How the AI Face Matching Works
Curious what actually happens when you hit "Start Scan"? Here's the honest, plain-English version: how the app checks photo readiness, processes selected photos privately, reviews visible resemblance, and keeps the result firmly in just-for-fun territory.
Ready to settle the debate?
Upload a few photos, preview the scan, and unlock a private feature-by-feature family likeness report.
Start Your ScanNo sign-up · Private AI processing · Entertainment only
Step 1: Check photo readiness
When you upload a photo, quick browser checks look for one clear face and assess practical issues such as blur, framing, face size, and viewing angle. If a photo is hard to analyze — or contains several faces — the app flags it so you can replace it before running the scan.
Step 2: Prepare the selected photo set
The selected photos are resized for a reliable upload while preserving the full visible face and its surrounding context. The app keeps the chosen-first ordering consistent so the preview, saved report, and visual crop boards all stay anchored to the same primary photos.
Step 3: Review visible resemblance
When you generate the preview, the selected photos are sent for private AI review. The model considers the photo set together and describes visible similarities and differences across features such as the eyes, brows, nose, mouth, face shape, and other observable traits. It does not analyze DNA or establish identity.
Why more than one photo helps
A second or third clean photo can make the analysis steadier because the AI has more context for separating a lasting facial trait from one expression, angle, or lighting condition.
Step 4: Build and validate the report
The AI response is organized into a consistent report: overall family-likeness strength, scan reliability, parent lean when both parents are present, feature-by-feature attribution, and plain-English analyst notes. The app validates that structure before presenting or saving the result instead of inventing missing sections in the browser.
Step 5: Show reliability and limits
The report keeps likeness strength separate from scan reliability. Reliability reflects how usable the photos are and how clearly the visible observations line up; it is not a probability of parentage. Difficult angles, heavy shadows, strong expressions, age gaps, and a one-photo set can all make the signal less steady.
How private processing works
Dad or Mad may run quick photo-readiness checks in your browser, then uploads selected photos for private AI processing when you generate a preview or full report. Saved scans can be revisited or deleted from the product. The full details are in our privacy policy.
Why it's for fun
Face matching is great at one job — measuring how similar two faces look. It is not genetics. Children's faces change, lighting and angle move the numbers, and family resemblance is gloriously unpredictable. So enjoy the verdict, share it widely, and read can a face really tell you who a child looks like? for the honest limits.
Frequently asked questions
Does the AI use my photos to train itself?
No. Selected photos may be uploaded for private AI preview or report generation, but they are not used to train Dad or Mad models.
What makes the AI more or less confident?
Photo quality is the biggest factor: clear, front-facing, well-lit images with one face produce the most reliable results. Blurry, angled, or multi-face photos lower the confidence.
Can it really tell who a child looks like?
It can tell you how similar the faces look, which is genuinely fun and often convincing. It cannot determine genetics or parentage — that's not what face matching does.
Does it work on a phone?
Yes. It runs in any modern mobile or desktop browser, and most people use it on their phones.
Give it a whirl
It takes about a minute. Upload, analyze, share the verdict.
Preview the scanNo sign-up · Private AI processing · Entertainment only