Skip to content
OpenMicroDrama

numz /ComfyUI-SeedVR2_VideoUpscaler

SeedVR2 video and image upscaling, as ComfyUI nodes or a CLI
ComfyUI workflowApache-2.0English README
Open on GitHubSave
Stars
2.9k
30 days
+1
Last push
9 months ago
ComfyUI-SeedVR2_VideoUpscaler screenshot from its README
From the ComfyUI-SeedVR2_VideoUpscaler README on GitHub.

About ComfyUI-SeedVR2_VideoUpscaler

The official ComfyUI release of ByteDance's SeedVR2 upscales video and images. Run it as ComfyUI nodes, or headless as a multi-GPU command-line tool. Long clips stream in chunks, and FFmpeg output can be 10-bit x265. Version 2.5.0 broke old workflows, so plan to rebuild them.

Best for: Editors finishing AI video who want a final upscale and restore step, in ComfyUI or on a multi-GPU CLI.

What it does

  • Upscales video and images with the official SeedVR2 restoration model
  • Ships 3B and 7B model configs and example ComfyUI workflows
  • Runs as ComfyUI nodes or as a multi-GPU standalone CLI
  • Processes long videos in chunks with --chunk_size, so RAM stays bounded
  • Writes 10-bit x265 video through FFmpeg with --10bit, cutting gradient banding
  • Works on CUDA, AMD ROCm and Apple Silicon MPS, with attention backend fallbacks

Quickstart

Rewritten from the README. Check the repo for the latest steps.

  1. 1
    Install the node through ComfyUI Manager.
  2. 2
    Restart ComfyUI, then load a workflow from example_workflows/.
  3. 3
    For headless runs, use the standalone CLI inference_cli.py, which supports multiple GPUs.
  4. 4
    Add flags like --video_backend ffmpeg, --10bit, --chunk_size, --cache_dit or --cache_vae.

Models and languages

Models it works with

  • SeedVR2 3B
  • SeedVR2 7B

Interface and docs

  • English

Head-to-head

Alternatives

Other comfyui workflow projects people compare with ComfyUI-SeedVR2_VideoUpscaler.

Lightricks

ComfyUI-LTXVideo

Gives ComfyUI extra nodes and example workflows for the LTX-2 video model

ComfyUI workflow
  • LTX-2.3
  • LTX-2.5
  • LTX-2.0
  • +1
4.2k GitHub stars+4/30dyesterdayCustom license

Comfy-Org

workflow_templates

Browse official ComfyUI workflow templates and subgraph blueprints

ComfyUI workflow
  • Wan 2.1
  • LTXV
1.2k GitHub stars+4/30dtodayMIT

kijai

ComfyUI-WanVideoWrapper

Puts WanVideo and related video models inside ComfyUI

ComfyUI workflow
  • Wan 2.1
  • Wan 2.2
  • SkyReels V2
  • +7
6.7k GitHub stars+2/30d4 months agoApache-2.0

Songssx

ComfyUI-MiniMaxH3-TimelineDirector

ComfyUI nodes for editing a reference-media timeline in MiniMax H3

ComfyUI workflow
  • MiniMax H3
547 GitHub stars+6/30d10 days agoGPL-3.0

Featured on OpenMicroDrama

Do you maintain ComfyUI-SeedVR2_VideoUpscaler? Add this badge to your README so English speakers can find our write-up.

Featured on OpenMicroDrama
markdown
[![Featured on OpenMicroDrama](https://openmicrodrama.com/badges/featured.svg)](https://openmicrodrama.com/projects/numz-comfyui-seedvr2-videoupscaler)

The badge links to this page. Want your description changed? Email support@openmicrodrama.com.

Reviewed Oct 1, 2026. Last push 2025-12-24, outside our 90-day window; kept because it's still the go-to SeedVR2 upscaler for ComfyUI. We write these descriptions ourselves. The repo's own docs are the final word.