligned Representation for Avatar Motion from Egocentric Video (CVPR 2025 · similarity 0.88)
- Estimating Ego-Body Pose from Doubly Sparse Egocentric Video Data (NeurIPS 2024 · similarity 0.87)
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.
