π€ Leon Ideogram Image API
Ideogram v2's legendary text rendering, piped into a ComfyUI node
- image
- image_url
- seed
Leon Ideogram Image API wraps Ideogram's v2 generation models - ideogram-v2 and ideogram-v2-turbo - and routes them through the HyprLab API. Ideogram's whole identity is text-in-image accuracy: through v3 it was the closed benchmark for putting legible words, logos, and UI text into generated images, the thing local checkpoints kept fumbling. If your ComfyUI workflow needs an image with correct spelling, this node is the shortcut.
One important framing note: this is the old closed v2 line, not the Ideogram 4.0 that open-weighted in June 2026 and became a local ComfyUI checkpoint. If you want Ideogram 4 locally, this isn't it - that's a different install entirely (and a different workflow, since 4.0 wants JSON-conditioned prompts). This node is for the hosted v2 API, which still has the text-rendering chops and the classic Ideogram style controls, just no open weights.
How it works
Standard pack plumbing: JSON POST to https://api.hyprlab.io/v1/images/generations with a Bearer token, 5-attempt backoff retry, download, RGBA tensor out. The inputs:
model-ideogram-v2(default) orideogram-v2-turbo. Turbo is the fast/cheap tier; plain v2 is the quality pick.prompt- max 10,000 characters. This is where you put the text you want rendered - "a coffee shop sign that reads 'MORNING BREW'".negative_prompt- Ideogram actually supports one, so you get to say what NOT to include.aspect_ratio- a wide menu (11 options) including16:10/10:16and ultrawide3:1/1:3, which the pack's other nodes mostly don't offer.style_type-Auto,General,Realistic,Design,Render 3D,Anime. TheDesignstyle is the marketing-materials one;Animeis self-explanatory.magic_prompt_option-Auto/On/Off. Ideogram's magic prompt auto-rewrites your prompt for better results;Offkeeps your words literal, which you'll want when the exact text matters.
Outputs: image, image_url, seed.
Installing it
Same pack, same dance:
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" via ComfyUI Manager. No model downloads; the models run hosted.
Where people get burned
The recurring pack traps: placeholder key (YOUR_API_KEY_HERE), slow-failing retries on a bad key, and the thin public documentation around HyprLab as a reseller. On top of that, keep the v2-versus-4.0 distinction straight - if someone told you to "use Ideogram in ComfyUI" they may have meant the open-weight 4.0, which this node is not. And remember Ideogram v2 is a closed subscription model with its own content rules baked in; the hosted route inherits them. For text-heavy generation - logos, posters, anything with words - this is one of the more useful nodes in the pack.
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | cat | Main text input to guide the generation process (max 10,000 characters) |
| model | COMBO | ideogram-v2 | Ideogram AI model to use for generation |
| output_format | COMBO | webp | Format of the output image |
| seed | INT | 00β18446744073709550000 | Random seed for reproducible results |
| api_url | STRING | https://api.hyprlab.io/v1/images/generations | API URL |
| api_key | STRING | YOUR_API_KEY_HERE | Your HyprLab API key |
| response_format | COMBO | url | How the response data should be formatted |
| negative_promptopt | STRING | Input that defines what NOT to include during generation (max 10,000 characters) | |
| aspect_ratioopt | COMBO | 1:1 | Aspect ratio of the generated image |
| style_typeopt | COMBO | Auto | Artistic style or genre for the output |
| magic_prompt_optionopt | COMBO | Auto | Toggle for automatic magic prompt system |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | β |
| image_url | STRING | β |
| seed | INT | β |