Research map / Video / Video generation
UniVST: A Unified Framework for Training-Free Localized Video Style Transfer
Quanjian Song, Mingbao Lin, Wengyi Zhan, Shuicheng Yan, Liujuan Cao, Rongrong Ji
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 all components to the baseline (ablation) | UniVST baseline · DAVTG-WikiArt ArtFID | ArtFID: 45.676 -> 37.152 | Improved |
| add localized latent blending (ablation) | UniVST · DAVTG-WikiArt | ArtFID: 38.014 -> 37.152 | Improved |
| add AdaIN-guided latent-shift (ablation) | UniVST · DAVTG-WikiArt | ArtFID: 37.649 -> 37.152 | Improved |
| add AdaIN-guided attention-shift (ablation) | UniVST · DAVTG-WikiArt | ArtFID: 39.589 -> 37.152 | Improved |
| add sliding-window consistent smoothing (ablation) | UniVST · DAVTG-WikiArt | CLIP-F: 0.970 -> 0.980 | Improved |
| add AdaIN-guided attention-shift (ablation) | UniVST · DAVTG-Laion | ArtFID: 32.534 -> 30.636 | Improved |
| add AdaIN-guided latent-shift (ablation) | UniVST · DAVTG-Laion LPIPS | LPIPS: 0.362 -> 0.383 | Hurt |
| use first frame plus previous frame anchors instead of previous frame only (ablation) | point-matching mask propagation · DAVIS-Laion mask propagation | Avg IoU: 0.5321 -> 0.7799 | Improved |
| use n=15 anchors instead of n=9 (ablation) | point-matching mask propagation · DAVIS-Laion mask propagation | Avg Dice: 0.8726 -> 0.8729; time 2.742 -> 2.878 | No clear effect |
| use k=15 in k-NN instead of k=40 (ablation) | point-matching mask propagation · DAVIS-Laion mask propagation | Avg Dice: 0.8719 -> 0.8772 | Improved |
| use decreasing time coefficient beta instead of increasing (ablation) | AdaIN attention-shift · DAVIS-Laion | ArtFID: 32.158 -> 30.658 | Improved |
| use AdaIN attention-shift instead of K-V replacement (ablation) | UniVST · DAVIS-Laion | ArtFID: 32.561 -> 30.658 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- BIVDiff — BIVDiff: A Training-Free Framework for General-Purpose Video Synthesis via Bridging Image and Video Diffusion Models (CVPR 2024) · better on 2 of 2 shared results
- CAST — CAST: Cross-Attention in Space and Time for Video Action Recognition (NeurIPS 2023) · better on 2 of 2 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
- RegionRoute: Regional Style Transfer with Diffusion Model (CVPR 2026 · similarity 0.89)
- StyleMaster: Stylize Your Video with Artistic Generation and Translation (CVPR 2025 · similarity 0.89)
- Semantix: An Energy-guided Sampler for Semantic Style Transfer (ICLR 2025 · similarity 0.87)
- Scalable Motion Style Transfer with Constrained Diffusion Generation (AAAI 2024 · similarity 0.87)
- OmniStyle: Filtering High Quality Style Transfer Data at Scale (CVPR 2025 · similarity 0.87)
- Local Consistency Guidance: Personalized Stylization Method of Face Video (Student Abstract) (AAAI 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.
