Two numbers multiplied together
Motion blur in a security frame is not complicated. Take how fast the subject moves across the frame in feet per second, multiply by how long the shutter is open in seconds, and you have how far it moved in feet. Multiply that by the pixel density at that spot and you have the smear in pixels. That is the whole calculation, and it is why the answer surprises people: the shutter times are small enough to feel harmless and the pixel densities are large enough to make small distances matter.
On the defaults — 15 mph at 25 degrees to the lens, 1/250 s, 60 px/ft — the car moves 0.45 in across the frame during the exposure, and that is 2.2 pixels. Slow the shutter to 1/30 and the same car smears 18.6 pixels, which is seven times the painted stroke of a plate character. Same camera, same car, same spot, and the only thing that changed is a setting the camera picks for itself when it gets dark.
Only the sideways part blurs
A vehicle driving straight at the lens barely smears at all. It gets bigger in the frame, which is a scale change rather than a translation, and the character edges stay where they are. A vehicle crossing side on smears the full amount. Everything in between splits by the sine of the angle, which the page does for you.
This produces the awkward trade that every plate-capture position runs into. Head-on is best for blur and worst for the angle you view the plate at, because a plate seen at a steep oblique compresses its own characters and the retroreflective sheeting throws light back along the axis it came from rather than towards a camera off to one side. Somewhere between those there is a workable angle, and the way to find it is to drive a car past at the speed people actually drive.
The shutter is not free
Every doubling of the shutter speed halves the light reaching the sensor. That is why cameras left on automatic drift to slow shutters after dark, and why the footage that matters most is the footage most likely to be smeared. Buying two stops of shutter means finding two stops of light from somewhere: more illumination, a faster lens, more gain and the noise that comes with it, or a spot where the vehicles are slower.
The page prints the stop count against a reference shutter you supply, so the trade is visible as a number rather than as a feeling. It does not tell you which side of the trade to come down on, because that depends on what the footage is for and how the scene is lit.
Frame rate decides how many attempts you get
Blur decides whether one frame is any use. Frame rate decides how many frames you get. A car at 15 mph covers 22 ft a second, so at 15 fps it moves about a foot and a half between frames, and a 20 ft capture zone gives it fourteen chances. At 4 fps the same zone gives four. Drop the zone to a few feet, because that is where the angle and the density are both acceptable, and the number gets uncomfortable quickly.
This is the part that gets cut first when storage is tight, and it is the part that turns a working capture into an intermittent one. The storage arithmetic is on its own page; what is worth knowing here is that halving the frame rate does not halve the file, because most of a compressed stream is the moving parts rather than the count of frames, but it does exactly halve the number of attempts.
What the arithmetic leaves out
All of it is a floor. Rolling shutter skews a fast subject as well as blurring it. Compression is tuned to throw away high-frequency detail, which is precisely what plate characters are made of. Infrared lights retroreflective sheeting far more brightly than the paint on top of it, so a plate at night can arrive as a white rectangle with nothing on it. A lens that is soft at full aperture has already spent some of the resolution before motion enters. None of those is in the number at the top of the page, and every one of them makes the real result worse than it says.
Questions people ask
What shutter speed do I need to read a plate on a moving car?
It depends on the crossing speed, the angle and the pixel density, and this page works it out from those rather than quoting a figure. On the defaults — 15 mph at 25 degrees, 60 px/ft, one pixel of blur allowed — it wants about 1/560 s. Double the speed and it wants double the shutter. Turn the camera side on to the path and the crossing component goes up by a factor of about 2.4, and so does the shutter it asks for.
Why is my security camera footage of cars blurry at night?
Almost always because the camera slowed its own shutter to keep the picture bright. Every doubling of shutter speed halves the light, so an automatic exposure in a dark yard drifts to a long exposure and a moving vehicle smears across the frame. The camera is not faulty and a higher resolution will not fix it, because the smear is measured in pixels and more pixels means more of them get smeared.
Does more resolution reduce motion blur?
No, it makes the smear larger in pixels. Blur is a fixed distance on the ground during the exposure, and raising the pixel density turns that same distance into more pixels. What more resolution buys is that the object is bigger to start with, so the ratio of detail to smear improves if the shutter keeps up. If it does not, the extra pixels are all smeared ones.
What angle should a camera watch a driveway from?
Head-on to the traffic gives the least blur, because motion towards the lens changes scale rather than sliding the image sideways. It also gives the worst view of anything painted on the back of the vehicle. The workable answer sits between the two and depends on how fast people drive in, which is why the page takes the angle as an input and prints what it costs rather than recommending one.
How many frames will a camera get of a passing car?
Divide the length of the usable zone by the distance the car covers between frames. On the defaults — 20 ft of zone, 15 mph, 15 fps — that is about a foot and a half between frames and fourteen chances. At 4 fps it is four chances, and whether one of them is usable starts to depend on luck rather than on the camera.