1Upload the best source you have
The quality hierarchy is: flatbed scan > Google PhotoScan > phone photo. If you have the original print and access to a scanner, scan it. If you are working from a phone capture, PhotoScan's glare removal makes a visible difference compared to a regular camera shot.
2Crop before uploading
Remove album page edges, tape, handwriting on borders, and any non-photo material before uploading. The AI models work best when the entire image is the photo itself. Cropping also reduces file size, which speeds up processing.
3Use batch for large collections
If you have a box of photos to work through, batch processing lets you upload many at once and process them in sequence. The smart pipeline adapts to each photo's condition automatically. Save individual processing for your most important or complex photos.
4Review before archiving
AI restoration is impressive but not perfect. Always compare the before and after — especially for faces. If something looks off, try a different tool order or re-upload a higher quality scan. The goal is faithful preservation, not a perfect AI rendering.
5Organize with collections
After restoring, group photos into collections by family branch, decade, or event. This makes your archive browsable and shareable. Family Vaults let you collaborate with relatives — each person can contribute scans and everyone benefits from the restored versions.
6Start with your best photos
When learning the tools, begin with photos in decent condition. This helps you see what the AI can do before tackling heavily damaged originals. Once you are comfortable with the workflow, move on to the tougher cases.
7Save originals separately
Keep your original scans in a separate folder — do not overwrite them with restored versions. Nostalgia preserves both in your Library, but having local backups of originals means you can always re-process with future improvements.