🤖 Leon Google Image API

Leon Google Image API is the pack's Imagen node, and it's also the simplest one in the whole pack - which is kind of its charm. No optional parameters at all. You get a prompt, a model dropdown, an aspect ratio, an output format, a seed, the API URL and key, and a response format. Fill those in and out comes a Google Imagen image as a ComfyUI tensor. If you want to try Imagen 4 without a local model and without learning yet another API SDK, this is the shortest on-ramp there is.

It's also the node the pack's own example workflow is built around - a bare Leon_Google_Image_API_Node → PreviewImage + ShowText graph - so it's the pack author's reference "hello world". That tells you something about how the rest of the pack is intended to behave: same three outputs (image, image_url, seed), same retry-on-failure base class, same model-family abstraction.

How it works

The node builds a JSON payload and POSTs it to https://api.hyprlab.io/v1/images/generations with a Bearer token - HyprLab is the reseller here, not Google directly. The response is downloaded and converted to an RGBA tensor, so it drops straight into PreviewImage, SaveImage, or anything downstream.

The inputs that matter:

• model - imagen-4-ultra (default), imagen-4, imagen-4-fast, imagen-3, imagen-3-fast. Imagen 4 is the current line; the ultra/fast split is the usual quality-versus-speed trade. The Imagen 3 options are there if you want the cheaper, older behavior.
• prompt - max 10,000 characters, which is generous. Imagen is a strong natural-language follower; descriptive sentences beat tag soup here.
• aspect_ratio - 1:1, 3:4, 4:3, 9:16, 16:9. No match_input_image and no ultrawide options - this is a deliberately small menu.
• response_format - url (default) or b64_json. With url, the node downloads the image itself and image_url comes back as a hosted URL; with b64_json you get a data URI in image_url instead.

That's the whole surface. There's no image-input socket, no negative prompt, no steps - if you need editing or multi-image guidance, Imagen's relatives in this pack (the Nano Banana nodes) are where that lives.

Installing it

The standard pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/l3ony2k/comfyui-leon-nodes
pip install -r requirements.txt

Restart ComfyUI, or install "ComfyUI Leon Nodes" from ComfyUI Manager. Dependencies are light and there are no models to download.

Where people get burned

The recurring theme: the YOUR_API_KEY_HERE default is a placeholder, and because failures retry with backoff, a wrong key means a slow death rather than a fast one - watch the console. Imagen is also a heavily filtered model (it's Google, so expect content moderation baked in), and the "no optional parameters" simplicity means you can't tune your way out of a censorship refusal - you just get a different image or an error. And keep expectations honest: this routes through HyprLab, a small paid aggregator with barely any public community footprint, so troubleshooting means provider docs, not a Reddit thread. For a clean single-image Imagen generation, this is exactly as easy as it looks.