Question 1

What kinds of old photos work best?

Accepted Answer

Any photograph with visible age damage: color fade, yellow cast, scratches, dust, creases, stains, water damage, or emulsion loss. It works on black-and-white prints, color prints from any era, slides, and even smartphone photos of old pictures. The higher the resolution you feed in, the more detail the restoration can recover.

Question 2

Will the person in the photo still look like themselves?

Accepted Answer

Yes. The prompt specifically tells the AI to preserve the subject exactly — facial features, expressions, clothing, posture, and framing. The restoration fixes damage to the print, not the content of the photograph. That said, gpt-image-1 is an AI model and on heavily damaged faces (missing eyes, corrupted features) it may have to guess at some detail. For irreplaceable family photos with severe damage, compare results carefully before keeping them.

Question 3

Can it colorize black-and-white photos?

Accepted Answer

Not intentionally — our prompt specifically asks gpt-image-1 to preserve the original tonal range. If the source is black and white, the output stays black and white (but cleaner and sharper). If you want to add color to a grayscale photo, that's a different workflow — try the main Studio with a custom prompt that asks for colorization explicitly.

Question 4

How does this compare to traditional photo restoration?

Accepted Answer

A professional restoration artist in Photoshop can be more precise for faces and hands, but takes hours or days per photo and costs $50-200 per image. AI restoration finishes in 30 seconds and is free to try. Quality is excellent for 80-90% of common damage patterns (fade, scratches, dust, discoloration), and good enough for almost anything short of severe physical destruction.

Question 5

What about photos with rips or missing chunks?

Accepted Answer

gpt-image-1 can reconstruct small rips and missing corners by inferring from context. For large missing sections (more than ~15% of the image), results become speculative — the AI has to invent what was there. If your photo has a large missing piece covering important detail, consider using the Watermark Remover tool instead to paint a precise mask over just the missing region.

Question 6

How much does each run cost?

Accepted Answer

Each restoration run costs 4 credits (~$0.06 on a Basic plan). That is the true cost of a gpt-image-1 medium edit plus a small margin — we pass provider cost through rather than subsidizing it. New accounts get 10 credits on signup so you can restore a photo or two without paying. For large batches, a subscription plan bundles credits at a lower effective rate.

Question 7

Will it work on photos I took on my phone last week?

Accepted Answer

Yes, but it's designed for age damage not general enhancement. If your photo is sharp, well-exposed, and color-accurate already, restoration won't change much — try the AI Image Upscaler for resolution improvement or open the main Studio for creative enhancement. The restore tool shines on photos that actually look old.

Question 8

Is my old photo private?

Accepted Answer

Yes. Uploads are stored in your private account workspace and never shared, sold, or used for model training. OpenAI receives the image for processing but doesn't retain it per their API data policy. You can delete any upload from your account at any time.

一键修复老照片

Not a filter — a real restoration

Fixes age-related damage

Corrects color fading

Sharpens without over-processing

Preserves subject and composition

Two steps from faded to fresh

Upload the old photo

Click Restore Photo

Download and compare

常见问题解答

想要更精细的控制？