Research map / Video / Video generation
Phased One-Step Adversarial Equilibrium for Video Diffusion Models
Jiaxiang Cheng, Bing Ma, Xuhua Ren, Hongyi Henry Jin, Kai Yu, Peng Zhang, Wenyue Li, Yuan Zhou, Tianxiang Zheng, Qinglin Lu
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 stability priming (VSD warm-up) before adversarial phase (ablation) | V-PAE UAE · VBench-I2V 1-NFE | Quality Score: 69.65 -> 79.56 (no priming vs V-PAE-I) | Improved |
| use VSD priming instead of consistency-distillation priming (ablation) | V-PAE · VBench-I2V 1-NFE | Quality Score: 75.21 -> 79.56 | Improved |
| use unified self discriminator backbone instead of frozen (ablation) | V-PAE · VBench-I2V 1-NFE | Quality Score: 77.36 -> 79.56 | Improved |
| train full discriminator backbone instead of unified (ablation) | V-PAE · Wan2.1-I2V-14B | memory: OOM | Hurt |
| use multimodal cross-attention head instead of convolution head (ablation) | V-PAE · VBench-I2V 1-NFE | Quality Score: 63.29 -> 79.56 | Improved |
| add conditional SDS loss with lambda=10 (ablation) | V-PAE · VBench-I2V 1-NFE | Quality Score: 67.33 -> 79.56 | Improved |
| raise conditional SDS lambda from 10 to 100 (ablation) | V-PAE · VBench-I2V 1-NFE | Quality Score: 79.56 -> 77.31 | Hurt |
| distill with V-PAE at 1 step vs 100-NFE teacher | V-PAE · VBench-I2V | Quality Score: 80.82 -> 79.56 (1-NFE); 80.82 -> 82.24 (4-NFE) | Mixed |
Closest papers by meaning
- OSV: One Step is Enough for High-Quality Image to Video Generation (CVPR 2025 · similarity 0.92)
- Diffusion Adversarial Post-Training for One-Step Video Generation (ICML 2025 · similarity 0.89)
- SF-V: Single Forward Video Generation Model (NeurIPS 2024 · similarity 0.89)
- From Slow Bidirectional to Fast Autoregressive Video Diffusion Models (CVPR 2025 · similarity 0.88)
- Autoregressive Adversarial Post-Training for Real-Time Interactive Video Generation (NeurIPS 2025 · similarity 0.88)
- Motion Consistency Model: Accelerating Video Diffusion with Disentangled Motion-Appearance Distillation (NeurIPS 2024 · 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.
