ChatGPT photo-editing prompt library

ChatGPT Photo Editing Prompts for Real Editing Tasks

Copy and use practical ChatGPT prompts for real photo editing tasks. Remove objects, change backgrounds, retouch portraits, and improve product photos.

Real editing examples

See what these prompts can do

Example edits made with these prompts. Results vary with your source image and model.

Remove unwanted objects

Before editing: a person crossing the street in the background. Before
After editing: the background person removed and the scene reconstructed. After
Remove the person crossing in the background. Reconstruct the missing wall and floor so the texture, lighting, and perspective stay consistent. Keep the main subject unchanged.

Change photo background

Before editing: the subject on its original background. Before
After editing: the subject placed on a clean studio background. After
Replace the background with a clean warm-white studio sweep. Keep the subject’s edges sharp, preserve natural shadows on the ground, and do not alter the subject’s colors or proportions.

Natural portrait retouch

Before editing: an unretouched portrait. Before
After editing: the portrait retouched with natural skin texture and lighting. After
Soften skin and even out tone while preserving pores and natural texture. Brighten the eyes slightly, reduce shadows under them, and keep the face recognizable and unchanged in identity.

Start here

Choose your editing task.

Each focused page includes eight short and detailed prompt pairs, constraints, and fixes for common failures.

How it works

How PromptEdit works

Three steps from an editing goal to a result you can check.

  1. 01

    Choose your editing goal

    Select the task you want to complete.

  2. 02

    Copy the prompt

    Use a short or detailed version.

  3. 03

    Refine the result

    Adjust the prompt based on the output.

FAQ

Upload the source image, paste a prompt, replace any bracketed details, and describe what must remain unchanged. Review the output closely and make the next request smaller if too much changed.