Motion blur and QR scanning

The one-line physics

During an exposure, relative motion drags every edge across the sensor:

smear = speed × exposure time.

A decode survives when the smear stays under roughly half a module width, edges stay distinguishable and the sampler still lands on the right side of the threshold. Beyond that, adjacent dark and light modules average into grey and the grid dissolves.

The drive-by arithmetic

Codes aimed at moving traffic fail on arithmetic, not luck. A car at 50 km/h covers 13.9 m per second. A phone in daylight shoots around 1/60–1/125 s once you account for a hurried, moving subject:

Speed	1/500 s (bright, still hand)	1/60 s (typical)	1/30 s (dim)
Walking, 5 km/h	0.3 cm	2.3 cm	4.6 cm
Cycling, 20 km/h	1.1 cm	9.3 cm	18.5 cm
Driving, 50 km/h	2.8 cm	23 cm	46 cm
Motorway, 100 km/h	5.6 cm	46 cm	93 cm

Even a billboard-scale code with 10 cm modules is smeared past recovery by the 23 cm swipe of a 1/60 s exposure at 50 km/h. And that ignores the part where the driver should be driving. Codes for drivers are a design error, one of the clear cases in when not to use a QR code. The honest pattern for vehicle wraps and roadside media is a code that works when parked or stopped, backed by a short memorable URL for everyone in motion.

The walking row explains everyday experience: a pedestrian who pauses half a second scans fine; one who scans mid-stride in a dim station concourse does not.

Handheld shake

Even standing still, hands tremble. At normal exposures (1/100 s and faster) the tremor moves the image far less than half a module and scanning feels instant. The trouble starts in dim light, where exposures stretch to 1/30–1/15 s and the same tremor becomes a multi-module smear, which is why low light and motion blur are really one combined failure, covered from the light side in QR scanning in low light. Brace the phone, hold for a second, and let autofocus settle. Optical stabilisation helps but does not repeal the arithmetic.

How industry freezes motion

Parcel sorters read codes on belts moving metres per second, seemingly impossible by the table above. They cheat on exposure: industrial 2D imagers use global-shutter sensors with intense pulsed illumination, cutting effective exposure to tens of microseconds. At 1/10,000 s, a belt at 3 m/s smears just 0.3 mm, nothing. The lesson transfers: motion tolerance is bought with light and shutter speed, which a phone in a dim room has neither of.

Design margin

If your code will plausibly be scanned by people in motion (transit riders, queue walk-pasts) buy margin the same way you do for distance:

• Bigger modules: print larger than the distance minimum via the size calculator, or shorten the payload.
• Light the placement so exposures stay short.
• Give people a reason to stop, a caption and a spot to stand beat any amount of oversizing.

Then confirm the artwork itself is sound with the validator, so motion is the only variable left.

FAQ

Can you scan a QR code from a moving car?

Effectively no. At 50 km/h the car moves 14 m per second, so an ordinary 1/60 s exposure smears the image about 23 cm, wider than even billboard-scale modules. Roadside codes only work for stopped vehicles, and inviting drivers to scan is irresponsible anyway.

How much movement can a QR scan tolerate?

The smear during the exposure must stay under about half a module width. At bright-light exposures near 1/500 s that permits slow walking; at dim-light exposures near 1/30 s even handheld tremor can exceed it.

Why do my scans fail at night but work in daytime?

Exposure time. In dim light the camera exposes for 1/30 s or longer, so the same hand shake drags the image many times further across the sensor than it does in daylight. Brace the phone or add light.

How do warehouse scanners read codes on fast conveyor belts?

With global-shutter sensors and bright pulsed illumination that cut effective exposure to tens of microseconds. At 1/10,000 s a belt moving 3 m/s smears only a third of a millimetre, so the code is frozen mid-flight.