How to Use GPT Image 2

To use GPT Image 2, choose a workflow, write a clear prompt or upload an image, generate a first result, then refine it with follow-up edits.

Step-by-step guide

Step 1

Choose your workflow

Decide whether you want text-to-image generation, image editing, upscaling, or image-to-image transformation.

Step 2

Write a clear prompt or upload an image

For generation, describe the subject, scene, style, and lighting. For editing, upload the image you want to improve.

Step 3

Generate a first result

Aim for a usable first draft instead of trying to solve everything in a single prompt.

Step 4

Refine the output

Use follow-up instructions to improve composition, detail, style, background, or clarity.

Step 5

Export or keep iterating

Export when the result is ready, or continue with more specific edits if it still needs work.

Beginner tips

  • Be clear about the subject, scene, style, and lighting
  • Start with a usable draft, then iterate
  • Use reference images when consistency matters

Common mistakes

  • Prompts that are too vague
  • Trying to solve too many things in the first prompt
  • Skipping reference images when consistency is important

Related pages

Prompt examplesImage GeneratorImage Editor

FAQ

Which tool should beginners start with?

Start with the Image Generator for creating from scratch, or the Image Editor if you already have an image to improve.

Do prompts need to be very long?

Not necessarily. Start by being clear about the subject, scene, style, and lighting, then refine from there.

When should I use a reference image?

Reference images are especially helpful when you need consistency for people, products, or layout.