Fake Detector

Preamble

Fake Detector handles the most dangerous claim in the stack: whether the card is real, fake, or too poorly photographed to judge.

---

The Fake Can Survive The Photo

A fake card can look convincing in a photo. Texture, foil behavior, print quality, ink density, surface finish, and microscopic defects can disappear under glare, compression, sleeves, poor focus, or seller-controlled staging.

Authenticity is harsher than centering. A bad centering measurement can misprice a submission. A bad authenticity call can bless the wrong object.

Real, Fake, Uncertain

The output needs three honest states: real, fake, and uncertain. Each state needs evidence attached: print features, holo pattern, OCR mismatch, color profile, corner and edge clues, reprint markers, source provenance, and the image defects that limit the claim.

A clean authenticity badge without reasons would be the most dangerous output the tool could produce. When the photo cannot support the claim, the app should ask for better evidence and leave the card unresolved.

Compare Before Confidence

Authenticity needs known references, variant awareness, and provenance. A suspected fake only means something beside a trusted record of what the real card should look like under comparable conditions.

Fake Detector becomes credible when it can show the comparison, name the mismatch, preserve the source, and refuse cases where the visible evidence is too thin.