Research map / Video / Video generation
SMRABooth: Subject and Motion Representation Alignment for Customized Video Generation
Xuancheng Xu, Yaning Li, Sisi You, Bing-Kun Bao
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 subject representation alignment (DINOv2 target) (ablation) | SMRABooth · Customized video generation | DINO-I: 0.467 -> 0.519 | Improved |
| add subject representation alignment (ablation) | SMRABooth · Customized video generation | CLIP-I: 0.667 -> 0.700 | Improved |
| add motion representation alignment (optical flow) (ablation) | SMRABooth · Customized video generation | Motion Fidelity: 60.02 -> 62.89 | Improved |
| add motion representation alignment (ablation) | SMRABooth · Customized video generation | Imaging Quality: 61.29 -> 67.46 | Improved |
| use sparse LoRA layers instead of full layers for both LoRAs (ablation) | SMRABooth · Customized video generation | CLIP-I: 0.652 -> 0.700 | Improved |
| add FFN.0 to subject LoRA (ablation) | SMRABooth · Customized video generation | DINO-I: 0.357 -> 0.519 | Improved |
| add FFN.0 to motion LoRA (ablation) | SMRABooth · Customized video generation | Motion Fidelity: 56.12 -> 62.89 | Improved |
| use time-split subject LoRA weights (low before Tpoint=15, doubled after) (ablation) | SMRABooth · Customized video generation | video quality | Improved |
| compare SMRABooth to DualReal on WAN2.1 | WAN2.1 customization · Customized video generation | Motion Fidelity: 45.75 -> 62.89 | Improved |
| add LoRAs only (baseline) vs plain WAN2.1 | WAN2.1 · Customized video generation | Subject Consistency / Imaging Quality: DINO-I 0.165 -> 0.464; Imaging Quality 65.12 -> 55.90 | Mixed |
| add full SMRABooth vs ablated variants on subject consistency (ablation) | SMRABooth · Customized video generation | Subject Consistency: 95.31 vs 95.29 (w/o SuRA) | No clear effect |
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.
- DualReal — DualReal: Adaptive Joint Training for Lossless Identity-Motion Fusion in Video Customization (ICCV 2025) · better on 5 of 5 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
- MotionBooth: Motion-Aware Customized Text-to-Video Generation (NeurIPS 2024 · similarity 0.87)
- VideoMage: Multi-Subject and Motion Customization of Text-to-Video Diffusion Models (CVPR 2025 · similarity 0.87)
- Spectral Motion Alignment for Video Motion Transfer Using Diffusion Models (AAAI 2025 · similarity 0.86)
- DisMo: Disentangled Motion Representations for Open-World Motion Transfer (NeurIPS 2025 · similarity 0.86)
- Synergizing Motion and Appearance: Multi-Scale Compensatory Codebooks for Talking Head Video Generation (CVPR 2025 · similarity 0.85)
- CustomCrafter: Customized Video Generation with Preserving Motion and Concept Composition Abilities (AAAI 2025 · 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.
