RM Video Combine

If you've built a video workflow in ComfyUI, you've met VideoHelperSuite's VideoCombine - the node that turns a batch of frames into an actual file. This is that node, ported into the RMAutomation pack (the README credits VHS directly), with the same formats, the same VHS_FILENAMES output, and the same ffmpeg requirement. The selling point isn't novelty; it's that you get video output without installing a second pack, and it slots into the RMAutomation style workflow with its own metadata and progress plumbing.

How it works

You feed it an image batch (or a latent plus a VAE, which it decodes frame by frame in batches to keep VRAM sane). It picks a format, and then:

• image/gif and image/webp are written straight through PIL - no ffmpeg needed, and WebP saves lossless by default.
• video/webm-vp9, video/h264-mp4, and video/h265-mp4 go through ffmpeg, driven by JSON format files in the pack's video_formats/ folder that define the codec pass, pixel format, and dimension alignment. You can drop your own JSON in there to add formats.

Metadata is the differentiator: the full prompt and extra PNG info are written into the file's metadata via ffmpeg, and a first-frame PNG with the same metadata is saved alongside so a drag-into-ComfyUI workflow round-trip still works. An optional audio input gets muxed in with an audio pass, and pingpong doubles the frames back on themselves for a loop. It also pings a local "Playground" viewer app at 127.0.0.1:3001 for progress - harmless if nothing's listening; it fails silently.

Inputs and outputs that matter

• images (IMAGE, required) - also accepts LATENT when vae is connected.
• frame_rate (default 8), loop_count (0 = no loop), format, pingpong, save_output.
• format is the one you'll touch most: H.264 MP4 for anything you'll upload, GIF/WebP for quick previews.
• filename_prefix (default "video") - files land in your ComfyUI output folder as prefix_timestamp.mp4, or in output_dir if you set one.
• Output: Filenames (VHS_FILENAMES) - compatible with VHS's preview/info nodes if you have them.

How to install it

cd ComfyUI/custom_nodes
git clone https://github.com/Moser9815/ComfyUI-RMAutomation

Restart ComfyUI. The video formats need an ffmpeg binary - the pack has no requirements.txt, so install one manually:

pip install imageio-ffmpeg

or put a system ffmpeg on your PATH (the node prefers the imageio one if imageio-ffmpeg is importable).

Common issues

The classic: "ffmpeg is required for video outputs and could not be found." That's the node telling you the binary is missing - install imageio-ffmpeg and restart; GIF and WebP keep working without it. Second: dimension errors. The format files enforce dimension alignment, and the node pads odd-sized frames automatically, but a frame count that doesn't match expectations can produce a black or clipped tail - keep every batch the same size. And if you only get the first-frame PNG and no video, check the ffmpeg stderr - the node prints the subprocess error, and it's almost always a codec your build doesn't include (H.265 especially).