Reve Text-to-Image (fal)

Reve is one of the closed text-to-image models the local community spent 2025 wishing it could run - it turns up in the canonical "makes me really hope open source can catch up" thread right alongside Ideogram and OpenAI. It's known for strong prompt adherence and clean typography. This node hands you Reve through fal.ai without any of that catching-up: you call it from a ComfyUI graph and pay per image.

It's a hosted call, so there's no checkpoint on your disk and nothing runs on your GPU. Reve's weights are closed, so this is the only way to touch it locally-ish - through the API.

How it works

Type a prompt, pick an aspect ratio, hit run. The node sends the request to Reve on fal and returns finished images. That's genuinely most of the story - this is one of the simplest nodes in the whole pack, which is refreshing if you're coming off a Wan graph with fifteen settings. You lean on the model's prompt-following rather than a pile of dials.

The inputs and outputs that matter

There's really only one required input and three optional ones:

• prompt (required, multiline) - the whole show. Reve rewards clear, descriptive prompts, so spend your effort here.
• aspect_ratio - 16:9, 4:3, 3:2 (default), 1:1, 2:3, 3:4, 9:16. Frame it for the crop you actually want; Reve composes to the ratio rather than cropping afterward.
• num_images - 1 to 4 per call. Handy for picking the best of a batch, but remember each image is billed.
• output_format - png (default) or jpeg. PNG for quality, JPEG if you want smaller files.

Output is a native IMAGE, ready for Preview Image, Save Image, or an upscaler downstream - no URL step needed like the video nodes in this pack.

Notice what isn't here: no seed, no steps, no CFG, no negative prompt. Reve on fal doesn't expose them through this node, so don't go looking for a seed to lock - reproducibility isn't a knob you get here.

How to install it

Part of gokayfem's ComfyUI-fal-API pack. ComfyUI Manager: search ComfyUI-fal-API, install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/gokayfem/ComfyUI-fal-API.git
pip install -r ComfyUI-fal-API/requirements.txt

Add your fal key - copy config.ini.example to config.ini and paste the key from fal.ai/dashboard/keys, or export FAL_KEY=your_key - then restart. No model download; Reve lives on fal.

Common issues & troubleshooting

No seed means no reproducibility. If you got a great image and want to iterate on it, you can't re-roll the exact same one. Save the winners as you go.

Prompt not being followed the way you expect. With so few knobs, prompt wording is your only handle. Be concrete about subject, composition, and any text. Vague prompts get vague results and you can't blame a bad CFG.

Insufficient funds / late credits. fal is pay-per-call and legit, but its billing has occasionally been slow to reflect top-ups. If credits haven't appeared, let them settle instead of paying again.

A refusal image or policy error. fal passes its real error message through to ComfyUI. If a prompt trips content policy, the text will say so - read it rather than guessing.