ausboss/FlashVSR-v1.1-ComfyUI

FlashVSR v1.1 for ComfyUI, in one file

This is the official FlashVSR v1.1 model, repacked into one file for the FlashVSR Upscale 🆎 node in my ComfyUI pack, ComfyUI-AusBoss. It makes a video larger and sharper.

I did not train this model. FlashVSR is the work of Junhao Zhuang and the OpenImagingLab team. The original files are at JunhaoZhuang/FlashVSR-v1.1 and the code at OpenImagingLab/FlashVSR.

How to use it

1. Download flashvsr_v1.1_fp16.safetensors and put it in ComfyUI/models/diffusion_models.
2. Install ComfyUI-AusBoss from ComfyUI-Manager (search for AusBoss).
3. Add FlashVSR Upscale 🆎, connect your video frames, pick the file under Model, and set Scale from 1 to 4.

No other pack and no other model file is needed. Connecting the Wan 2.1 VAE to the node's vae input is optional.

Files

File	Size	What it is
flashvsr_v1.1_fp16.safetensors	3.5 GB	Everything the node needs
LICENSE		The Apache-2.0 license of the original model

sha256 of the model file: dd833652205e2efdb6d524a669908156b13620e77b8b3f2aedfcc3086eae1569

What I changed

Nothing was retrained or tuned. The file holds four parts of the official release, stored together:

In this file	Official file
the video model (keys as they are)	diffusion_pytorch_model_streaming_dmd.safetensors
lq_proj_in.*	LQ_proj_in.ckpt
tc_decoder.*	TCDecoder.ckpt
positive_conditioning	posi_prompt.pth, from the code repository

Every value is stored as 16-bit floats (fp16), which is why the file is smaller than the official ones together. The two .ckpt files and the .pth file became safetensors, so nothing in this file can run code when it loads.

This file works with my node only. For the official code, use the official files.

What I measured

On an RTX 5090, one 97-frame clip made 2x larger:

Result size	Speed	Graphics memory
576 x 1024	about 7 frames a second	about 10 GB
1152 x 2048, in two tiles	about 2 frames a second	about 16 GB

A long clip needs no more graphics memory than a short one.

License

Apache-2.0, the same as the original model. The LICENSE file is in this repository.

@article{zhuang2025flashvsr,
  title={FlashVSR: Towards Real-Time Diffusion-Based Streaming Video Super-Resolution},
  author={Zhuang, Junhao and Guo, Shi and Cai, Xin and Li, Xiaohui and Liu, Yihao and Yuan, Chun and Xue, Tianfan},
  journal={arXiv preprint arXiv:2510.12747},
  year={2025}
}