Skip to main content
Any ComfyUI plugin that outputs IMAGE / VIDEO / AUDIO (mesh2motion, IPAdapter, ControlNet preprocessors, 3D nodes, any future plugin) can plug into a ComfyTV pipeline through the bridge nodes.
The bridge is itself a ComfyTV stage with its own Run button. Click Run, the upstream runs once, the URL is stored as the snapshot; downstream ComfyTV stages then use that snapshot on subsequent Runs instead of re-running.

​
Into-bridges

ComfyTV/Bridge contains 5 into-bridge nodes that take a native or third-party plugin’s output and pull it into ComfyTV: Into-bridges have a Run; their output persists with the project.

​
Out-bridges

The reverse direction also exists: 5 out-bridge nodes turn a ComfyTV snapshot back into native ComfyUI types, so any plugin can consume what a ComfyTV pipeline produced: ← ComfyTV Mask reads the image’s alpha channel as a mask — handy for feeding a Cutout or painter result into native inpainting nodes.

​
Common patterns

​
Plugin output → ComfyTV downstream

​
Round trip: ComfyTV result → native plugin → back

​
Plugin output is a multi-frame IMAGE batch

For plugins that output an IMAGE batch but no VIDEO object, chain ComfyUI’s Create Video first:

​
Third-party LLM / prompt enhancer → ComfyTV prompt

Any ComfyUI node that outputs STRING (a prompt enhancer, a captioner, a custom LLM) can feed text into a ComfyTV pipeline:

​
File locations

Into-bridges write to ComfyUI’s output directory under output/ComfyTV/bridge/…: