Research map / Video / Video generation
MAGREF: Masked Guidance for Any-Reference Video Generation with Subject Disentanglement
Yufan Deng, Yuanyang Yin, Xun Guo, Yizhi Wang, Jacob Zhiyuan Fang, Shenghai Yuan, Yiding Yang, Angtian Wang, Bo Liu, Haibin Huang, Chongyang Ma
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 region-aware masking (vs vanilla masking) (ablation) | MAGREF · Multi-subject (small-scale ablation) | Total Score: 0.558 -> 0.587 | Improved |
| use I2V backbone instead of T2V backbone (ablation) | MAGREF · Multi-subject (small-scale ablation) | Total Score: 0.550 -> 0.558 | Improved |
| add region-aware masking to full pipeline (ablation) | MAGREF · Multi-subject (full pipeline ablation) | Total Score: 0.570 -> 0.627 | Improved |
| add cross-pair data processing strategy (ablation) | MAGREF · Multi-subject (full pipeline ablation) | Total Score: 0.574 -> 0.627 | Improved |
| add subject disentanglement (ablation) | MAGREF · Multi-subject (full pipeline ablation) | Total Score: 0.580 -> 0.627 | Improved |
| add subject disentanglement (ablation) | MAGREF · Multi-subject (full pipeline ablation) | Subj-Sim: 0.417 -> 0.496 | Improved |
| add subject disentanglement (ablation) | MAGREF · Multi-subject (full pipeline ablation) | Motion / GmeScore: Motion 0.919 -> 0.945; GmeScore 0.679 -> 0.681 | No clear effect |
| MAGREF vs best open-source baseline VACE | MAGREF · Single-ID | Total Score: 0.687 -> 0.694 | Improved |
| MAGREF vs best open-source baseline VACE | MAGREF · Multi-subject (120-pair benchmark, 60 multi-subject) | Total Score: 0.590 -> 0.627 | Improved |
Later papers that compared with this method
5 later accepted papers report MAGREF as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- LibraGen: Playing a Balance Game in Subject-Driven Video Generation (ECCV 2026 · better on 6 of 6 shared results)
- AlcheMinT: Fine-grained Temporal Control for Multi-Reference Consistent Video Generation (CVPR 2026 · better on 5 of 5 shared results)
- MoFu: Scale-Aware Modulation and Fourier Fusion for Multi-Subject Video Generation (AAAI 2026 · better on 5 of 5 shared results)
- FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation (NeurIPS 2026 · better on 2 of 2 shared results)
- Scaling Zero-Shot Reference-to-Video Generation (CVPR 2026 · better on 1 of 1 shared result)
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
- Scaling Zero-Shot Reference-to-Video Generation (CVPR 2026 · similarity 0.88)
- MAGVIT: Masked Generative Video Transformer (CVPR 2023 · similarity 0.87)
- Taming Teacher Forcing for Masked Autoregressive Video Generation (CVPR 2025 · similarity 0.86)
- ARGUS: Stacked Multi-View Identity Mosaic Injection for Subject-Preserving Video Generation (NeurIPS 2026 · similarity 0.86)
- DiTCtrl: Exploring Attention Control in Multi-Modal Diffusion Transformer for Tuning-Free Multi-Prompt Longer Video Generation (CVPR 2025 · similarity 0.85)
- Mask^2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation (CVPR 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.
