Input mode guide
Text-to-image workflow
Start from a written brief and choose the image model that matches typography, realism, speed, and composition needs.
A repeatable workflow
- 1
Define the subject and destination format
- 2
Add composition, lens, lighting, and text constraints
- 3
Choose the model by the hardest constraint
- 4
Generate variants and keep one variable per iteration
Frequently asked questions
What should a text-to-image prompt include?
Start with the subject and destination format, then add composition, lighting, lens, style, and any text constraints.
How do I choose between realism and typography?
Choose by the hardest constraint: Nano Banana Pro for natural photography, or GPT Image 2 for legible text and strict layout.
How many variables should I change per iteration?
Change one major variable per iteration so you can identify what improved or degraded the result.