Ideogram V4 is their most capable image generation model to date — world-class typography and text rendering with the best text training data in the industry, clean and detailed photorealistic outputs, and strong artistic/illustration quality for posters, editorial illustrations, and design-forward imagery.
Key Features
Text Rendering
- Accurate Typography: Renders legible, correctly spelled text inside generated images.
Image Generation
- Photorealistic Output: Produces photo-grade images suited for production use.
Technical Capabilities
| Spec | Details |
|---|---|
| Inputs | Text-to-Image |
| Resolution | square_hd · square · portrait_4_3 · portrait_16_9 · landscape_4_3 · landscape_16_9 · custom (up to 14142×14142) |
| Aspect Ratios | square_hd · square · portrait_4_3 · portrait_16_9 · landscape_4_3 · landscape_16_9 |
Limitations
- Text-to-image only on fal: The fal V4 endpoint accepts text prompts only — no image input, editing, or reference images in the V4 schema. Editing and reference features are available via separate V3 endpoints.
- No video output: Image generation only; no video or animation support.
- Style references not in V4 API: Up to 3 style reference images are documented for V3 but not present in the V4 fal endpoint schema.
Prompting Tips
Prompt Formula: Subject + Style/Medium + Text intent + Composition details
- Prompt expansion is on by default — short prompts work well, but disable it for precise control.
- Specify exact text in quotes when you need typography in the image.
- Use rendering speed QUALITY for maximum detail; TURBO for fast iteration.
- Acceleration levels (low/regular/high) trade fidelity for speed — start with "none" for final assets.
- For style consistency across a set, use the V3 style-reference endpoint with up to 3 reference images.
- Pair with the Reframe endpoint to extend a finished image into wider aspect ratios without regenerating.
Where can I use it?
Image Generation in the AI Toolkit — try it here.