Can Your Phone Camera Actually Check Your Form?
Yes, for some things, and no for others. Here is exactly what a single phone camera can see about a lift, what it cannot, and why we cut a whole category of exercises because of it.
Five things decide whether a camera-based form check measures anything real: the angle, the framing, the height, the light, and what else is moving. Here is what each one does and why it matters.

Most bad form checks are not the model's fault. They are a phone propped against a water bottle, pointing the wrong way, in a dark corner, with somebody walking through the shot.
This is written from building Therox, which checks your form and counts your reps off your phone camera. The advice is not Therox-specific though. Any camera-based form check is reading the same joint positions and fails for the same reasons.
Five things decide whether the check measures anything real. None of them take longer than about fifteen seconds to get right.
This is the one that matters most, and the easiest to get wrong.
A single camera sees flat. It has no idea how far away anything is, so it can only measure an angle that happens to be facing it. Squat depth lives in the plane you see from the side; film it head on and the knee angle collapses to almost nothing. Arm symmetry is the reverse, because comparing your left side to your right only works if both are visible, and from the side one arm hides the other.
So every movement needs a specific angle. Therox names the one it needs on each exercise card before you start, and here is the whole set:
| Movement family | Film from |
|---|---|
| Squat patterns | The side |
| Hip hinges | The side |
| Horizontal press | The side |
| Overhead press | The front |
| Vertical pull | The front |
| Rows | The side |
| Arms | The side |
| Shoulders | The front |
If you film from the wrong side, a good check does not guess. Therox marks that check as not judged and tells you why, both on the set summary and against the individual check: filmed face-on, this needs a side view.
The rep is still scored on whatever could be measured, and each one says how much that was, as a line like "Clean (2 of 5)". So a set filmed from the wrong angle still gives you a number, just one standing on fewer legs than it looks. That is the real reason to spend fifteen seconds on the setup.
Head to feet, including the feet.
The check is built on the position of your joints, so a joint that leaves the frame takes its check with it. Cut your feet off on a squat and there is no ankle, which means there is no reliable knee angle, which means there is no depth reading. Cut your head off on an overhead press and the lockout has nothing to be measured against.
The common failure is standing too close. Most phones need more distance than people expect, and the fix is almost always to move the phone further away rather than to change what you are doing.
A phone on the floor looking up, or on a high shelf looking down, distorts the very angles the check is reading. A squat filmed from ankle level looks deeper than it was. The same squat filmed from a shelf looks shallower.
Hip height, roughly level, is close enough. You do not need a tripod. A bench, a plate stack or a gym bag does the job.
Even light beats bright light. The model is finding the edges of your limbs, so what hurts it is not dimness so much as contrast it cannot read: a window blazing behind you, a single hard spotlight, deep shadow across half the platform.
If you are backlit, turn around. That one change fixes more bad readings than any other lighting adjustment.
Two separate problems with the same fix.
A moving camera moves everything in the picture, and some checks read small changes in position. If the phone is drifting, that movement is added to yours. Prop it against something solid rather than leaning it on something that will slide.
Other people in frame are a harder problem, because a second body gives the model a second set of joints to consider. In a busy gym, the best you can usually do is angle the phone so the background is a wall rather than the walkway.
Before you start the set:
Getting the setup right removes the reasons a check fails for camera reasons rather than form reasons. It does not make the underlying numbers more certain than they are, and those limits are still starting values rather than measured ones.
It also does not turn any of this into a substitute for learning the lift. Think of it as a second pair of eyes on every rep, useful between the sessions where someone is watching you properly.

Yes, for some things, and no for others. Here is exactly what a single phone camera can see about a lift, what it cannot, and why we cut a whole category of exercises because of it.

The ten minutes you lose before your first rep usually go to someone else's feed. Here is what fixes it, what it costs you when it does not, and the part we have not fixed yet.