Browse sample poster results on the right, then tap one to load that prompt back into the generator on the left.

Old photo restoration workflow overview

Restore damaged photos while preserving identity

Use reusable prompts for scratches, fading, colorization, portraits, albums, and archival scenes.

Preserve faces

Keep facial structure, age, expression, and pose faithful to the reference photo.

Repair real damage

Guide GPT Image 2 to remove scratches, folds, stains, dust, blur, and faded contrast.

Recover natural color

Colorize black-and-white or sepia photos with restrained, historically plausible tones.

Reuse the workflow

Start from a prompt, attach a reference image, then refine the restoration constraints.

How old photo restoration works

Choose a prompt, attach the source photo, then generate a careful restoration with GPT Image 2.

Choosing an old photo restoration template

Choose a restoration template

Pick the prompt that matches the source photo, from family portraits to ID photos.

Editing restoration prompt variables

Edit repair constraints

Replace variables for damage type, color treatment, crop, and preservation rules.

Generating a restored photo with GPT Image 2

Generate and refine

Use GPT Image 2 to create a restored draft, then adjust for a more faithful result.

Old photo restoration FAQ

Practical answers for using GPT Image 2 restoration prompts.





Restore an old photo

Load a GPT Image 2 restoration template and generate a faithful first draft.