Research map / Video / Video generation
OpenSubject: Leveraging Video-Derived Identity and Diversity Priors for Subject-driven Image Generation and Manipulation
Yexin Liu, Manyuan Zhang, Yueze Wang, Hongyu Li, Dian Zheng, WEIMING ZHANG, Changsheng Lu, Harry Yang
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 |
|---|---|---|---|
| fine-tune on OpenSubject plus T2I data (ablation) | OmniGen2 · OSBench | average score: 6.43 -> 7.22 | Improved |
| fine-tune on only 100k synthetic T2I data (ablation) | OmniGen2 · OSBench | average score: 6.43 -> 5.68 | Hurt |
| add OpenSubject to T2I fine-tuning, multi-subject manipulation (ablation) | OmniGen2 · OSBench multi-subject manipulation | overall: 2.98 -> 6.22 | Improved |
| add OpenSubject, multi-subject generation identity fidelity (ablation) | OmniGen2 · OSBench multi-subject generation | identity fidelity: 7.97 -> 8.65 | Improved |
| add OpenSubject, single-subject generation prompt adherence (ablation) | OmniGen2 · OSBench single-subject generation | prompt adherence: 8.50 -> 8.30 | No clear effect |
| add T2I data alone, prompt adherence (ablation) | OmniGen2 · OSBench single-subject generation | prompt adherence: 8.50 -> 8.62 | Improved |
| fine-tune on OpenSubject | OmniGen2 · OmniContext | average score: 7.18 -> 7.34 | Improved |
| fine-tune on OpenSubject, single object | OmniGen2 · OmniContext SINGLE Object | score: 7.58 -> 7.54 | No clear effect |
| fine-tune on OpenSubject | OmniGen2 · ImgEdit | overall score: 3.44 -> 3.72 | Improved |
| fine-tune on OpenSubject, Style and Action edits | OmniGen2 · ImgEdit Style/Action | score: 4.81 -> 4.66 (Style); 4.68 -> 4.45 (Action) | Hurt |
Closest papers by meaning
- ReactID: Synchronizing Realistic Actions and Identity in Personalized Video Generation (ICLR 2026 · similarity 0.88)
- PolyVivid: Vivid Multi-Subject Video Generation with Cross-Modal Interaction and Enhancement (NeurIPS 2025 · similarity 0.88)
- Multi-subject Open-set Personalization in Video Generation (CVPR 2025 · similarity 0.87)
- ARGUS: Stacked Multi-View Identity Mosaic Injection for Subject-Preserving Video Generation (NeurIPS 2026 · similarity 0.87)
- Large-Scale Text-to-Image Model with Inpainting is a Zero-Shot Subject-Driven Image Generator (CVPR 2025 · similarity 0.87)
- VideoMage: Multi-Subject and Motion Customization of Text-to-Video Diffusion Models (CVPR 2025 · 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.
