🖼👁 Preview (first) image

ComfyUI's built-in Preview Image node previews everything you feed it - fine for a single still, painful when you've got a batch of a hundred video frames wired in and it tries to render all of them. This node is the lighter alternative: it grabs just the first image and shows you that, which is usually all you need to sanity-check that a video pipeline is producing something sane before you commit to the full render.

How it works

Feed it a batch of images and it displays only the first one - a quick "is this on track" check without the UI cost of rendering every frame. But it's built to be more flexible than that single job suggests: beyond an IMAGE input, it can also take a plain file path pointing at an image on disk, or a path pointing at a video file, in which case it extracts and shows the first frame of that video. That last mode is genuinely handy mid-pipeline - you can point this node at a video file that hasn't even been loaded into ComfyUI as frames yet and get an instant visual check.

It's marked as an output node with no data outputs of its own - it's a terminal, display-only step, same category as ComfyUI's native Preview and Save Image nodes. Its own description notes it saves the input images into your ComfyUI output directory, which is standard behavior for anything in this "preview" family - the image shown in the UI is backed by a real file, not just an in-memory render.

The inputs and outputs that matter

Both inputs are optional, which is a little unusual for a node that needs something to do anything useful:

• images - a batch of images; only the first is shown.
• path - a string path to either an image file or a video file on disk; when it's a video, the node extracts the first frame.

There's no data output - this is purely a display step, the same shape as ComfyUI's own Preview Image node.

How to install it

Through ComfyUI Manager: search Bjornulf_custom_nodes, install, restart ComfyUI. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/justUmen/Bjornulf_custom_nodes

cd in and run pip install -r requirements.txt (a venv avoids stepping on other packs' dependencies), then restart. Since this node touches video-frame extraction, it leans on the pack's opencv-python / ffmpeg-python dependencies - those install fine through the standard requirements.txt on Linux, but the pack's own author has noted historically that some of the video-related nodes were only tested on Linux, so if you're on Windows and the video-path mode specifically misbehaves while plain image previews work fine, that's a more plausible explanation than a broken install.

Common issues

If nothing shows up when you feed a batch of images, double check you're wiring images, not trying to route through path - the two inputs are independent ways of getting an image, not a fallback chain, so leaving images empty and expecting it to fall back cleanly isn't something the schema guarantees. If the video-path mode specifically doesn't extract a frame, confirm the path is correct and points at a format your installed ffmpeg/opencv build actually supports - that's a dependency issue outside ComfyUI's control, not something this node can work around.