Preview Image (Standalone)

So you found a "Preview Image (Standalone)" node and you're wondering why it exists, since ComfyUI already ships a PreviewImage. Fair question. The honest answer: it's a near-verbatim extraction of ComfyUI's core PreviewImage, and its one real reason to exist is in the name plus one extra output socket. The stock preview node is a dead end - nothing comes out the other side, it just shows you the image and stops the chain there. This standalone version does the same display job but also passes the image through, so you can drop it inside a workflow, peek at an intermediate result, and keep the tensor flowing onward. Check step one, check step two, keep going. That's the entire pitch, and it's a good one for debugging.

How it works

Under the hood it's a subclass of the pack's own SaveImage base with three lines changed. It writes the incoming IMAGE tensor to ComfyUI's temp directory (not your output/ folder - those files get cleaned up, they're not your art), tags them as type="temp", and saves with compress_level=1 - the fastest, smallest PNG setting, because nobody needs archival-quality compression on a throwaway debug frame. Each temp file gets a _temp_ prefix plus five random characters so runs don't collide.

The "standalone" bit is real, not marketing: if folder_paths (a ComfyUI-internal module) isn't importable - say you're running this outside ComfyUI as a plain Python script - the code falls back to tempfile and a local ./output directory. It's one of the few "custom" nodes that would still work if you yanked it out of ComfyUI entirely.

Inputs, outputs, and what you actually set

There's exactly one input and one output, and you don't configure anything:

• images (IMAGE) - the tensor you want to eyeball. That's the whole list.
• images (IMAGE) - pass-through, same tensor, for the next node downstream.

No prefix, no filename, no settings panel. Wire something in, run, look. It's an output node in the image category, so ComfyUI shows it in the Image/Preview-right-click area when you're adding nodes.

Installing it

ComfyUI Manager is the easy path: open Manager, search ComfyUI-PreviewImageNode, install, restart. Manual route is the same as any custom node:

cd ComfyUI/custom_nodes
git clone https://github.com/eddyhhlure1Eddy/ComfyUI-PreviewImageNode.git
cd ComfyUI-PreviewImageNode
pip install -r requirements.txt

Then restart ComfyUI. The requirements.txt is just pillow and numpy - both already pulled in by ComfyUI core - so on a normal install that pip step is a no-op. Nothing heavy, no models to download.

Gotchas

The classic failure mode: the node shows no image. It's almost always one of three things - the input isn't actually an IMAGE type, the temp directory isn't writable, or you're running a stale session. A restart clears ComfyUI's cached preview state, which fixes the last one. And because previews write to disk (temp or not), don't mistake "preview" for "no file I/O" - if you're on a network share or a tiny drive, that's still a write per frame.

Bottom line: if you just want to see your output, the built-in node is fine and you don't need this pack. If you want to inspect mid-pipeline without breaking your graph, this is the node - and since the code is a clean, short extraction, it's also a great first file to read when you're learning how ComfyUI nodes are structured.