Ideogram v2

The standard V2 workflow generates new images and typographic layouts from text prompts from scratch, whereas Edit and Remix modify existing images. The Edit variant uses a binary mask alongside your prompt to perform precise, localized inpainting without altering unmasked areas. The Remix variant transforms the aesthetic style or medium of an entire source image, with a strength parameter controlling how closely the output adheres to the original composition.

Use the Edit workflow when you need to replace specific text strings, swap objects, or retouch isolated areas in an existing image without modifying the rest of the canvas. You supply a base image and a matching binary mask where black areas indicate regions to replace. For generating entirely new scenes or exploring broad artistic style transfers, use the base generation or Remix workflows instead.

Wrap the exact words you want rendered inside double quotation marks and keep text blocks under six to eight words for peak accuracy. You can steer font appearances by naming typographic styles such as bold geometric sans-serif or vintage slab-serif, and enforce visual layout hierarchy by describing text positions in your prompt. For flat typography and graphic posters, select the 'design' style preset.

Ideogram v2 offers five specialized presets alongside an automatic setting: 'design' for graphic layouts and vector typography, 'realistic' for photorealistic portraits and commercial ads, 'render_3D' for dimensional assets, 'anime' for stylized illustration, and 'general' for balanced digital art. You can also toggle prompt expansion to control whether the engine enriches basic prompts automatically.

Ideogram v2 struggles with tiny fine-print disclaimer text, legal paragraphs, and complex multi-language scripts rendered on a single line. Highly complex punctuation such as nested slashes or backslashes can confuse the text parser, and low strength settings on intricate anatomical details like hands can produce visual distortions.