Image editing handles a wide range of tasks. Here are the most popular use cases, with a tip for getting the best result from each.

Restore old or damaged photos — remove scratches, denoise, fix fading. A direct instruction like "restore this old photograph" works well as a starting point, then refine specific damage in follow-up passes.

Clean up backgrounds, adjust lighting, or tidy styling for a polished result. Lock identity first — "preserve the same facial features, expression, and pose" — before changing anything else.

Specify texture, color depth, and length precisely, and always protect facial features. See the dedicated hair editing guide for detail.

Generate a set of images of the same person across different settings or styles. Use the same identifying labels every time and lock facial features, expression, and pose so identity stays consistent across the series.

Remove unwanted text or watermarks cleanly. Be explicit: "remove the watermark in the bottom-right corner, fill in the background to match."

Apply a named, recognizable style — cartoon, Minecraft style, age progression, gender swap, and more. Name a specific style rather than a vague one, and keep composition unchanged: "convert to [style], maintain the same composition and subject."

Tip for all of these: one transformation per pass. Complex results come from chaining small, successful edits — not from one overloaded prompt.