How to Use AI Pre-Grading and Centering Checks Before Mailing a Card
use AI pre-grading as a screening pass, not as a promise of the grade another company will assign. Photograph the front and back clearly, run a separate
Jamie Budesky·Founder, CardGrade·Published Jul 26, 2026 · Updated Aug 3, 2026·5 min read
In short: use AI pre-grading as a screening pass, not as a promise of the grade another company will assign. Photograph the front and back clearly, run a separate centering check, review corners, edges, and surface yourself, and mail only after the card still makes sense under your cost and value assumptions. CardGrade does not predict or guarantee a third-party grade.
What this workflow is and is not
This guide answers a narrower question than our AI card grading app guide: what should you do after choosing a tool but before putting a card in the mail? It is also broader than the card centering guide, because centering is only one condition category.
An AI result can help rank cards, highlight areas for another look, and create a repeatable first pass. It cannot authenticate a card from photos, see a defect the image does not reveal, reproduce a grader lighting and handling, or bind PSA, TAG, or any other grader. PSA itself says grading has a subjective component and reserves judgment calls near centering limits. Treat every score or ratio as decision support, never a third-party-grade prediction.
Exact image-quality requirements for a useful screen
Use this capture gate before trusting any result:
Both sides: provide one complete, straight-on image of the front and one of the back.
Full perimeter: keep all four corners and every edge inside the frame; do not crop the border.
Sharp detail: card text, corner tips, edge fibers, and print detail must look crisp when enlarged. If they do not, retake the image.
Even light: reject frames with glare, deep shadow, reflections, or a hotspot hiding any surface area.
Square camera position: keep the sensor parallel to the card so perspective does not make one border appear wider.
Clean separation: use a plain, contrasting background and keep fingers, sleeves, top loaders, labels, and dust from covering the card.
No editing: do not apply filters, sharpening, retouching, or perspective correction. Preserve what the camera recorded.
Extra inspection views: for foil, chrome, or textured surfaces, add angled-light views for your manual review. Do not substitute those for the straight-on front and back used for centering.
These are acceptance requirements for this workflow, not claims about what a grading company accepts. They follow from what the official rubrics ask an evaluator to judge: centering, corners, edges, surface, focus, gloss, staining, and small defects. TAG explicitly describes artifacts found under high-resolution imagery.
Run centering as a measurement, not a verdict
Measure horizontal and vertical borders on both sides where the design has identifiable borders. Record the worse axis rather than averaging away a weak direction. Borderless, tilted, irregular, or intentionally asymmetric designs may not support a reliable ratio; mark those results inconclusive.
Use a grader current published standard only as a reference. PSA published PSA 10 description gives an approximate 55/45 front and 75/25 reverse tolerance, but PSA also says eye appeal can affect a judgment call and a small variance may be allowed at grader discretion. Passing a ratio therefore does not establish a grade.
Centering: front and back, horizontal and vertical, with design ambiguity noted.
TAG public rubric uses these same four headings and shows why resolution matters: even high grades may turn on minor corner, edge, or surface artifacts visible in detailed imagery.
Apply the hard-stop limitations
Do not let a favorable AI result overrule these stops:
Possible alteration or authenticity issue: escalate to an expert; photo screening is not authentication. PSA lists trimming, recoloring, restoration, tampering, and questionable authenticity among reasons it will not grade a card.
Hidden or ambiguous area: retake the photo or mark the screen inconclusive.
Borderless or misleading geometry: do not force a centering ratio.
Foil glare or shallow defects: inspect under changing light by hand.
Grade-dependent economics: rerun the math at a lower outcome and include fees, shipping, insurance, and time.
Make the mail-or-hold decision
Create three buckets: mail, manual review, and hold. Mail only cards with usable images, no unresolved red flags, a manual inspection consistent with the screen, and acceptable economics under a conservative outcome. Put uncertain authentication, alteration, surface, or image cases into manual review. Hold the rest. Then follow the physical preparation steps in How to Prepare Cards for Grading.
The CTA comes last because the useful decision comes first: if you want a repeatable first pass, run CardGrade on clear front-and-back images, then complete the centering and hands-on checks above before you submit.
Sources and methodology
Proposed 2026-07-26. We reviewed PSA official Grading Standards and TAG official Grading Rubric as retrieved on 2026-08-03. Official rubrics support claims about evaluated condition categories and published tolerances; our capture checklist is a conservative editorial workflow designed to make those features visible. We did not use CardGrade blog copy as factual evidence, and we make no claim that CardGrade predicts a third-party grade.
Share
Ready to grade your cards with AI?
Pre-screen your cards before submitting to PSA, BGS, or CGC. Plans from $4.99/mo.
Jamie Budesky is the founder of CardGrade and the engineer behind its AI vision grading pipeline. An Army veteran and IT specialist (DoD, since 2017), he writes about card grading, AI/ML grading technology, and collecting strategy — grounded in CardGrade's own grading data.
