How we score
No mystery numbers. This page explains exactly how your physique score, body-fat estimate and training plan are produced — and what a photo can and can't tell us.
Body-fat % — always a range
Photo-based body-composition estimation is validated science — studies comparing 2D-photo estimates against DEXA scans report strong agreement — but it is still an estimate. Lighting, pose and clothing move the number. That's why we always show a range, never a false-precision decimal, and why we never claim "DEXA accuracy".
Same photo, same score
Re-analysing the same photo returns the same result — scores are cached per photo, so your progress deltas reflect your body changing, not model noise. For week-to-week comparisons, our capture guide helps you match angle, distance and lighting.
Your plan — published training science
Programs are generated from the hypertrophy evidence base: weekly volume landmarks per muscle, a minimum of two sessions per muscle per week, hard sets taken 1–2 reps from failure, double progression, and scheduled deloads. Every change to your plan comes with the reason attached.
How your program is built
- Weekly volume per muscle comes from published landmarks — for most muscles, 10–20 hard sets a week is where growth actually happens.
- Every muscle gets trained at least twice a week — frequency spreads the work so each set counts.
- Hard sets stop 1–2 reps short of failure (RIR) — close enough to grow, far enough to recover and do it again in 48 hours.
- Double progression: hit the top of the rep range at your target RIR, then the weight goes up. No guesswork, no ego lifts.
- Deloads are scheduled into the block, not improvised after you're already run down.
- The muscles your scan flags as lagging get priority volume — the plan attacks your weak points first.
The research basis: volume dose-response meta-analyses and the RIR-based progression literature. We use the published consensus ranges, not one cherry-picked study.
The closed loop
Scan → program → log → re-scan → verified delta. Your scan builds the program, your logs drive the progression, and the next scan measures whether any of it worked — against your own baseline, in the same light, at the same angle.
Your rank only moves when a re-scan proves change. We can't be fooled by streaks — and neither can you.
What we don't do
- Store your photos on our servers — analysis is in-memory; progress photos live on your device.
- Score your face, genetics, or anything you can't train.
- Promise outcomes. Projections are bounded by realistic natural rates of change.