Research map / Video / Video generation
Everybody Dance Now
Caroline Chan, Shiry Ginosar, Tinghui Zhou, Alexei A. Efros
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 temporal smoothing to frame-by-frame synthesis (ablation) | Everybody Dance Now · Held-out target subject, face region | SSIM: 0.784 -> 0.811 | Improved |
| add temporal smoothing to frame-by-frame synthesis (ablation) | Everybody Dance Now · Held-out target subject, body region | LPIPS: 0.057 -> 0.051 | Improved |
| add Face GAN on top of temporal smoothing (ablation) | Everybody Dance Now · Held-out target subject, face region | SSIM: 0.811 -> 0.816 | Improved |
| add Face GAN on top of temporal smoothing (ablation) | Everybody Dance Now · Held-out target subject, face region | LPIPS: 0.039 -> 0.036 | Improved |
| add Face GAN on top of temporal smoothing (ablation) | Everybody Dance Now · Held-out target subject, body region | SSIM: 0.838 -> 0.838 | No clear effect |
| add temporal smoothing and Face GAN to frame-by-frame synthesis (ablation) | Everybody Dance Now · Perceptual study | % preferring full model over ablation: 58.8% | Improved |
| add Face GAN to FBF+TS (ablation) | Everybody Dance Now · Perceptual study | % preferring full model over ablation: 53.9% | Improved |
| add temporal smoothing to FBF (ablation) | Everybody Dance Now · Perceptual study | % preferring FBF+TS: 58.5% | Improved |
| add Face GAN to FBF+TS vs Nearest Neighbors baseline (ablation) | Everybody Dance Now · Perceptual study vs Nearest Neighbors | % preferring our model: 91.2% -> 95.1% | Improved |
| add Face GAN to FBF+TS vs PoseWarp baseline (ablation) | Everybody Dance Now · Perceptual study vs PoseWarp | % preferring our model: 79.1% -> 83.3% | Improved |
| use full model instead of PoseWarp | Everybody Dance Now · Perceptual study | % preferring our method: 83.3% | Improved |
| train separate fake-detector, source motion from Copeland instead of same subject | Everybody Dance Now · Fake detection | accuracy: 95.68% -> 97.00% | Improved |
Closest papers by meaning
- DisCo: Disentangled Control for Realistic Human Dance Generation (CVPR 2024 · similarity 0.86)
- DisMo: Disentangled Motion Representations for Open-World Motion Transfer (NeurIPS 2025 · similarity 0.85)
- HumanDreamer: Generating Controllable Human-Motion Videos via Decoupled Generation (CVPR 2025 · similarity 0.85)
- Move-in-2D: 2D-Conditioned Human Motion Generation (CVPR 2025 · similarity 0.85)
- Make Pixels Dance: High-Dynamic Video Generation (CVPR 2024 · similarity 0.85)
- DanceCamera3D: 3D Camera Movement Synthesis with Music and Dance (CVPR 2024 · similarity 0.85)
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.
