Original editorial illustration for this topic. Not an official product screenshot or a depiction of the exact reported event.
Editing with an image model is a conversation: provide an image and explain what should change. The model may alter details outside the intended area, so review every iteration.
Step 1 — supply the original image
Upload an image or open an image you generated. State whether you want a new style or a careful local adjustment. If preserving product identity matters, describe the important unchanged features.
Step 2 — make the change concrete
Keep the person, clothing, facial expression and lighting unchanged. Remove the bright red sign in the background and reconstruct the wall naturally. Do not add new text.
For precise tasks, a written instruction can be paired with selecting an area where the available interface supports it.
Step 3 — inspect and refine
Zoom in on faces, text, hands, logos and patterned materials. If a detail changed unexpectedly, ask for a correction, or revert to the original image and narrow the instruction.
What to expect
OpenAI notes that selections may not be exact and edits can extend beyond a highlighted area. Read the official Images in ChatGPT guide for current controls and limitations.
Primary sources
Official vendor documentation was used to verify the workflow. Some options depend on plan, region, product version or account access.