Resolution is the number everyone compares, and it is the number that matters least when choosing where a camera goes. A 4 MP camera and a 5 MP camera with the same lens on the same wall produce nearly the same result, because what limits identification is not the sensor's pixel count. It is how many of those pixels land on the thing you need to recognize, and that is a function of focal length and distance. The industry has a formal way to talk about this, and it is rarely printed on the box.
The four levels of "seeing"
Surveillance standards describe four distinct tasks, each with a pixel density requirement. Detection means knowing something moved, and needs very little detail, around 25 pixels per meter. Observation means telling what category of thing it is, a person versus a dog versus a falling branch, around 63 pixels per meter. Recognition means recognizing a specific known person you have seen before, roughly 125 pixels per meter. Identification, the strictest, means being able to describe an unknown person to someone else with confidence, and it wants around 250 pixels per meter on the face.
The arithmetic is unforgiving. A camera with a wide 2.8 mm lens covering a ten-meter-wide yard produces a horizontal pixel density that falls far below recognition at the far edge. The same sensor with a 6 mm lens covering a narrower slice of that yard doubles the density. Nothing about the image "looks low quality" in the wide version. It looks fine. You only discover the deficit the night you need to describe someone, which is the worst possible time.
Why one camera cannot be two cameras
This is the core trade that spec sheets blur. A single fixed-lens camera makes one bet about its scene: broad coverage or reach. A wide lens on a door sees the whole approach but the face at the door may be under 100 pixels per meter. A narrow lens identifies the face but shows nothing of the approach. Neither is wrong. What is wrong is buying one camera to do both, then discovering the compromise at the moment of use.
Practical deployments resolve it three ways. Multiple cameras with different lenses, the correct answer for sites that matter, each camera given one job. A varifocal lens, which lets you frame the scene at installation time, trading flexibility for a higher price and a focus step that must be done carefully at mounting height. Or accepting the division of labor that most homes actually need: one wide camera for general activity, one tighter camera on the entry point where identification actually happens.
Distance is the other half of the same decision. The rule of thumb worth internalizing: for identification, the distance in meters should be roughly under the focal length in millimeters multiplied by a factor set by sensor resolution. A 4 MP camera with a 4 mm lens is an observation tool past about four or five meters. The same body with an 8 mm lens doubles its usable identification distance while halving its field of view. When evaluating any security camera, the focal length on the specification line is the number that decides this, not the megapixels above it.
What to actually check before buying
Four questions, all answerable from the listing or one email:
- What is the focal length, and is a varifocal or alternate focal length available for this model?
- At what distance does the manufacturer claim face recognition, not just detection? If only a "night vision range" is given, that number describes IR illumination, not identification.
- Is the sensor ratio matched to the target, meaning a corridor gets a vertical composition rather than a wide one?
- Can the mounting height and angle place the face in the upper third of the frame, where the pixel budget actually is?
A face at the very bottom edge of a frame, partially cropped by the mounting angle, loses usable pixels to the geometry before the lens ever gets involved. Mounting is free; focal length is not.
The manufacturer-side version of the question
On the B2B side this arithmetic drives real cost decisions. Offering a scene as "covered" with detection-grade coverage when the customer's actual requirement is identification is the most common specification mismatch we see in project inquiries. A facility that wants license-plate capture at a gate needs a different lens and often a different mounting position than the same facility wanting general yard awareness, and the two needs are frequently quoted as one camera to save a line item.
This is why when we engineer camera projects as a CCTV camera manufacturer, the requirement conversation starts with the identification task and its distance, then works backward to lens and placement. The general product context lives at XingLiao, and project-level discussions with drawings start at our contact page.
The summary worth carrying away: pixels per meter at the target distance is the spec that predicts whether footage will be useful. Megapixels tell you the ceiling; focal length and distance decide where you land under it. A camera chosen for "seeing the yard" and a camera chosen for "identifying the person in the yard" are different purchases, and pretending one product is both is how the important recording turns out to be the useless one.
Top comments (0)