Frequently asked questions
How does it work?
Upload a scan or phone photo of your print. Our AI pipeline sharpens and reconstructs faces, upscales the image 2×, optionally removes scratches and dust, and colorizes black-and-white or sepia photos. You see a free watermarked preview first and pay only if you like it.
How much does it cost?
$3.9 for one photo, $9 for up to five photos in the same order. One-time payment, no subscription.
How do I pay and how fast do I get the full-resolution file?
Pay the exact amount shown on your order page with PayPal (debit/credit card works too). We match your payment by that unique amount. Optionally put your order ID in the PayPal note, then press “I’ve paid — unlock download”. We confirm payments manually and email your download link, usually within a few hours (always within 24 hours).
What happens to my photos? (privacy)
Photos are used only to produce your result. They are never shared, sold, or used to train AI models, and every upload and result is permanently deleted from our server after 7 days. See our privacy note.
Will the AI change how my relatives look?
AI face restoration reconstructs fine detail, and on very small or very damaged faces it can guess. We blend the restored face with the original to keep the likeness. If a face looks “off”, email us and we’ll redo it with a lighter touch or refund you.
Are the colors historically accurate?
No AI can know the real colors of a black-and-white photo; it predicts plausible, natural colors (skin, sky, foliage, fabric). Clothing colors in particular are a best guess.
Which photos work best?
A sharp scan (300–600 dpi) or a straight, glare-free phone photo of the print. Large tears or missing pieces need manual retouching; email us for a custom quote.
Can I get a refund?
Yes. If you are not happy with the result, email [email protected] within 7 days of payment with your order ID.
Example photo sources
All example photos are in the public domain and were restored by our own pipeline:
Technology
Open-source models: DDColor (colorization), Real-ESRGAN (upscaling), GFPGAN 1.4 (face restoration), YuNet (face detection).