Searching…
Built on rigorous research from Tsinghua University and Shanghai AI Lab, FlashVSR combines multiple innovations to make diffusion-based video super-resolution practical and efficient.
FlashVSR runs at ∼17 FPS for 768 × 1408 videos on a single A100 GPU by combining three complementary innovations: (i) a train-friendly three-stage distillation pipeline that enables streaming super-resolution, (ii) lo...
我们开源了 FlashVSR，首个能支持流式实时处理的基于生成式超分技术，支持无限长视频。 我们利用了稀疏注意力、 Casual Attention 和 单步推理技术，在扩散式视频框架，在单张 A100 上可以对768×1408分辨率视频达到约 17 FPS处理速度，比其他生成式视频超分快约12倍。
📰 News Nov 2025 — 🎉 FlashVSR v1.1 released: enhanced stability + fidelity Oct 2025 — FlashVSR v1 (initial release): Inference code and model weights are available now! 🎉 Bug Fix (October 21, 2025): Fixed local_attenti...
FlashVSR Video Upscaler converts low-resolution or AI-generated footage into cinematic 4K at ~17 FPS with a diffusion-based streaming pipeline. Deliver broadcast-ready masters, restore archives, and enhance AIGC clips...
FlashVSR 是一个面向实时扩散模型的流式视频超分辨率框架，旨在通过实现高效性、可扩展性和实时性能，使基于扩散模型的视频超分辨率技术变得实用化。
TL;DR — FlashVSR is a streaming, one-step diffusion-based video super-resolution framework with block-sparse attention and a Tiny Conditional Decoder. It reaches ~17 FPS at 768×1408 on a single A100 GPU.
在这个视频里，我会详细讲解如何安装FlashVSR需要的块稀疏注意力（Block-Sparse-Attention）依赖库，利用ComfyUI_FlashVSR（smthemex）体验FlashVSR模型的全功能，然后尝试简单的WanVideoWrapper（Kijai）放大流程，最后对比一下同赛道的SeedVR2。
Real-Time Diffusion-Based Streaming Video Super-Resolution / 基于Diffusion架构的实时视频流超分模型 - ComfyUI-FlashVSR_Ultra_Fast/README_zh.md at main · lihaoyun6/ComfyUI-FlashVSR_Ultra_Fast
FlashVSR 是一个基于扩散模型（Diffusion Models）的流式视频超分辨率框架，能将低分辨率视频实时转换为高清视频。 项目地址： Towards Real-Time Diffusion-Based Streaming Video Super-Resolution
