GPT Image2 Config

Name the problem: you've got three GPT Image 2 generation nodes in a workflow, and every one of them has its own size, background, and quality dropdowns sitting somewhere in its panel. Change the output size and you get to hunt down all three. GPT Image2 Config is a configuration hub that does what the name says - it centralizes the three output settings a GPT Image workflow actually cares about, so you set them once and every OpenAI node reads from the same source.

The reassuring part for beginners: this node doesn't call any API and needs no key. It's pure plumbing. It takes your chosen settings and hands them back out on typed sockets. The actual API call - and the credit card, and the key handling - lives in the GPT Image nodes you wire it into, which is the normal shape of the API-wrapper category the KB covers: the cloud model stays closed, and these nodes are just the levers in front of it.

How it works

Mechanically it's a V3 schema node: three combo inputs in, the same three values out. The only real logic is the constrained option lists, which are worth knowing because they're the GPT Image spec:

• size - auto, 1024x1024, 1024x1536, 1536x1024, 2048x2048, 2048x1152, 1152x2048, 3840x2160, 2160x3840. Default auto.
• background - auto or opaque. opaque forces a solid background instead of letting the model decide (which can produce transparency).
• quality - low, medium, high. Default low, which is the sensible starting point for drafts; bump it for final renders and eat the longer generation time.

In the pack's app-build mode, the three widgets collapse into a single combined panel input so a packaged workflow shows "output settings" as one control instead of three loose dropdowns.

The inputs and outputs that matter

All three inputs (size, background, quality) are combos, and all three outputs are the same types. The outputs wire into the matching input sockets of any GPT Image 2 node that exposes them - so the settings you set here are the settings every consumer runs with.

How to install it

Part of the DAELab pack. ComfyUI Manager, search "ComfyUI-DAELab-Custom-Nodes-Library", or:

cd ComfyUI/custom_nodes
git clone https://github.com/TeutonicMepheles/ComfyUI-DAELab-Custom-Nodes-Library

then restart. No models, no key, no requirements file - but the nodes it feeds need their own OpenAI setup (API key, and whichever custom-node pack provides the actual GPT Image generation node). It's a V3 node, so No module named 'comfy_api' in the startup log means update ComfyUI Core, not pip-install a package.

Common issues & troubleshooting

"I changed the size here and nothing changed." Check that the downstream GPT Image node is actually wired to this node's outputs, and not still reading a built-in widget. Unplugged consumers keep their own defaults.

"The socket types don't match." The outputs are combo types meant for the DAELab pack's own GPT Image nodes. If you're driving a different vendor's node, it may want plain string or its own combo type - match the types and it'll wire.

"It says my size is invalid at the API." The option list is a valid set, but a specific model version may reject some sizes (or require a paired background). Fall back to auto and let the API pick, then check the error message the generation node surfaces.