Comflowy Ideogram

Ideogram is the model you reach for when the image has to contain words that are actually spelled right. Posters, logos, album art, infographics - the open-source world still struggles with legible typography inside generated images, and Ideogram has been the market leader at it since the original v1. It's closed, it's hosted, and normally it's stuck behind Ideogram's own web app. This node drags it into your ComfyUI graph, same as the rest of the Comflowy pack drags in Flux, Kling, and company.

The README positions it exactly that way: closed-source, usually unusable in ComfyUI, so Comflowy built a node for it. The catch is the meter - Ideogram v2-turbo runs about 800 credits per image, Ideogram v2 about 500. Text-rendering quality is the most expensive thing in this pack, apparently.

How it works and the inputs

Standard FlowyApiNode: prompt and settings POSTed to app.comflowy.com/api/open/v0/ideogram with your key, output URL downloaded back to an IMAGE tensor. The node surfaces Ideogram's own parameter set, including a couple of genuinely useful ones:

• prompt and negative_prompt (both required, multiline) - the text to render and what to avoid. Put your actual text in quotes within the prompt; that's how Ideogram models are trained to treat it.
• version - ideogram-v2-turbo or ideogram-v2. Turbo is cheaper and faster; v2 the full-strength model.
• resolution - a big list (79 entries) of exact pixel sizes, default None. This one overrides aspect ratio when set, per the node description - pick a specific resolution and the aspect ratio field stops mattering.
• style_type - None, Auto, Realistic, Design, Anime, Render 3D. Auto lets Ideogram pick.
• aspect_ratio - the usual suspects, 1:1 through 3:1 / 1:3, used when resolution is None.
• magic_prompt_option - On or Off. Magic Prompt rewrites and optimizes your prompt to maximize variety and quality, and the description notes it can also translate prompts across languages. Leave it On unless you need exact wording preserved.
• seed - 0 to 2^31-1.

Output: a single IMAGE.

Installing and the gotchas

Install Comflowy's Custom Nodes via ComfyUI Manager (search "Comflowy") or git clone https://github.com/6174/comflowy-nodes into custom_nodes, restart, and set your key with Comflowy Set API Key first. Only requests as a dependency.

The two things that bite people:

• Resolution vs. aspect ratio confusion. They're two separate dropdowns and the node description spells out the rule: resolution wins. If you set both, your aspect ratio is silently ignored. Pick one system and stick to it.
• The credit meter. At ~500–800 credits an image, iterating on typography gets expensive fast. Use magic_prompt_option On to avoid re-rolling for variety, and get your prompt right before you pay.

Plus the pack-wide usuals: network errors against app.comflowy.com (the README's top-cited failure), and the "API Key is not set" refusal if you skipped the Comflowy Set API Key node.

If your workflow is photoreal or anime and never involves text, this node will collect dust - the open models are fine for that. The moment someone asks for "a poster with the title in fancy serif," it becomes the only sensible answer in the pack.