Start with the providers’ own descriptions
FaceCheck’s API exposes uploads, a search reference, progress and result items. Its test mode searches a restricted collection and is not intended to produce meaningful results. PimEyes’ FAQ explains its face-search product, coverage and personal-use guidance. Read those current descriptions before relying on third-party “winner” tables.
A product offering an API and a product offering a consumer search interface may fit different workflows. That difference does not establish which one will find a particular image. Index membership often matters as much as the visual comparison.
Run an honest evaluation
Use a small set of photos you own or have permission to search. Include known public appearances, altered crops and images with no known public source. Write down what you expect before viewing the results. Keep false positives and inaccessible source links in the record.
Check whether each result is the same file, another portrait that resembles it, or a page that actually identifies the subject. Those are different outcomes. Do not give every similar thumbnail the same label of “found”.
Compare the whole experience
- Access. Can the service be used in your region and for your intended purpose?
- Results. What is visible before payment and what is unlocked afterwards?
- Control. Can you remove uploads or request exclusion of your own images?
- Costs. Is the unit a search, a credit bundle, a recurring plan or monitoring?
- Failure. What happens during downtime, an empty search or an expired result?
If you are building an app, add service reliability, documented error handling and restrictions on redistribution. Avoid unofficial wrappers that give you no clear answer about where an uploaded photo goes.
What we can say about Facehunch
Facehunch uses FaceCheck’s documented test mode. A successful generated-image test established that the connection works; it did not establish that FaceCheck is more accurate, cheaper or faster than PimEyes in production. We will not turn that test into a claim we did not measure.
For general copy search rather than facial comparison, revisit how to choose an image-search tool. Changing the method may be more useful than changing between face-search brands.