output image 输出图片

Every workflow ends somewhere, and in the 2lab world it ends here. OutputImage (2lab) takes your IMAGE, writes it to disk as a PNG, and hands the result back to ComfyUI's UI so you can preview it. It's the pack's answer to the core Save Image node - with a default filename prefix of 2lab/img and a design intent pointed at the 2lab publishing platform.

If you've used core ComfyUI's Save Image, you already know the drill: images in, PNG out, and a filename_prefix that controls where in the output directory things land. The differences are subtle but real. The default prefix is 2lab/img, so renders pile up in ComfyUI/output/2lab/img/ unless you change it. And this is the node the pack expects you to use as the terminal of a workflow you're going to publish - when the published app runs, this is what captures and returns the render to the user.

The inputs that matter:

• images (IMAGE) - what you're saving. The output of your VAE decode.
• filename_prefix - where it goes, default 2lab/img. Set it to something meaningful per workflow (portraits/golden_hour) or you'll be spelunking through a folder called img.

The schema shows no declared outputs - it's a terminal output node that returns its results to the UI. The saved files land in ComfyUI/output/ like any other save node. One nice detail from the source: this node defaults to not embedding workflow metadata in the PNG unless you opt in, which is intentional - the 2lab platform runs with metadata disabled so published renders don't leak your entire graph. That's actually a thoughtful privacy choice worth respecting when you share outputs.

Why use this over core Save Image? If you're building toward a 2lab publish, you should, because the platform recognizes this as the output point. If you're just saving locally, core Save Image does the same thing with fewer moving parts. Be honest about which camp you're in.

Installing it

Ships in AI2lab/comfyUI-tool-2lab. ComfyUI Manager → search "comfyUI-tool-2lab" → install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/AI2lab/comfyUI-tool-2lab

Restart ComfyUI. No heavy dependencies and no model downloads - qrcode, Pillow, numpy, torch only.

Common issues

The usual save-node complaints apply. Forgot the prefix and everything's in output/2lab/img/ - fix it by setting filename_prefix per workflow. Wondering where a render went? It's under ComfyUI/output/, following whatever subfolder your prefix implies. If you publish a workflow and the app doesn't show the result, this node being present is basically required - the platform keys on it as the output. And if you expected your PNG to carry the workflow JSON like core ComfyUI's, remember this one defaults to stripping metadata; re-enable embedding only if you want the workflow-in-image culture to apply to your published outputs.