PreViewVideo
The no-frills output node inside MimicMotion
- video
This one's exactly what it sounds like, and not much more. Plug a VIDEO into it and it renders that video right in the node so you can watch your MimicMotion output without hunting through your output folder. That's the whole job. It's the terminal node at the end of a MimicMotion graph - the thing you look at, not something you keep wiring further downstream.
Don't go looking for a real description in the node itself, by the way. The node_description field the author shipped it with literally says "hello world!" - a placeholder that never got replaced. That tells you something about this pack: it's a research-repo-to-ComfyUI port done fast, not a polished product with documentation to match.
How it works
There's no processing here. It takes whatever VIDEO object lands on its single input and displays it inline, the same instinct as ComfyUI's built-in Preview Image but for video instead of stills. It doesn't decode, re-encode, or touch the file - it just gives you eyes on the result of whatever produced that video, which in this pack is almost always MimicMotionNode.
The inputs and outputs that matter
There's exactly one thing to know:
video(required,VIDEO) - the clip to preview. Wire it straight from MimicMotionNode's output.
No optional inputs, no outputs. It's flagged as an output node, meaning ComfyUI treats it as a legitimate endpoint for the graph - you're allowed to end a workflow here. If you actually want to keep the result as a file rather than just eyeball it, you still need a save node somewhere; this one doesn't write anything to disk on its own.
How to install it
You don't install this node by itself - it ships as part of the whole ComfyUI-MimicMotion pack. Search "ComfyUI-MimicMotion" in ComfyUI Manager and install, or do it by hand:
cd ComfyUI/custom_nodes
git clone https://github.com/AIFSH/ComfyUI-MimicMotion.git
cd ComfyUI-MimicMotion
pip install -r requirements.txt
then restart ComfyUI. Make sure ffmpeg is actually reachable from your command line first (apt install ffmpeg on Linux; WingetUI works on Windows) - the pack's video handling leans on it, and this preview node is downstream of that same plumbing. The README also calls out pinning xformers to match your exact torch build (their example: torch==2.1.0+cu121 → xformers==0.0.22.post7), which matters for the pack to load at all, not just for this node.
Common issues & troubleshooting
Worth knowing before you install: this whole pack is genuinely quick-and-dirty, and if you only want a "play a video inline" node, there are lighter packs that do just that without dragging in an SVD-based motion pipeline and its dependencies. Install this one because you're using MimicMotionNode, not because you specifically want this preview widget.
The node doesn't show up at all. That almost always means the pack failed to load, not that this node individually is broken - check the ComfyUI console at startup for an import error. The most common cause is exactly the dependency conflict the ecosystem is known for: custom nodes share one Python environment with no isolation, so a mismatched xformers/torch pairing here can quietly break the whole pack's import, taking this node down with it. This repo hasn't been touched since August 2024 and is sitting on 15 open GitHub issues, so if you hit something odd, don't expect a quick upstream fix - check the issues list before assuming it's your setup.
Nothing plays, or you get a blank player. Confirm ffmpeg is actually on your PATH - that's the one dependency the README calls out explicitly, and video display breaking silently is the usual symptom when it's missing.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| video | VIDEO | — |
Outputs (0)
No outputs