Combine Video

If you googled "Combine Video" hoping for the node that finally muxes your audio and video back together inside ComfyUI, read this before you wire anything up. This Combine Video comes from Immac's ComfyUI Core Video Nodes pack (Immac/ComfyUI-CoreVideoMocks), and the whole pack is a mock. The GitHub description is literally "A set of mock nodes for RFC purposes." It's a proposal - a sketch of what ComfyUI's native video nodes could look like - not a working video toolkit.

What the node would do, if the streams were real: take a video stream, an audio stream, and a subtitle stream and mux them into a single container file. That's the classic "remux" step you do with ffmpeg all the time, and it's the correct mental model to bring here. A video file is a container (webm, mp4, mkv) holding separate streams, each encoded with a codec. Combine Video is the node at the end of the pipeline that puts them back in the box.

The inputs, and the honesty tell

All four inputs are required:

• video_stream, audio_stream, subtitle_stream - all STRING with forceInput, meaning you can't type values in; each must be wired from a stream-producing node (Split Video, the Get*Stream nodes, or the batch nodes).
• container - a dropdown "Select the video container format" with exactly one choice: webm.

That one-option dropdown is the pack's little tell. A real muxer would offer you half a dozen containers. This one offers one because it's a mock - it doesn't actually write any container at all.

What actually happens

Feed it anything and you get back a single combined_video string that looks like "Mock Combined Video: Mock Video Stream, Mock Audio Stream, Mock Subtitle Stream in webm container". That string is the entire output. There's no file written, no codec involved, no bytes muxed. The node exists so you can see what the API shape of a core "combine" node might be and plan a workflow around it.

Install

ComfyUI Manager → search "ComfyUI Core Video Nodes", or:

cd ComfyUI/custom_nodes
git clone https://github.com/Immac/ComfyUI-CoreVideoMocks

Then restart ComfyUI. That's it - no pip packages, no models to download, no PyAV or ffmpeg. It's pure Python that leans only on ComfyUI's own folder_paths.

The takeaway

Install it if you want to see the proposed core-video API or test workflow shapes. Don't expect a working remux. If you actually need to combine real video and audio today, the community standard is VideoHelperSuite (VHS) - that's the one that does real muxing - and it's not part of this pack. The pack has roughly two stars on GitHub and no real discussion anywhere; it's an RFC for a feature that hasn't shipped in core ComfyUI, not a tool for your pipeline.