Read this before your first clip

Film it like a coach would watch it.

MotionGrade reads your movement from the footage you give it. A coach standing in the wrong place can't see your base either — the camera has exactly the same problem.

Nine measures. How many of them we can read is decided before you press record.

9 measures 4 survive almost any angle 3 need a turned camera 0 uploads — measured in your browser
The rule underneath all the others

The distance between you and the camera must not change.

Every measure is a ratio against how big your body is on screen. Let the camera drift closer, or run away from it, and the ruler changes mid-measurement — so a movement that never changed reads as though it did.

Two set-ups compared side by side. On the left, a phone on a tripod and three identical figures, all the same size, because the distance holds. On the right, the same phone and three figures that shrink into the distance as the athlete runs away. Distance holds Distance opens up
The athlete on the right never changed their form. The measurement did, because the ruler did.

It is also why panning is not a fix. Swivelling to keep someone in shot keeps them in frame while the distance carries on changing underneath the measurement. Framing and distance are two different problems, and only one of them is solved by turning the phone.

The one that matters most

Stand where the camera can see the gap between your feet.

Almost everything else is lighting and framing. This one is geometry, and it silently costs you three measures.

A plan view seen from overhead. An athlete stands at the centre with their feet apart on a diagonal stance line. A ring around them is green in the two zones at right angles to that stance line, where a camera can see the gap between the feet, and amber in the two zones aligned with it, where one foot hides the other. Camera sees both feet One foot hides the other Your stance line
Green is every camera position where the read works. It is a quarter-turn off the line your feet point in.

You can check it before you press record: look at your own screen and see whether there is daylight between your feet. If there is, the base is measurable. If one shoe is hiding behind the other, step the phone around until it isn't.

That's the whole rule — and the next section is what it costs when you get it wrong, measured on real clips rather than asserted.

It depends what you train

Some sports hold the distance for you. Running does not.

If you work inside a box — striking, racket sports, throwing, jumping — set the phone down and walk away. If the movement travels, the camera has to travel with it at a fixed gap.

Two set-ups that work for running. Above, a treadmill with a phone on a tripod holding station directly behind it. Below, a runner with someone on a bicycle behind them holding a phone at the same fixed gap. The dimension line between camera and runner is identical in both. Treadmill, camera holding station Bike, keeping the same gap
The gap is the same length in both. That is the entire running protocol.
Two set-ups that fail for running. Above, a phone left on a tripod while the runner disappears into the distance. Below, a tripod swivelling to follow a runner who is still getting further away. Phone left behind Panning is not a fix
Both keep you in frame. Neither keeps the distance, and the distance is what the measurement divides by.
A plan view from overhead. An athlete stands inside a marked working area with faint ghosted copies of themselves at other positions inside it, while a phone outside the area keeps a fixed distance from its near edge. Your working area This distance never changes
Boxing, martial arts, tennis, padel, throwing, jumping. Move freely inside the box — what breaks the read is moving toward or away from the camera.

And one that is simply out of scope: this is not for barbell work. No squats, no bench, no presses. It measures the timing and sequencing of fast functional movement — throwing, jumping, running, punching — and a loaded lift asks a different question with a different answer.

A real one, not a diagram

Two of our own clips. Same measure, two camera positions.

Both of these tracked cleanly. The skeleton is locked on in both. The only difference is where the phone was standing.

A golf swing filmed down the line. The skeleton and joint angles track cleanly, but the two feet almost completely overlap and head stack reads 8 per cent.
✗ Filmed down the line L/R match 85% · Balance 76% · Head stack 8%

Look at the feet — they sit almost on top of each other. That's a textbook golf camera position and the tracking is fine, but the stance has almost no measurable width left, so the measure built on that width collapses.
A boxer filmed side-on. Both feet are clearly planted and separated, and head stack reads 79 per cent.
✓ Filmed side-on L/R match 91% · Balance 94% · Head stack 79%

Both feet clearly planted, clearly apart. Same tracking, same overlay, same product — but now there's a base to measure against, and the measure comes back.

We're showing you a bad score on purpose. That 8% is the tool being honest, not the tool being broken — it would have been trivial to return a confident-looking number from that clip, and most things do. A score you can't trust is worse than no score, because you'll coach off it.

The rules, in order of how much they cost you

Eight things that decide your read.

A side elevation. A phone on a tripod at hip height, its field of view fanning out to an athlete standing so their head and feet sit just inside the frame. Below, two smaller failure cases: one where the athlete is too close and their head is cropped, one where they are too far away to measure. Whole body inside the frame Phone at hip height
Four of the eight in one picture: whole body in frame, hip height, fill about half the frame, phone still on something.
What survives a bad angle, and what doesn't

Not all nine fail together.

Timing measures are robust. Anything that needs your base is fragile — which is why rule 02 and rule 03 carry so much weight.

The app tells you this up front, before it analyses anything — "this angle should give you 6 of 9 measures" — while you can still re-film. If you see a number you don't like there, move the camera and go again. That's cheaper than a bad read.

The short version

Film it the way you'd want it watched.

◱

Phone still, on something

Prop it, lean it, tripod it. Handheld drift is the quiet killer of feet-in-frame.

◐

Light on you, not behind you

Shooting into a window or the sun turns you into a silhouette. Silhouettes have no joints.

↔

Side-on, feet apart in view

If a coach standing there couldn't judge your base, neither can we.

← Back to MotionGrade