ViewComfy - Download and Save Video

Despite the name, this node never generates a frame of video. What it does is grab a video file from a URL and save it to your output directory - no decoding, no tensor conversion, no preview, just a fast download. That sounds reductive until you realize it's exactly what you want when a workflow's job is "fetch this video and make it available," which is a shockingly common need when you're building ComfyUI-powered apps or APIs rather than single generations.

The mechanism is plain HTTP, straight from the source: two inputs, url (a string pointing at the video file) and filename_prefix (default DownloadedVideo). It streams the response with a 30-second timeout into a file named like DownloadedVideo_00000.mp4 in your output folder, bumping the counter if a file already exists. It figures out the extension from the URL path first, falls back to the MIME type in the response headers, and defaults to .mp4 if neither yields anything. There are no outputs - this is an output node, so it just runs and finishes.

The two things the README is upfront about: it won't show a video preview on the canvas after downloading (the file still lands in your history tab on the left sidebar), and that limitation only matters in the interactive ComfyUI interface - via API or in a ViewComfy app, the saved file is what counts. The design intent is explicit: "designed to be fast and reliable when called via API, skipping the usual tensor conversion processes." The pack's original purpose was supporting the ViewComfy platform, where a workflow sometimes needs to fetch a video asset and persist it without burning GPU time on pixel math.

Where it's genuinely useful: any workflow-as-API scenario - an endpoint that takes a video URL as input, saves it, and hands the file off to a video-to-video or analysis pipeline. If your use case is "download a video so I can watch it," you're better off with a browser; this node is for pipelines that need the file on disk and need it fast.

Watch out for two gotchas. First, the 30-second timeout: big or slow-hosted videos will fail with a "Failed to download video from URL" error, and there's no retry logic. Second, it's a dumb downloader, not a validator - whatever the server returns gets saved, so a URL that points at an HTML page instead of a video will happily save you an HTML file with an .mp4 name. And there's no authentication support; anything behind a login will just fail. If you need yt-dlp-level smarts, this isn't it.

Installation is shared across the pack: ComfyUI Manager → search "ViewComfy-Utils" → install, or clone by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/ViewComfy/ViewComfy-Utils

Restart ComfyUI after. No extra dependencies beyond stock ComfyUI, no model downloads, MIT licensed. It's a niche tool, but when your workflow has to ingest a remote video, there are very few cleaner one-node answers.