How Nostalgia chooses a repair path
Nostalgia uses several kinds of AI, but the product is designed so you do not have to choose a technical model yourself. Photo Insight reads the photo first, then recommends a repair path that fits the image in front of you.
Every upload is reviewed for capture quality, visible damage, glare, blur, faces, color, and likely era before a repair is recommended.
A faded portrait, a scratched print, and a blurry phone capture need different steps. Nostalgia recommends the safest useful order instead of applying every tool at once.
AI can repair a lot, but it can still make mistakes. The original stays preserved so you can compare, rerun, rescan, or stop when the result feels faithful.
What the different tools are for
The tools are grouped by job, not by technical implementation. Start with the tool that solves the visible problem, then stop when the photo looks faithful.
| Tool group | When it helps |
|---|---|
| Repair tools | Restore, denoise, deblur, and deglare are used when the source has visible damage, grain, softness, or reflections. These are the first tools to consider for most old prints. |
| Detail and color tools | Enhance, face repair, and colorize are optional follow-up steps. They work best after the basic repair is clean enough to support them. |
| Creative tools | Animation, background removal, and Reimagine are separate from faithful restoration. Use them when you want a clearly labeled creative version, not as the archival source. |
| Caption and story help | Text help can draft captions, stories, and date ideas from visible clues. Treat those drafts as a starting point and replace guesses with family knowledge whenever you have it. |
Restore, denoise, deblur, and deglare are used when the source has visible damage, grain, softness, or reflections. These are the first tools to consider for most old prints.
Enhance, face repair, and colorize are optional follow-up steps. They work best after the basic repair is clean enough to support them.
Animation, background removal, and Reimagine are separate from faithful restoration. Use them when you want a clearly labeled creative version, not as the archival source.
Text help can draft captions, stories, and date ideas from visible clues. Treat those drafts as a starting point and replace guesses with family knowledge whenever you have it.
Where AI needs your judgment
AI restoration is useful because it can repair damage at a scale that manual editing cannot. It still needs human review, especially on faces, dates, captions, and color.
| Limit | What to do |
|---|---|
| A better scan can matter more than another repair | If the upload is tiny, angled, heavily glared, or out of focus, Photo Insight may recommend recapturing before you spend time on AI repair. |
| Missing detail cannot always be recovered | Severe tears, missing emulsion, heavy water damage, and very small faces may produce approximate results. Keep the original and review faces closely. |
| Color is plausible, not proof | Colorization uses visual context to make a likely color version. It does not know the exact color of a dress, car, wall, or uniform unless your family records say so. |
| Sensitive originals may need a conservator | For fragile, moldy, stuck, or historically important originals, physical preservation comes first. Digitize carefully and consider professional conservation advice. |
If the upload is tiny, angled, heavily glared, or out of focus, Photo Insight may recommend recapturing before you spend time on AI repair.
Severe tears, missing emulsion, heavy water damage, and very small faces may produce approximate results. Keep the original and review faces closely.
Colorization uses visual context to make a likely color version. It does not know the exact color of a dress, car, wall, or uniform unless your family records say so.
For fragile, moldy, stuck, or historically important originals, physical preservation comes first. Digitize carefully and consider professional conservation advice.
Privacy and ownership
Your uploaded photos and restored outputs are not used to train AI models. Results stay in your private Archive unless you choose to share them, and you can export or delete your data from your account.
Try it with one photo
Upload a photo, read the Photo Insight recommendation, then compare the repaired result against the original before you save it as the keeper.