Research map / Video / Video generation
Audio-sync Video Instance Editing with Granularity-Aware Mask Refiner
Haojie Zheng, Shuchen Weng, Jingqi Liu, Siqi Yang, Boxin Shi, Xinlong Wang
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 the granularity-aware mask refiner (ablation) | AVI-Edit · AVISet (coarse masks) | FVD: 372.44 -> 335.32 | Improved |
| add the granularity-aware mask refiner (ablation) | AVI-Edit · AvED-Bench (coarse masks) | FVD: 539.83 -> 402.74 | Improved |
| add the precision factor to the mask refiner (ablation) | AVI-Edit · AVISet (coarse masks) | FVD: 354.43 -> 335.32 | Improved |
| add the precision factor to the mask refiner (ablation) | AVI-Edit · AvED-Bench (coarse masks) | FVD: 490.92 -> 402.74 | Improved |
| use the self-feedback audio agent instead of a general audio editing model (ablation) | AVI-Edit · AVISet (coarse masks) | Sync-C: 3.83 -> 4.18 | Improved |
| use the self-feedback audio agent instead of a general audio editing model (ablation) | AVI-Edit · AvED-Bench (coarse masks) | AC (%): 21.22 -> 21.63 | Improved |
| use instant degradation schedule instead of linear for mask refinement (ablation) | GAMR · AVISet test (40 clips) | mask IoU (%): 71.87 -> 76.23 | Improved |
| use instant degradation schedule instead of constant precision factor (ablation) | GAMR · AVISet test (40 clips) | mask IoU (%): 68.01 -> 76.23 | Improved |
| AVI-Edit versus Ovi on lip sync distance | AVI-Edit · AVISet test | Sync-D: 9.12 -> 9.19 | No clear effect |
| AVI-Edit versus Ovi on inference time | AVI-Edit · Inference cost (A100) | time (s): 232.6 -> 311.3 | Hurt |
| AVI-Edit versus AvED and VACE-Foley on inference time | AVI-Edit · Inference cost (A100) | time (s): 1140.5 -> 311.3 | Improved |
Closest papers by meaning
- Audio-Sync Video Generation with Multi-Stream Temporal Control (NeurIPS 2025 · similarity 0.86)
- AVENUE: Audio-Video Editing Understanding and Evaluation (NeurIPS 2026 · similarity 0.86)
- Native Audio-Visual Alignment for Generation (NeurIPS 2026 · similarity 0.85)
- DiVAS: Video and Audio Synchronization with Dynamic Frame Rates (CVPR 2024 · similarity 0.85)
- CAV-MAE Sync: Improving Contrastive Audio-Visual Mask Autoencoders via Fine-Grained Alignment (CVPR 2025 · similarity 0.85)
- Foley-Flow: Coordinated Video-to-Audio Generation with Masked Audio-Visual Alignment and Dynamic Conditional Flows (CVPR 2025 · similarity 0.84)
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.
