The claim
Charge people to restore and colourise old family photos with AI.
Steady background interest; PetaPixel covered problems with a new AI restoration tool in 2026.
How we checked
This is an evidence check, not a hands-on test. We reviewed platform rules and policies, published earnings data, regulator warnings and reporting: 4 sources, listed below.
Costs
- Start-up cost
- Varies
- Effort
- medium
What the evidence shows
People value old family photos, but the cheapest version of this service is already sold direct to consumers: MyHeritage offers automatic repair, colourisation and enhancement, with several free repairs before a subscription. AI tools can also change what they restore; PetaPixel found ON1's Restore AI altered faces so much that people were unlikely to be recognisable as family members. Colourisation is guesswork by nature because black-and-white photos carry no hue information, and historians warn it can mislead.
Realistic earnings: No source gives earnings data for this service. The sources suggest that any value lies in careful human checking and retouching, since one-click AI output is cheap to get elsewhere and can invent details.
Limitations
Earnings ranges are typical outcomes for a beginner, drawn from the sources; some people do far better or worse. Platform rules change often, so check the latest policy before you start.
Works for some, not most.
Sources
- MyHeritage launched Photo Repair alongside its colourisation and enhancement tools, with several free repairs before a subscription is needed businesswire.com
- PetaPixel found ON1 Restore AI redrew faces and altered identities so they were unlikely to be recognised as family members (2026) petapixel.com
- Scienceline explains AI colourisation is fundamentally ambiguous because grayscale images lack hue and saturation, with documented errors scienceline.org
- British Library and UCD historians argue colourisation and upscaling add details that were never there and can mislead petapixel.com