🖌️ Save Image

The pack's description is the whole story, so let's not bury it: "Saves the input images to your ComfyUI output directory." That's it. This node is a SaveImage with one extra knob - a format dropdown - and if you've used core ComfyUI's SaveImage you already know 95% of how it works.

So why does it exist? Because the pack is a grab-bag of "nodes modified to achieve things I felt were missing," and the author evidently felt that core's save node needed a file-format choice without learning a whole new pack's convention. Fair enough. It's the unglamorous end of your workflow, and sometimes unglamorous is exactly what you want.

How it works

Feed it images, it writes them to ComfyUI/output (the standard output directory, so they show up in the browser's history panel like any other save). The three inputs:

• images - what you're saving.
• filename_prefix - the filename stem, default ComfyUI. Per the author's own tooltip, this supports formatting like %date:yyyy-MM-dd% or %Empty Latent Image.width%, so you can embed timestamps or pull values from other nodes into the filename. That's a genuinely handy feature and it's the same syntax core ComfyUI uses.
• file_format - PNG, JPG, or WebP.

One behavior worth knowing, straight from the code: if your image has an alpha channel and you ask for JPG, it quietly saves a PNG instead. JPG can't carry transparency, so the node falls back rather than erroring. You'll just get a .png where you asked for .jpg, which is the sane failure mode.

Why reach for it

Honestly? You'd reach for this node because you're already in the pack, or because you want a save node whose whole UI is three fields and a dropdown. PNGs get workflow metadata embedded (that's how the "workflow included" sharing culture works in ComfyUI), JPGs and WebP are smaller - if you're batch-generating for the web, WebP at quality 100 is a nicer tradeoff than PNG for photographs. For a node this thin, the format choice is basically the product.

Install

ComfyUI Manager → search ComfyUI-NeuralMedia, or:

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

Restart ComfyUI. It's YarvixPA's pack - the same person who publishes the FLUX.1-Fill-dev-GGUF quants on HuggingFace - and if a manual clone hits a 404 (the repo has moved around), install via Manager, which carries its own registry entry.

The honest take

This is not a node you should go out of your way to install. It's a convenience wrapper around functionality ComfyUI already ships, and the only real addition is the format dropdown. If you're building a shared workflow and want minimal dependency footprint, core SaveImage does the job. But if you're already running this pack and you want a save node with the format dialed in front rather than hidden in a script, this is a perfectly serviceable end to a graph - no more, no less.