Research map / Video / Video generation
FlashVideo: Flowing Fidelity to Detail for Efficient High-Resolution Video Generation
Shilong Zhang, Wenbo Li, Shoufa Chen, Chongjian GE, Peize Sun, Yifu Zhang, Yi Jiang, Zehuan Yuan, Bingyue Peng, Ping Luo
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 |
|---|---|---|---|
| use LoRA instead of full fine-tuning for the 270p Stage I (ablation) | CogVideoX-5B Stage I · VBench semantic | semantic score: 78.54 -> 82.03 | Improved |
| lower Stage I resolution from 480p to 270p (ablation) | CogVideoX-5B · VBench semantic | semantic score: 77.04 -> 82.03 | Improved |
| add latent degradation on top of pixel degradation in Stage II training (ablation) | FlashVideo Stage II · 5-frame 1080p | MUSIQ: 49.12 -> 55.45 | Improved |
| add pixel degradation Stage II over raw Stage I output (ablation) | FlashVideo Stage II · 5-frame 1080p | MUSIQ: 23.61 -> 49.12 | Improved |
| narrow latent degradation training range from 600-900 to 650-750 (ablation) | FlashVideo Stage II · Texture100 | MUSIQ: 53.98 -> 58.49 (inference 650) | Improved |
| use latent noise steps below 400-500 (weak degradation) (ablation) | FlashVideo Stage II · artifact correction | artifact removal | Hurt |
| increase Stage II NFE from 1 to 4 (ablation) | FlashVideo Stage II · Texture100 | MUSIQ: 48.60 -> 58.69 | Improved |
| increase Stage II NFE beyond 4 (ablation) | FlashVideo Stage II · Texture100 | MUSIQ: 58.69 -> 59.64 (8 NFE) | No clear effect |
| use 3D RoPE and extrapolate from 1080p to 2K (ablation) | FlashVideo Stage II · Texture100 | MUSIQ: 58.69 -> 62.40 | Improved |
| apply Stage II flow-matching enhancement to Stage I output (ablation) | FlashVideo · VBench-Long | image quality: 61.87 -> 66.96 | Improved |
Later papers that compared with this method
1 later accepted paper reports FlashVideo as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- SimpleGVR: A Simple Baseline for Latent-Cascaded Generative Video Super-Resolution (ICLR 2026 · better on 4 of 4 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- CogVideoX-5B — CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer (ICLR 2025) · better on 3 of 3 shared results
- Upscale-A-Video — Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution (CVPR 2024) · better on 2 of 2 shared results
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
- HiFlow: Training-free High-Resolution Image Generation with Flow-Aligned Guidance (NeurIPS 2025 · similarity 0.87)
- Encapsulated Composition of Text-to-Image and Text-to-Video Models for High-Quality Video Synthesis (CVPR 2025 · similarity 0.87)
- Ultra Flash: Scaling Real-Time Streaming Video Generation to High Resolutions (NeurIPS 2026 · similarity 0.87)
- Snap Video: Scaled Spatiotemporal Transformers for Text-to-Video Synthesis (CVPR 2024 · similarity 0.87)
- StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text (CVPR 2025 · similarity 0.86)
- LinGen: Towards High-Resolution Minute-Length Text-to-Video Generation with Linear Computational Complexity (CVPR 2025 · similarity 0.86)
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.
