GPT Image 2.6 Sunburst

Create detailed images with GPT Image 2.6 Sunburst

Choose Sunburst for complex compositions and edits where small details matter. Build a new scene from text, or upload a product photo or portrait to refine lighting, materials, and background.

Try GPT Image 2.6 Sunburst
Korean idol collage portrait
Creative reference, not a verified output or before-and-after comparison. Upload your own source in the editor when a reference is needed.

When to use this model

Start with one clear reference or a brief that describes the subject, framing, and light. For a product edit, list what should stay in place, such as the silhouette, label, and camera angle. Start with a 1K draft to settle the composition, then raise resolution or quality when the image needs finer detail.

A visual starting point

Korean idol collage portrait
Korean idol collage portrait
Vintage Amalfi travel poster
Vintage Amalfi travel poster
Surreal koi nebula illustration
Surreal koi nebula illustration

Creative reference, not a verified output or before-and-after comparison. Upload your own source in the editor when a reference is needed.

A prompt to get started

An example prompt to try, not a generated result. Edit it to fit your project.

Edit the uploaded product photo into a catalog hero image. Set it on a matte slate surface with soft side lighting from the left and a natural contact shadow. Keep the product silhouette, label placement, color, and camera angle. Leave space around the product for cropping.
Try this prompt

Supported settings

  • Text to image
  • Reference image editing
  • 1K
  • 2K
  • 4K

Settings reflect the current DrawNew integration.

OpenAI’s Image API also lets both models adjust how much planning goes into an image, use circles or arrows to guide local edits, and output scalable SVG graphics or animated GIFs. These are model API capabilities. In DrawNew you can generate from text or edit a reference image, then choose an aspect ratio, resolution, and rendering quality.

From your input to the next version

  1. Choose the model and open the editor.

  2. Write your prompt, add supported references and choose output settings.

  3. Generate, review the result, and download it. Refine your prompt to try another version.

Your questions, before you create

What should I prepare before starting?

Upload your reference image in edit mode. Describe what should change and what should stay the same.

Which reference files can I upload?

Reference limits: 14 images, 0 videos, 0 audio files A value of zero means that reference type is not supported.

Will the result preserve the original exactly?

A reference guides generation but does not guarantee exact preservation. Describe what must stay the same and compare the result with your original.

What should I check before using the result?

Compare the result with the reference at full size. Check small text, logos, material texture, and whether light and shadows agree. Correct any changed product or brand details before publishing.

Make the first version of your idea

Start with a clear brief, explore a direction, then refine the details.

Try GPT Image 2.6 Sunburst