Which way a plate camera points, and how far it reads

The beam is one number drawn as a shape. That number is the camera’s heading: the compass bearing it was recorded as pointing along, counted in degrees clockwise from north. North is 0, east is 90, south is 180 and west is 270.^⁠[1]

To read one:

1. Zoom in to street level. On the map, beams are not drawn on wider views.
2. Start at the pin. The narrow end of the wedge sits on the pin’s dot, and the wedge opens the way the camera was recorded as facing. North is always up on this map, so a wedge that opens to the right means east.
3. Two wedges mean two recorded headings, as at a spot with a reader for each direction of traffic.
4. No wedge means no usable heading. Either nothing was recorded, or the value is one this site does not draw. The counts below show how common that is.
5. Select the pin for the figure. The Direction line in its panel gives the heading in words and in degrees, and a link opens the pin’s OpenStreetMap entry, where the tag appears as the mapper wrote it.

Going by what makers publish, a fixed plate reader covers a short stretch of a few lanes, not a whole street. The figures below are theirs, not tests by this site, and any one camera may be set up differently.

Flock’s row is its general answer, not every model. Its product page also lists a Wide-Range reader, described as having a wider field of view, and a Long-Range one for fast roads with several lanes, with no lane counts or distances for either.^⁠[6] An operator’s account can differ, too: South Portland, Maine, which ran seven Flock readers until June 2026, put a reader’s cone at about two and a half lanes and says detectives chose spots where roads narrow to two.^⁠[7]

Three points in those documents bear on direction:

• One camera, one direction of travel. Flock says a single camera reads traffic heading one way, and that reading both ways takes at least two.^⁠[2] That is one reason two cameras can share a spot and face opposite ways.
• Close to the lane. Flock advises placing a reader as near as it can be to the lane it reads, and against aiming one across other lanes.^⁠[2]
• The lens sets the reach. Motorola lists five lenses for the L5F, each with its own working band: 6 to 27 feet for the shortest, 105 to 125 feet for the longest.^⁠[3]

Rear or front plates? Flock says its cameras photograph vehicles from the back.^⁠[8] Two cities that have used them agree: Provo, Utah, writes that its Flock cameras photograph the rear of passing vehicles,^⁠[9] and South Portland wrote that its own were aimed at vehicles from behind.^⁠[7] The Motorola, Genetec and Leonardo documents above do not say. A camera that reads rear plates looks the same way its traffic is moving, at vehicles driving away from it. A heading alone cannot confirm that for any pin.

Day and night. All four makers say their readers work after dark. Flock’s uses infrared light, which it says shows what reflects or is lit, such as the plate and the vehicle’s lights, and leaves the rest of the frame dark.^⁠[2] Motorola and Genetec list pulsed LED lighting for zero-lux scenes, meaning total darkness,^⁠[3]⁠[4] and Leonardo says its reader works around the clock.^⁠[5]

Not sure whose camera you are looking at? Start with how to tell a Flock camera from the rest.

Facing a road is not the same as reading all of it. A heading leaves out more than it says.

• Not a field of view. A heading is a single line. How wide a camera sees, and which lanes fall inside that, depends on the model, the lens and how it was aimed, and this site’s data holds none of those. Flock calls its own view “intentionally narrow”.^⁠[2]
• Not a range. The longest distance in the table above is 148 feet. A beam that points down a street does not mean the camera reads the length of it. How far a camera stands from you is a separate question, covered in how close the nearest plate camera is.
• Nothing about what is stored. Whether a camera is switched on, what it photographed, how long that is kept and who can search it are not in the map data at all.
• Not a survey measurement. A heading is one mapper’s reading on one day. Treat it as an estimate, and remember that a camera can be re-aimed after it was mapped.

Every pin on this site is a point in OpenStreetMap, and the heading is one of its tags. The direction key holds a number between 0 and 359, counted clockwise from north. Compass letters such as NE are accepted too.^⁠[1] For plate readers, the OpenStreetMap wiki prefers that plain key to the camera-specific camera:direction.^⁠[10] This site reads both, and uses the camera-specific one where a mapper filled it in.

A double camera, two readers at one spot facing different ways, can reach the map in either of two forms:

• One pin with two headings split by a semicolon, such as 90;270. The semicolon is OpenStreetMap’s usual way to give a single tag several values.^⁠[11] Both headings are kept.
• Two pins on the same spot, each with one heading.

Some pins carry a sector instead: two bearings joined by a hyphen, read clockwise from the first to the second, which the wiki describes as a field of view.^⁠[1] This site keeps only the sector’s midpoint as the heading, so the wedge you see is the standard beam, not the sector a mapper wrote. A pin with three or more headings, or a sector of half a circle or wider, is treated as having no single heading.

In the data built on October 2, 2026, 130,627 of 143,570 mapped cameras (91%) have a recorded heading. Of those 130,627, 6,060 (5%) are recorded with two.

A bare 0 (the number itself, or a sector whose midpoint is 0) is the only heading on another 7,295 cameras, 5% of all 143,570. Zero means due north. But among cameras with one heading, a bare 0 turns up more than twice as often as the average for the other multiples of ten (10, 20 and so on), which looks less like that many cameras aimed exactly north than like a value nobody set. So this site treats a bare 0 as not recorded: it is left out of the 130,627 above and gets no beam. A camera tagged that way may well face north. The data cannot say.

The remaining 5,648, 4% of all 143,570, have no heading this site can use: nothing was recorded, the value could not be read as one bearing, or the pin lists three or more.

A camera with two headings counts once for each, so the rows add up to 136,687 headings from 130,627 cameras. The north row leaves out the bare zeros. These are headings as mappers recorded them, and like every count on this site they are floors: they cover only cameras that somebody has mapped.

Nobody at this site visits cameras, so a heading is only as good as its last edit. If you have stood at the camera and the beam points the wrong way, fix it at the source:

1. From public ground, face the way the lens faces and read the bearing from a compass app, in whole degrees.
2. Open the pin on the map and follow the link to its entry in OpenStreetMap.
3. Correct the direction value there, or camera:direction if that is the tag the pin carries. Change only what you saw yourself.
4. Look for the change here after the next refresh of the data, which runs once a week.

The guide to reporting a camera walks through the account and the edit, and corrections sets out what this site can and cannot change. To see which mapped cameras stand closest to an address, start at near me.