Guided views keep the card framed, upright, and supported by enough whole-card evidence.
SECURING YOUR WORKSPACE
Opening GradeLens…
Your private Vault and report state are being checked.
SECURING YOUR WORKSPACE
Your private Vault and report state are being checked.
A hybrid grading engine that combines measurable image quality, computer vision, published industry standards and GPT-powered visual reasoning—without pretending a phone photo can replace physical authentication.
The companion site intentionally does not grade uploaded photos. Guided framing, orientation, image-quality checks, and source-photo capture stay in the iPhone app. Your resulting evidence report then syncs into one private card workspace for review on any screen.
Owner alpha active · wider TestFlight access is coming soonGuided views keep the card framed, upright, and supported by enough whole-card evidence.
The report, source photos, notes, and methodology travel together through Cloud Vault.
Review confirmed sales, graded comparisons, price history, and the full evidence dossier here.
Every iPhone report stays connected to its source images, condition findings, confidence limits, and full evidence dossier.
The phone app handles guided photography and grading. This companion keeps reports readable, searchable, and ready for research without mixing market value into the grade.
The model does not simply look at two photos and guess a number. GradeLens separates capture quality, observable defects, scoring rules and uncertainty into auditable stages.
Checks source resolution, sharpness, clipping and glare before condition analysis begins.
On-device preflightEstimates card boundaries, front and back centering, squareness and visible dimensional anomalies.
Computer visionGPT vision classifies only visible corner, edge, surface and print evidence with confidence.
GPT visionA published, versioned formula applies severity penalties, weights and grade ceilings.
Deterministic codeThe condition model uses the shared core of major grading rubrics—centering, corners, edges, surface and dimensions—then applies its own documented score. It is not TAG, PSA, Beckett or CGC, and it does not claim equivalence to an in-person grade.
A future capture box can lock the camera, block ambient light and sequence opposing LEDs. The same report and scoring engine can then consume normal maps, edge macros and calibrated dimensions.