How BackMaxx works, from one photo to a training plan

How BackMaxx works, end to end

BackMaxx runs one loop, five steps long, and repeats it every week. You photograph your back, an AI vision model reads the photo, the app hands back a Back Score out of 100 with a rank alongside six zone scores and eight muscle scores, and the weekly plan that follows leads with your lowest-scoring zones. Seven days later you shoot the same photo and the loop runs again.

None of it is a measurement. Every score is an estimate read off a photograph, which is exactly why the second scan is worth more than the first.

1. Capture. One rear photo, selfie camera, 5-second self-timer, phone propped somewhere stable.
2. Analysis. A vision model grades what the frame shows: width from the armpit down, thickness between the blades, the slope into the neck, the roundness behind the shoulders, the detail running down the middle of the lower back, and left against right.
3. Scores and rank. A 0–100 Back Score, one of nine rank bands from Newbie to GOAT, plus six zone scores and eight muscle scores on that same 0–100 scale.
4. Plan. Your three lowest zones become the week’s priorities, filled from a corpus of 77 movements, with five fixed focus programs available if you would rather run a block.
5. Re-scan. Same photo, same spot, one week later. Scores update, the zone map redraws, and the two images sit side by side in a timeline.

Everything below is that loop in detail, including the four things that move your score without changing your back.

Shooting the one photo the scan reads

The photo is the only manual part of the loop. You use the selfie camera so the screen faces you while you frame it, start the 5-second self-timer, then turn around before it fires. One frame, not a burst, and no second angle.

What the model can grade is limited to what the frame shows, so the useful photo is the boring one: even light, plain wall, waistband in shot, arms hanging rather than flared. A lat spread inflates Lat Width in a photo the same way it does in a mirror.

Repeatability beats quality here. A slightly dim photo taken in the same spot every week produces a more honest trend than a great photo taken somewhere new each time, because the week-to-week difference is the thing you are actually reading. Locking the phone height, the distance and the time of day removes most of the noise before it ever reaches the score, which is the whole argument in the guide to shooting back progress photos you can compare.

Nothing sits between the photo and the score. No account, no email address, no sign-in step.

Six zone scores and eight muscle scores from one frame

Analysis turns one image into fourteen sub-scores: six zone scores and eight muscle scores, all on the same 0–100 scale as the headline Back Score. The zones are what you read week to week; the muscle scores are what sits underneath them.

Each zone is a specific visual claim rather than a body part. Symmetry is the clearest case — it is left against right across the whole frame, so no single muscle is assigned to it and no movement in the corpus lists it as a primary target.

Upper Back is the busiest zone in the corpus by a distance: 19 movements name it as their primary target and another 36 pick it up as a secondary. It is also first in line whenever the session builder has a spare slot left over, so Upper Back work is hard to avoid whatever your lowest zone turns out to be.

The eight muscle scores exist because zone scores hide things. A mid-range Upper Back can mean thick rhomboids with nothing sitting behind the blades, or the exact reverse, and the muscle breakdown is where that split shows up.

Infraspinatus sits at the bottom of that list with 8 tagged movements, and that is not a gap in the corpus. It trains as a rotator rather than a puller, so face pulls and external rotation carry it and heavy rowing barely touches it. The other seven, and what each one actually changes in a photo, are laid out in the eight back muscles BackMaxx scores.

The Back Score and the nine rank bands

Six numbers tell you where to work; one number is the one you remember. The Back Score is the headline 0–100 figure the scan returns alongside the zone scores, so the weekly comparison becomes a single subtraction, and the rank turns that figure into something you can say out loud.

The bands are deliberately uneven. Newbie swallows 38 points on its own because almost everything below the high thirties reads the same in a photograph, while Grandmaster is 7 points wide and Titan only 6.

The practical consequence is that the same eight points buy different amounts of rank depending on where you start. Moving from 41 to 49 takes you out of Starter and into Semi-Pro; the identical eight points from 22 to 30 keeps you in Newbie, and once you are past 91 there is no band left above you. The scoring itself, band by band, is broken down in how the Back Score is built and what each band looks like.

Building the plan around your three lowest zones

Plans get built from the bottom of your zone list. The scan sorts all six zones lowest-first, the three lowest become the week’s priorities, and the lowest of those leads a session and takes two of its slots. The other two priorities take one slot each, and any room left over goes to the remaining zones so nothing slides. Across a week of one to three sessions, the lead rotates through those three priorities rather than parking on one.

That single rule is what makes the score worth collecting, because a plan that ignores the score is just a workout.

How much the plan has to work with depends on which zone came in lowest.

Symmetry is the exception that explains the method. Nothing targets it primarily, and all 8 movements that touch it are one-side-at-a-time: Commando Pull-up, Meadows Row, One-Arm Dumbbell Row, Single-Arm Cable Row, Kettlebell Single-Arm Row, Landmine Romanian Deadlift, Single-Leg Romanian Deadlift and Bird Dog. Visual balance gets trained by making each side do its own work, not by a dedicated symmetry exercise.

Equipment shapes the plan almost as much as the score does. 74 of the 77 movements work in a gym and 46 work at home, and the home set is not a consolation prize: it holds 13 of the 21 Lat Width primaries and 11 of the 21 Lower Back primaries. 20 movements need nothing at all, and the six pull-up-bar movements — Active Hang, Dead Hang, Scapular Pull-up, Negative Pull-up, Commando Pull-up, L-Sit Pull-up — are the only ones that need something fixed overhead.

Difficulty is filtered rather than blended. 50 of the 77 movements are beginner, 17 intermediate and 10 advanced, and the builder caps a session at intermediate unless you told it you train as an advanced lifter.

When you want a fixed block instead of a score-driven plan, five focus programs are pre-built:

Read the lead-zone column rather than the title. V-Taper Sprint opens every session with a vertical pull and finishes with a lat-length stretch. Thickness Block runs horizontal pulls first and collects the rear delts on the way past. Pull-Up Progression is the only one that needs a bar, and it spends six weeks on hangs, negatives and assisted reps with lat and upper-back volume stacked behind them so the strength actually arrives. A running program overrides the lead zone your scan picked for as long as its weeks last, then drops away and hands the week back to your latest scan.

Inside one generated session

A session comes out in three blocks. One to three warm-up movements at a single set each — a mobility opener, then activation chosen for that session’s lead zone — followed by the main block where the zone slots land, and up to three stretches to close. The main block holds between three and seven movements depending on how long you said you have, and a short session sheds the stretch block before it sheds main work: under 25 minutes there is no cooldown at all.

Sets and reps come from the corpus rather than from a formula. 70 of the 77 movements carry a default of 3 sets and the other 7 carry 2, and the prescription is a range instead of a number — 4–6 on a deadlift, 12–15 on a rear delt fly — so the load call stays yours. In a long session the opening compound picks up a fourth set. Rest is scaled by session length and experience and lands between 30 and 180 seconds.

Nine movements are timed instead of counted, from a 15–30 second Active Hang to a 30–60 second Child’s Pose. Five of the 77 are stretches — Child’s Pose, Cat-Cow Stretch, Torso Twists, Cross-Body Shoulder Stretch and World’s Greatest Stretch — carrying a corpus default of 2 sets of a 20–60 second hold, and they only ever appear in the cooldown slot, one set each, never as main work.

Every movement in a session shows three line-art frames rather than stock photography, 512 by 512, electric blue on black. 72 distinct frame sets cover all 77 movements, because five variations reuse the frames of the movement they sit closest to. The art is adapted from Everkinetic under CC BY-SA 4.0, and the credits sit at the illustration attribution page.

Re-scanning weekly, and what comes back

Weekly is the interval, and noise is the reason. Day to day, the lamp and the camera angle move the score more than your back does, so a daily scan mostly measures your bathroom. Across a week, real change gets big enough to read past the noise.

A re-scan hands back four things rather than one. The new Back Score and how far it has moved since your first scan, the six zone bars redrawn against that first scan, the before-and-after photo pair, and a fresh position in the photo timeline.

The zone map is the part that changes behavior, because it shows your back as six bars instead of one number. When a different bar comes in lowest, the week’s priorities re-sort behind it and the next plan leads somewhere else without being asked. If you have not scanned yet, the back rating walkthrough shows exactly what the first result looks like before you install anything.

What the score is based on, and what it is not

Grading a photo is pattern recognition. The model reads the visible shape of your back — where the width sits, how much separation shows between the blades, how far the traps rise into the neck, whether the two sides read the same — and returns an estimate of how developed that looks.

The estimate is based on one thing: what is visible in that frame. It is not based on your strength, your training age, your bodyweight, your body fat, a tape measure or anything under the skin. BackMaxx does not measure your body, and nothing in the app claims to.

So two people can carry the same Back Score and look different, and the same person can score differently in two photos taken ten minutes apart in different light. That is a property of reading a body off a photograph, not a bug in one app.

Where the number earns its keep is as a series compared against itself. A Back Score says very little as an absolute verdict on your back and a great deal as a run of weekly readings taken in the same spot, in the same light, by the same phone.

Known limits: lighting, pose, leanness, camera distance

Four things move your Back Score without changing your back at all. Knowing them is the difference between using the score and arguing with it.

Leanness is the one that catches people out, because the score reads the surface and nothing else. The Christmas tree in the Lower Back zone and the ridge the rhomboids raise are plainly visible at one condition and invisible at another with identical muscle underneath. Treat a sudden jump across Upper Back and Lower Back in a week you dieted as information about your condition, not about new tissue.

Camera distance matters more than it sounds like it should. A phone held at arm’s length by a friend is a different photograph from a phone propped across the room on a shelf, and the two will not hand you the same Lat Width number in the same week.

None of that makes the score useless. It makes consistency part of the method, which is why the app wants the same photo rather than the best photo.

Where your back photos actually live

Your back photos are the most personal thing this app touches, so the handling is deliberately thin. No account, no email address, no sign-in, no ads. Nothing ties a photo to you because nothing about you is collected in the first place.

The photo is analyzed transiently and is never stored on a server. Your images, your scores, your ranks and your before-and-after timeline live on your iPhone — which also means they leave when the app does, so keep your own copies of any photo you want a year from now.

BackMaxx is iPhone-only and needs iOS 26 or later. Scanning and plan generation are free, and a Pro subscription is optional on a weekly or yearly term. The full statement of what is collected and what is not sits in the BackMaxx privacy policy.

Why the loop is the product, not the number

A single scan hands you a number and an opinion, and both are worth about what an opinion from one photograph is worth.

The loop is what makes it useful: three lowest zones, a plan that leads with them, a week of work, then a second photo taken the same way to find out whether those bars actually moved. Six zone scores, eight muscle scores and a rank out of nine are the instrumentation, not the product.

Which is why the honest version of the pitch is a narrow one. BackMaxx will not tell you what your back is. It will tell you, week after week, which part of it is furthest behind and what to train next because of it.

Find out where your back actually stands

One photo, a Back Score out of 100, six zones and eight muscles ranked — then a plan built on whatever came out lowest.

Download on theApp Store Download on the App Store

Free to download. iPhone, iOS 26 or later.