Can Sunburst generate images directly from text?
Yes. Submit model=gpt-image-2.5-sunburst and prompt to /openai/images/generations to start text-to-image generation. Describe the subject, environment, lighting, and composition, and specify whether text is needed; if an existing image needs modification, use the editing endpoint.
How can I make reference image edits closer to the original design?
Submit the image and editing instructions to /openai/images/edits, and explicitly provide the Sunburst ID. The prompt should separately describe the intended changes and content to preserve; for multiple reference images, also specify the purpose of each. Focus on a clear goal in each round, and compare with the original image to check subject and brand details.
How should Sunburst masks be prepared?
The mask workflow uses gpt-image-2.5-sunburst:official, with both the original image and mask uploaded via multipart. The mask must be a PNG with an Alpha channel, the same dimensions as the first original image, and no larger than 4MB; transparent areas can be modified, while images with only black-and-white colors and no transparency channel cannot be used as a substitute.
Can I generate multiple images at once and return Base64?
The platform allows n to be set from 1–10, which is suitable for obtaining multiple candidate images; however, response_format=b64_json supports only 1 image. When multiple results are needed, use the URL return method; when Base64 image data must be processed directly, make a single-image request. Do not mix the two settings.
Will multi-round editing automatically remember previous images?
You should use the selected image as the input for the next edit, and state the changes and preservation requirements for the current round in the new instructions rather than relying only on prior requests. For long tasks, you can add callback_url, obtain task_id first, and receive the result upon completion; save the images and instructions from each round for easier comparison and rollback.
