Research map / Video / Video generation
MagDiff: Multi-Alignment Diffusion for High-Fidelity Video Generation and Editing
Haoyu Zhao, Tianyi Lu, Jiaxi Gu, Xing Zhang, Qingping Zheng, Zuxuan Wu, Hang Xu, Yu-Gang Jiang
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-driven alignment (subject mask) to HFA+APA (ablation) | MagDiff · UCF-101 | FVD: 444.67 -> 339.62 | Improved |
| add adaptive prompts alignment to HFA+SDA (ablation) | MagDiff · UCF-101 | FVD: 367.23 -> 339.62 | Improved |
| add high-fidelity alignment to APA+SDA (ablation) | MagDiff · UCF-101 | FVD: 388.41 -> 339.62 | Improved |
| add all three alignments to MSR-VTT FVD (ablation) | MagDiff · MSR-VTT | FVD: 372 -> 245 | Improved |
| use learnable adaptive weights (APA) instead of fixed equal-weight prompts alignment (ablation) | MagDiff · UCF-101 | FVD: 388.41 -> 339.62 | Improved |
| use trainable alphas instead of fixed values (ablation) | APA · UCF-101 | FVD: 364.89 -> 339.62 | Improved |
| use subject-segmented image instead of full image as condition | MagDiff · UCF-101 | FVD: 415.87 -> 339.62 | Improved |
| tuning-free MagDiff vs fine-tuning editing baselines | MagDiff · DAVIS (editing) | Textual alignment: 28.33 (Tune-A-Video) vs 27.65 | Mixed |
| MagDiff vs I2VGen-XL on MSR-VTT frame consistency | MagDiff · MSR-VTT | Frame consistency: 89.8 -> 88.4 | Hurt |
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.
- Make-A-Video — Make-A-Video: Text-to-Video Generation without Text-Video Data (ICLR 2023) · 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
- Inference-Time Text-to-Video Alignment with Diffusion Latent Beam Search (NeurIPS 2025 · similarity 0.89)
- VideoDPO: Omni-Preference Alignment for Video Diffusion Generation (CVPR 2025 · similarity 0.88)
- TokenFlow: Consistent Diffusion Features for Consistent Video Editing (ICLR 2024 · similarity 0.88)
- COVE: Unleashing the Diffusion Feature Correspondence for Consistent Video Editing (NeurIPS 2024 · similarity 0.87)
- TIV-Diffusion: Towards Object-Centric Movement for Text-driven Image to Video Generation (AAAI 2025 · similarity 0.87)
- Align-A-Video: Deterministic Reward Tuning of Image Diffusion Models for Consistent Video Editing (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.
