Research map / Video / Video generation
Temporally Consistent Semantic Video Editing
Yiran Xu, Badour AlBahar, Jia-Bin Huang
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 |
|---|---|---|---|
| apply the flow-based two-phase optimization to directly edited video (in-domain) | Direct editing · RAVDESS in-domain editing | Warping error: 0.0076 -> 0.0031 | Improved |
| apply the two-phase optimization to directly edited video (out-of-domain) | Direct editing · RAVDESS out-of-domain editing | Warping error: 0.0098 -> 0.0076 | Improved |
| add phase 2 generator finetuning after latent code update (in-domain) (ablation) | Phase 1 only · RAVDESS in-domain editing | LPIPS to direct editing: 0.2108 -> 0.1253 | Improved |
| add phase 2 generator finetuning after latent code update (in-domain) (ablation) | Phase 1 only · RAVDESS in-domain editing | Warping error: 0.0064 -> 0.0031 | Improved |
| add phase 1 latent code update before finetuning the generator (in-domain) (ablation) | Generator finetuning only · RAVDESS in-domain editing | LPIPS to direct editing: 0.2463 -> 0.1253 | Improved |
| add phase 1 latent code update before finetuning the generator (in-domain) (ablation) | Generator finetuning only · RAVDESS in-domain editing | Warping error: 0.0027 -> 0.0031 | Hurt |
| add phase 1 latent code update before finetuning the generator (out-of-domain) (ablation) | Generator finetuning only · RAVDESS out-of-domain editing | Warping error: 0.0057 -> 0.0076 | Hurt |
| add phase 1 latent code update before finetuning the generator (out-of-domain) (ablation) | Generator finetuning only · RAVDESS out-of-domain editing | LPIPS to direct editing: 0.1375 -> 0.1275 | Improved |
| ours vs DVP on in-domain warping error | DVP · RAVDESS in-domain editing | Warping error: 0.0036 -> 0.0031 | Improved |
| ours vs DVP on out-of-domain warping error | DVP · RAVDESS out-of-domain editing | Warping error: 0.0038 -> 0.0076 | Hurt |
| ours vs DVP on out-of-domain LPIPS to direct editing | DVP · RAVDESS out-of-domain editing | LPIPS to direct editing: 0.1972 -> 0.1275 | Improved |
| ours vs Latent Transformer on in-domain warping error | Latent Transformer · RAVDESS in-domain editing | Warping error: 0.0065 -> 0.0031 | Improved |
Closest papers by meaning
- Towards Smooth Video Composition (ICLR 2023 · similarity 0.89)
- Navigating the GAN Parameter Space for Semantic Image Editing (CVPR 2021 · similarity 0.89)
- Temporal Generative Adversarial Nets With Singular Value Clipping (ICCV 2017 · similarity 0.88)
- RIGID: Recurrent GAN Inversion and Editing of Real Face Videos (ICCV 2023 · similarity 0.88)
- RIGID: Recurrent GAN Inversion and Editing of Real Face Videos and Beyond (IJCV 2025 · similarity 0.88)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 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.
