📺 Resolution Selector (API Fallback)

The 📺 Resolution Selector is the simplest kind of ComfyUI node there is: a dropdown that outputs the value you picked. It's one of the little "controller" nodes in the ComfyUI-Gemini3-API-Fallback pack, meant to sit at the edge of your workflow and drive the resolution input of one or many 🤖 Gemini 3 Pro Image nodes.

Why bother? Same reason anyone uses a shared controller instead of per-node settings: you set 4K in one place, wire it to every image node, and later downgrade the whole batch to 1K for a quick test by flipping a single widget. With a dozen Gemini nodes in a grid, that's the difference between a two-second change and a scavenger hunt.

How it works

The node offers the three resolutions the Gemini image API actually accepts - 1K, 2K, 4K - and passes your choice straight through as a string on its resolution output. That's it. No logic, no validation, no network call. The heavy lifting happens downstream in the Gemini 3 Pro Image node, which sends your pick as the image_size in the generation request.

The resolution you choose is a real cost lever. Nano Banana Pro (Google's Gemini 3 Pro Image model, which these nodes call) generates native 4K output, but you pay per image, and higher resolution costs more - roughly $0.04 on the cheap end up to $0.24 a pop at the top tier. If you're iterating on composition, run 1K; only pay for 4K when you've locked the idea and want the final. A 4K-then-local-upscale pipeline is usually wasted money with a model that already outputs 4K natively.

The fields that matter

• resolution - the only input: 1K (default), 2K, or 4K.

Output: resolution, a single value that plugs into the resolution input of any Gemini 3 Pro Image node.

Installing it

No model downloads, no GPU - this is a cloud-API pack. Via ComfyUI Manager (search "ComfyUI-Gemini3-API-Fallback"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/Latentnaut/ComfyUI-Gemini3-API-Fallback
python -m pip install google-genai pillow numpy requests

then restart ComfyUI. The selector itself needs no API key; only the image node it feeds does.

Honestly, a lone Resolution Selector in a one-node workflow is redundant - just set the dropdown inside the image node. It pays for itself the moment you're controlling several generation nodes, or when you want resolution to be a workflow parameter you can edit without opening node internals.