License plate recognition is one of the few pieces of security technology that either works well or barely works at all, with very little in between. The difference is almost never the software.
Angle
The camera needs a shallow angle to the plate, both horizontally and vertically. A camera looking steeply down at a vehicle sees a foreshortened plate, and a camera off to one side sees a compressed one. Both are harder to read than they look on a screen, and both are decided by where the camera is mounted rather than by any setting.
Speed
The plate has to be in frame long enough to be captured cleanly. In practice this means the camera goes where vehicles are already slow: a gate, a barrier, a ramp, an entrance where people have to turn. Reading plates off a road at speed is a different and much more demanding piece of engineering than reading them at a gate.
Distance and lens
A plate needs to occupy enough pixels to be resolved, and that is a function of the lens and the distance together. This is why an LPR position is specified as a lens at a distance, not as a camera model. Move the same camera ten meters back and it stops working.
Light
Plates are retroreflective, which is useful and awkward at once. At night, headlights overwhelm a normal camera’s exposure and the plate washes out. Dedicated LPR cameras handle this with infrared and a much shorter exposure, giving up the rest of the scene in exchange for a readable plate. That trade is the point, and it is why the image from an LPR camera looks wrong if you judge it as a general camera.
Which is why it is its own camera
An overview camera that happens to see the gate will not read plates well, because the two jobs want different lenses and opposite exposure settings. LPR belongs alongside the overview camera, not instead of it. A quote that promises plate capture from a general camera position is promising something the physics does not support.
What you get when it works
A searchable log of plates and times. That is the actual product: not footage, but a list you can query after something happens, and a gate log and a lot that agree with each other.
Where it earns its place
LPR is most valuable where vehicles are the asset or the risk. In practice that means dealerships, equipment yards, fleet depots, storage sites and multi-tenant parking, which is where a plate log answers questions nothing else can.
It is worth being clear about what the log does and does not settle. It tells you a plate was at a point at a time. It does not tell you who was driving, and it does not identify a vehicle whose plate was obscured, missing, or false. Those are the cases people most want it for, and they are the cases where the overview camera alongside it does the work.
Which is the argument for specifying the pair together. The LPR camera gives you something searchable, and the overview camera gives you the context that makes the search result mean something.
Whether a given gate can actually read a plate is a question Elite Security answers on site, because it depends on the angle and the light rather than on the model number. More like this in the blog, and AI analytics and LPR is where to start.