Pixel density is the number that decides whether footage is worth anything
Every camera specification leads with resolution, and resolution on its own tells you almost nothing. An 8 megapixel camera pointed down a street covers so much scene that a person at forty feet gets fewer pixels than a 2 megapixel camera watching a single doorway. What matters is not how many pixels the sensor has, it is how many of them land on the thing you care about. That figure is pixel density, usually written as pixels per foot, and it is the horizontal pixel count divided by the width of the scene in feet at the distance in question.
The scene width comes out of the lens. A lens with a horizontal field of view of sixty-five degrees at thirty feet frames roughly thirty-eight feet across. Put 1920 pixels across that and you have about fifty pixels per foot; put 3840 across it and you have a hundred. Fit a ninety degree lens on the same camera and the scene grows to about sixty feet, and the density falls by more than a third without anything else changing.
This is why the single most effective change to a disappointing camera is usually a longer lens rather than a higher-resolution body. Narrowing the view concentrates the pixels you already paid for.
Detect, recognise, identify, and why no figure here is a standard
The trade press and several standards bodies publish pixel-per-foot thresholds for three tasks: detecting that a person is present, recognising someone you already know, and identifying a stranger well enough to describe them. The figures are useful as a way of thinking and useless as a rule, for three reasons. They disagree with each other by a factor of two or more. They were derived under controlled lighting with a stationary subject. And they say nothing about the two things that ruin most real footage, which are motion blur at low shutter speeds after dark and compression artefacts on a scene full of moving foliage.
So this calculator gives you the density and describes the bands in words. If a specific number matters to you, put it in the target field and the calculator will tell you how far your lens holds it. The honest test is cheaper than any of this: have somebody stand on the spot, record it, and look at the file at full size on a real screen rather than on a phone.
Mounting height, tilt and the blind zone nobody plans for
A camera on a wall at ten feet looking at a person at thirty feet is tilted down about eight and a half degrees. Tilt it further to fill the frame with the near ground and the far end of the view lifts off the ground entirely. Tilt it less and the bottom of the frame stops meeting the ground until much further out.
That bottom edge is the blind zone, and it is a large number more often than people expect. A ten foot mount with a sixty-five degree lens aimed at thirty feet sees nothing on the ground inside about eighteen feet of its own post. Somebody standing at the door directly beneath it is not in the recording at all. The calculator prints that distance because it is the failure that gets discovered after an incident rather than before one.
| Change | Pixel density | Blind zone | What you give up |
|---|---|---|---|
| Longer lens, narrower view | Rises | Grows | Context around the subject |
| Wider lens | Falls | Shrinks | Any chance of identification at distance |
| Lower mounting height | Unchanged | Shrinks | Reach, sweep, and the camera itself is reachable |
| Higher mounting height | Unchanged | Grows | Faces — you record the tops of heads |
| More megapixels, same lens | Rises | Unchanged | Bitrate, storage and network load |
Where the camera points is a legal question, not only a technical one
A lens does not stop at your property line. A camera aimed to cover your driveway will very often also cover part of the street, the pavement and a slice of a neighbour's garden, and how that is treated varies enormously between states and sometimes between cities. Recording audio is the part that carries the sharpest risk: consent rules for audio recording are stricter than for video in many places, penalties can be criminal rather than civil, and a camera with a microphone pointed at a shared space is a genuinely different thing from one without. Areas with an expectation of privacy — bathrooms, bedrooms, a neighbour's windows, a shared hallway in some buildings — are treated differently again.
None of that is settled here, and nothing on this page should be read as telling you what the rules are where you live. Find out what your state and city actually say, ask your landlord or HOA if either applies to you, and take proper advice before installing anything that captures a space you do not control. Turning the microphone off, and aiming to keep the frame on your own ground, are the two cheapest ways to shrink the question.
Working the other way, from the spot to the lens
The productive order is to start with the spot rather than the camera. Decide the one place where a face has to be readable — the gate, the top of the drive, the step at the back door — measure the distance to it, choose the density you want, and let that pick the lens. Then check what the resulting field of view leaves out and decide whether a second camera covers it. One camera that answers a specific question beats four that produce wide grey footage of everything.
Once the lens and the resolution are settled, the next two constraints are the recorder and the network. Resolution drives bitrate, bitrate drives both the camera storage and retention calculator and the camera bandwidth calculator, and if the cameras are powered over the network cable the PoE power budget calculator is the one that decides whether the switch can actually run them. For the framing arithmetic on stills photography rather than surveillance, the field of view calculator covers sensor formats and crop factors in more depth, and the home security guide puts cameras in context against the door hardware and lighting that do more of the work.
Questions people ask
How many pixels per foot do I need to identify someone?
There is no number this page will hand you as a rule, because the published figures disagree with each other and were measured under conditions your driveway does not reproduce. What holds up is the shape of the relationship: low density tells you a person was there, middling density lets you recognise someone you already know from build and clothing, and high density is what it takes to describe a stranger from the footage alone. Pick a target, put it in the field, and the calculator tells you how far your lens holds it. Then verify by recording an actual person standing on the actual spot, because lighting after dark and motion blur will move the answer more than the pixel arithmetic does.
Is a 4K camera better than two 1080p cameras?
Usually not, for the same money and the same area. Two cameras with narrower lenses put more pixels on each subject than one wide camera does, they give you two angles so a face turned away from one is toward the other, and losing one does not blind the whole approach. What the single 4K camera wins is fewer cable runs, fewer switch ports and one licence rather than two. It also costs you four times the bitrate of a 1080p stream, which lands on storage and on network load. If the choice is between one high resolution camera on a wide lens and two moderate ones aimed at the two places that matter, the pair is nearly always the better recording.
Why can I not read a licence plate even though the car fills the frame?
Plates are a separate problem from faces and they defeat general-purpose cameras routinely. A plate is small, it is retroreflective, it is usually moving, and at night it sits directly beneath headlights that drive the exposure down. A camera doing general coverage will show you the car, the colour and the direction, and produce a white smear where the characters are. Reading plates reliably takes a camera set up for that one job: a narrow lens on a fixed lane, a short shutter to freeze motion, and exposure biased for the plate rather than the scene, which usually means the rest of the frame goes dark. Trying to get plates and general coverage out of one camera gets you neither.
How high should I mount a camera?
Lower than instinct suggests, wherever faces matter. Height keeps the camera out of reach and widens the sweep, and the cost is that you record the tops of heads and grow the blind zone directly beneath. Somewhere near the top of a doorway gives usable faces on anyone approaching it. Under the eaves at twelve or fourteen feet gives you a good record of movement across a yard and a poor one of who was moving. A workable split is a high camera for the wide view and a low one on the specific approach where identification matters. Whatever you choose, check the blind zone figure this calculator prints, because it is the part that surprises people.
Does the calculator account for digital zoom or cropping in the recorder?
Not directly, and the correction is easy to apply. Enter the pixel count that is actually recorded rather than what the sensor can produce, since many recorders are configured below the camera resolution and it is the stored frame that you will be looking at later. If you crop into the footage afterwards you are not gaining detail, only discarding the surrounding pixels, so the density on the subject is whatever it already was. Digital zoom is the same operation with a nicer name. Optical focal length is the only thing that genuinely moves pixels onto the subject.