Landing pattern
Dashed ellipse is one standard deviation. Colour shows shot shape.
Short answer: 5 ft away, phone at ball height, square to the target
line, in bright daylight — and recalibrate every time the phone moves.
Everything below is why.
The rig
Distance to ball5 ft (4–7 ft usable)
Camera heightball height, 2–6 in
Angle to target line90° ± 5°
Mountweighted tripod, not leaning
Lightbright daylight only
Why 5 ft. Closer and the ball leaves the
frame before you have enough points. Farther and it shrinks: at 5 ft the ball is about
19 px across at 1080p, at 10 ft it is 9 px — you lose half your precision, and
precision is the whole game.
Why ball height. A camera looking down converts real launch angle into a projected one. Sitting it on the ground pointing up has the same problem in reverse.
Why square. The scale maths assumes the ball flies across the image, not toward the lens. Off-square costs you by cosine: 5° is 0.4%, but 25° is 9% on every speed.
Why ball height. A camera looking down converts real launch angle into a projected one. Sitting it on the ground pointing up has the same problem in reverse.
Why square. The scale maths assumes the ball flies across the image, not toward the lens. Off-square costs you by cosine: 5° is 0.4%, but 25° is 9% on every speed.
What makes it inaccurate — worst first
1. Scale error · up to ±15 yd
Pixels-to-inches is measured once, and every error in it passes straight into ball
speed. A 5% scale error is 5% on speed, which is about 14 yards of driver carry.
The big trap: calibrating on the mat instead of the ball. At 5 ft, a ball 6 inches nearer than the surface you measured is a 10% scale error on its own. The app cross-checks LiDAR depth against the ball's known 42.67 mm diameter and warns you when they disagree by more than 8%.
The big trap: calibrating on the mat instead of the ball. At 5 ft, a ball 6 inches nearer than the surface you measured is a 10% scale error on its own. The app cross-checks LiDAR depth against the ball's known 42.67 mm diameter and warns you when they disagree by more than 8%.
2. Spin is modelled, not measured · ±10 yd
No phone can see backspin from the side. It is estimated from club, speed and launch
angle. If your real spin is 1000 rpm off the model, driver carry moves roughly
8–12 yards. This is the single biggest systematic error and it does not go away with
a better tripod.
Side spin is measured, from how much the ball bends off its line — so draw/fade is real even though the spin number is not.
Side spin is measured, from how much the ball bends off its line — so draw/fade is real even though the spin number is not.
3. Low light · breaks it entirely
Freezing a 150 mph ball needs about 1/2000 s. At that shutter the ball still smears
33 mm — most of its own diameter. Indoors, at dusk, or in a net you cannot get that
exposure without ISO so high the ball's edge disappears into noise.
This app is a daylight tool.
4. Frame rate drift · up to ±20%
Phones thermally throttle. If the app assumes 240 fps and the sensor quietly delivers
197, every speed reads 22% high. The native app times actual frame delivery instead of
trusting the nominal rate — but a hot phone in the sun will still drop frames, and
you will see the warning.
5. Phone tilt · ±8 yd
Launch angle is measured against true horizontal, not against the phone. A 3° lean
is a 3° launch error, which is about 8 yards on a 7-iron. The accelerometer corrects
for this, but past ~6° the app asks you to straighten up rather than silently
correcting a large angle.
6. Start direction is the weakest number
From side-on, left/right motion is straight toward or away from the lens, which is
exactly the direction a single camera sees worst. Treat push/pull as indicative.
Carry and ball speed are far more trustworthy.
What accuracy to actually expect
Ball speed, good setup±2–4%
Launch angle±1°
Carry±5–8%
Shot shapereliable past a firm fade
Backspinmodelled — do not quote it
That is not Trackman. It is good enough to gap your bag and to see
your dispersion honestly, which is what most range sessions are actually for.
Where it is weakest is exactly where it says it is weakest.
Flight calculator
This is a calculator, not a measurement. You type the
numbers in and it simulates the flight. It is here because the flight
model is genuinely good — fitted to tour data, 6 yd mean error — so it
is useful for "what would 5 mph more be worth" or for entering numbers
from another launch monitor. It is not your phone measuring
your swing. Nothing in a browser can do that.
Club
Launch conditions
Ball speed
Launch angle
Spin axis — tilt right = fade
Start direction
Total spin
Player
Carry
—yd
Total
—yd
↑
—
Flight
Apex
—ft
Hang
—s
Descent
—°
Offline
—yd
Roll
—yd
Club spd
—mph
estimated
Altitude
Temperature
Carry vs sea level / 20°C
Can this phone track a ball?
A browser cannot do what ball tracking needs — no 240 fps, no shutter
lock, no LiDAR. Rather than take my word for it, this measures your device
and shows you the numbers.
Recalibrate whenever the phone moves. Bumping the tripod between
calibration and your swing invalidates the scale, and nothing downstream can detect a
wrong answer that is self-consistent.