How to Size an AR Marker for a Distance

The marker works perfectly on the desk. Mounted on a wall three metres away it detects intermittently, and at five metres not at all.

Nothing is wrong with the marker. It occupies too few pixels, and pixel count is what detection actually depends on.

Detection is a pixel count problem

A marker detector needs the pattern to resolve into enough pixels to read the cells inside the border. Below that threshold the pattern is a grey smear and no amount of processing recovers it.

How many pixels a marker occupies depends on three things:

pixels across = marker size × resolution ÷ (2 × distance × tan(fov/2))

Distance — inversely proportional. Double the distance, halve the pixels.

Camera resolution — proportional. A 4K sensor resolves twice the detail of 1080p at the same field of view.

Field of view — a wide-angle lens spreads the same pixels over more scene, so each object gets fewer of them.

The practical rule that falls out: to detect at twice the distance, double the marker size. This is linear and it surprises people who expect some tolerance.

The quiet border is not optional

Marker detection works by finding the black border against a lighter surround. Without adequate white space around the marker, the border cannot be found and nothing else happens.

A marker printed to the edge of the paper fails. A marker on a dark surface fails. A marker with a decorative frame touching it fails.

The convention is a white margin at least one cell wide around the whole marker, and more is safer. It costs paper and nothing else.

Print finish and lighting

Matte, always. Glossy stock produces specular highlights under directional lighting, and a highlight across part of the pattern wipes out the cells beneath it. A partly obscured marker is not partly detected; it is not detected.

Flat mounting. A marker on a curved or wrinkled surface distorts the pattern, and pose estimation degrades before detection fails entirely — which produces the worse failure mode of a marker that detects with a wrong position.

Even lighting. Strong shadows across a marker split it into regions of different brightness, and thresholding struggles.

Angle of view degrades detection too

Distance is the obvious variable. Viewing angle is the one that catches people in deployment.

A marker viewed square-on presents its full area. Viewed at 45 degrees it presents roughly 70% of its width, and the pattern is distorted by perspective. Past around 60 degrees off-axis, detection becomes unreliable regardless of size.

That matters because markers are frequently mounted flat on a wall while the camera approaches along it. The marker is at its largest apparent size when the viewer is directly in front, which may be exactly where they never stand.

Two responses: size for the worst-case combination of distance and angle rather than distance alone, or mount the marker angled toward where viewers will actually be.

Pose estimation degrades before detection does, so a marker at a steep angle may still be detected while reporting a position that is noticeably wrong — the worse failure mode of the two.

Common mistakes to avoid

• Sizing for the typical distance rather than the maximum.
• Printing to the edge with no quiet border.
• Using glossy or laminated stock.
• Mounting on a curved surface, which corrupts pose estimation before it breaks detection.
• Using a marker dictionary with more markers than needed, which reduces the inter-marker distance and increases false detections.

How to do it with AR Marker Size Calculator

The AR Marker Size Calculator works from your camera’s actual specification.

1. Enter the maximum distance detection must work at, not the typical one.
2. Enter the camera’s horizontal resolution and field of view.
3. Print at the calculated size or larger, with a quiet border.
4. Use matte stock and mount it flat.

Other engineering calculators are in the tools directory.

Frequently asked questions

Why does my marker fail at range?

Because it occupies too few pixels. Pixel count falls linearly with distance, so a marker readable at one metre needs to be three times larger to work at three.

Does the quiet border matter?

Considerably. Detection finds the black border against a lighter surround, so a marker printed to the paper edge or on a dark surface fails even when it is large enough.

Matte or glossy print?

Matte. Glossy stock produces specular highlights that obscure part of the pattern under directional lighting, and a partly obscured marker is simply not detected.

Final thought

Size for the furthest distance and add a border. Detection is linear in distance, so a marker that works at your test range will not stretch to twice it.

Try the free AR Marker Size Calculator