Model Prompt LabCompare models

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. 1

    Define the subject and destination format

  2. 2

    Add composition, lens, lighting, and text constraints

  3. 3

    Choose the model by the hardest constraint

  4. 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.