01
Posters and packaging with real words
Headlines, labels, signage, price tags. Test the text first — it is where image models usually break.
AI Image Tool
Text to Image
All-in-One AI Text to Image Generator
Describe a picture and get it back, up to 4K. Three models share the same prompt box: one gets text right, one keeps a set consistent, one is fast. You pick by what you are making.
GPT Image 2Nano Banana ProNano Banana 2
Creative workspace
Text to Image
8
What is Text to Image?
Text to image is generating a picture from a written description. You describe the subject, the framing, and the look; the model renders it. There is no photograph, no illustration, and no source file to start from.
How the people who built the model describe it:
“The OpenAI API lets you generate and edit images from text prompts using GPT Image models, including our latest, gpt-image-2.”
Text to Image at a glance
- You bring
- A description
- You get back
- One image, up to 4K
- Models
- Three, picked per job
What you can make with the Text to Image tool
These are the jobs people actually bring to Text to Image, and the settings each one needs.
02
A campaign set, not six loose images
Six images that share one palette, one lens, one light. Not six good images that obviously came from six different prompts.
03
The shot no stock library has
Your product on a surface nobody photographs it on. An idea that only works as a metaphor.
Text to Image
A workspace shaped for the model
Text to Image puts the handful of controls that change the result most in front of you, and keeps the rest out of the way.
01
Three models, one prompt box
GPT Image 2 for readable words. Nano Banana Pro when a set has to match. Nano Banana 2 when you are still deciding and want an answer quickly.
02
Describe a scene, do not list keywords
Subject, framing, light, mood — written as sentences. Keywords give you their average. A described scene gives you a composition.
03
Draft at 1K, deliver at 4K
Settle the idea where a run costs least, then re-run the one you kept at full size. Exploring and finishing should not cost the same.
04
Pick the ratio before the first draft
An image composed for 16:9 does not survive a crop to 9:16. Choosing up front costs less than fixing it later.
05
Failed runs are refunded
If no image comes back, the credits do. An experiment that goes nowhere costs nothing.
Text to Image
Choose with context
Where Text to Image is the right call, and where a different model does the job better.
Starts from
A blank page
The prompt decides
Everything
Pick this when
You have no image to work from yet
How to get an image you can actually use
01
Say what the picture is
One sentence for the subject and what it is doing. One for where it is and how it is lit. Skip the pile of adjectives.
02
Put exact text in quotes
Write the words the way they should appear, then choose GPT Image 2 for that run.
03
Draft at 1K, change one thing at a time
Iterating is cheapest here, and one change per run tells you which change worked.
04
Re-run the keeper at 2K or 4K
Same prompt, same model, bigger output. That is the version you ship.
Questions about text to image
Start Creating With Text to Image
Set up the brief now, then continue in the shared Studio workspace.