视频转存预览

LamViewVideo ("Save Video") is the output terminal of the ComfyUI_Lam video pipeline: it takes a VIDEO and writes it to your output directory with a preview. It's what you put at the end of a video workflow the way you put SaveImage at the end of an image workflow - feed it the rendered video, it writes the file, and the preview shows up in the UI.

This is one of those nodes that only makes sense if the rest of the pack's video stack (load video → process frames → save) is already running for you. ComfyUI's core has been evolving its own video saving, so the first question to ask is whether you need this at all: if you're already happily saving videos with core nodes, this adds nothing. Where it helps is inside this pack's own pipeline, where the video type flows through Lam nodes and you want a save node that speaks the same dialect.

How it works

It's a thin wrapper around ComfyUI's output machinery. The filename_prefix controls where and under what name the file lands, and it supports the same formatting tokens core ComfyUI does - %date:yyyy-MM-dd%, or tokens that pull values from other nodes like %Empty Latent Image.width% - so you can build file names out of the workflow's own state. format and codec default to auto, meaning it picks sensible values from the video itself; override them only when you have a specific container or encoding requirement.

It's an output node, so it both saves and is a terminal - the graph ends here, and the saved video is the result.

The inputs and outputs

• video (VIDEO) - the video to save.
• filename_prefix - output path/name prefix, default video/ComfyUI.
• format (COMBO, default auto) - the container to write.
• codec (COMBO, default auto) - the encoding to use.
• Output: video (VIDEO) - the saved video, available for chaining if you want a preview plus a downstream step.

Installing it

From ComfyUI_Lam: Manager → "ComfyUI_Lam", or:

cd ComfyUI/custom_nodes
git clone https://github.com/yanlang0123/ComfyUI_Lam

Restart, video category. Video saving needs the pack's video dependencies (ffmpeg-adjacent libs like imageio-ffmpeg / av, which are in the requirements). Be careful with the pack's full install though - the requirements file also pins ancient versions of TensorFlow, numpy 1.23.4, and dlib, and those are aimed at the pack's face/inpainting nodes. If you only need video save, you're better off installing just the video deps than running the whole install.bat and downgrading half your environment.

Common issues

The format/codec auto defaults are your friends; the moment you hand-pick a codec the video doesn't support, you get save errors. Also, this pack's video nodes expect their own VIDEO type - a video produced by core ComfyUI's loader may not be directly compatible, so keep the video flowing through the pack's own load/save path. And on the packaging side: the pack's frontend extension can leave a leftover popup if you uninstall without deleting ComfyUI/web/extensions/lam.