A camera spec sheet is a marketing document that happens to contain engineering facts. Both halves matter. The facts are usually true. Their arrangement — what appears in bold at the top, what appears in six-point type in a footnote, what is quoted under conditions the vendor chose — is a sales decision.
Nobody is being dishonest. A datasheet is written to make a product look good to somebody skimming it, and the number that makes a camera look good to a skimming reader is megapixels. So megapixels go at the top, in bold, and the numbers that actually decide whether the camera solves your problem go further down.
This article is about reading the sheet in the order that matters rather than the order it is printed in. It assumes you have read what "enough resolution" actually means, because pixel density is what everything below feeds into: what decides whether you can identify someone is pixels landing on that person, not pixels existing on the sensor.
Read it in this order
Not top to bottom. Like this:
- Aspect ratio and sensor format — what shape the picture is, before you count anything.
- Focal length and the field of view it produces — where the pixels land.
- Aperture — what happens when the light goes.
- The conditions footnotes — which figures were measured the vendor's own way.
- Resolution — last, because by this point you know what it will be spread across.
Putting megapixels last is not a rhetorical trick. Once the other four are known, it is genuinely the least decision-relevant number on the page.
Sensor size: the number in the smallest type
Sensor size is usually printed as a fraction of an inch. That designation is a naming convention rather than a measurement — inherited from an older imaging technology, it does not describe any physical dimension you can put a ruler against. A larger fraction means a physically larger sensor, but you cannot convert the fraction into millimetres by arithmetic.
What the sensor's physical size actually governs is how much light-collecting area each pixel gets. Take a fixed sensor area and divide it into more pixels, and each pixel gets a smaller share of the light arriving. That is the trade nobody mentions when they sell you the higher megapixel count: at a given sensor size, more pixels means smaller pixels, and smaller pixels have less to work with when the light drops.
This is why a higher-resolution camera can produce visibly worse night footage than a lower-resolution one built on the same sensor size. The daytime image is sharper and the night image is noisier, and the sheet reports only the first of those as a headline. The honest question to a vendor is: for these two models, which has the larger sensor, and what does the night image look like at the spot I care about?
Focal length: the number that decides everything downstream
Focal length is the single most consequential number on the sheet, and it usually appears as a modest line in the lens section.
Focal length determines the field of view — how wide a slice of the world the camera sees. Shorter sees wider, longer sees narrower. That is the entire relationship, and everyone gets this far.
The second consequence is less commonly understood. Because field of view sets how many metres of scene are spread across the sensor's fixed pixel count, focal length is what determines pixel density on your target. The resolution number sets the budget. The focal length spends it.
A worked example, with assumed inputs — these are not any manufacturer's figures, they are arithmetic you can redo with your own numbers:
Assume a camera with 3840 pixels across the image, and assume that with the lens set to one end of its range the view is 24 m wide at the distance you care about. That is 3840 ÷ 24 = 160 pixels per metre. Now assume the same camera, same sensor, same 3840 pixels, with the lens set to the other end of its range so the view at that distance is 8 m wide. That is 3840 ÷ 8 = 480 pixels per metre — three times the density, from an identical camera, because of one setting.
Nothing about the sensor changed. It is the same camera on both lines of the arithmetic. The only thing that moved was where its pixels landed.
Varifocal is a range, not a value. If the sheet quotes a focal length range, the camera's field of view at installation is whatever the installer set it to on the day, which may or may not be the setting the design assumed. Ask what it was set to, and ask for that to be recorded per camera.
A fixed lens is a permanent decision. Fixed-lens cameras are cheaper and often perfectly correct, but the field of view is chosen at purchase and can only be changed by moving the camera or buying another.
Aperture: the f-number, and what it does not tell you
Aperture is quoted as an f-number — f/2.0, f/1.4 and so on, and those two are illustrative examples rather than any product's specification. The convention confuses almost everyone on first contact, so: a smaller f-number means a wider opening and more light reaching the sensor.
The reason it runs backwards is that the f-number is a ratio — the focal length divided by the diameter of the opening. Because it is a ratio rather than an absolute measurement, it does not by itself tell you how much total light lands on the sensor: a physically larger sensor behind the same ratio is collecting across a larger area.
On a varifocal lens, the f-number typically changes across the zoom range, and the sheet usually quotes the best end of it. A lens that is bright when zoomed wide can be noticeably dimmer at full zoom — which is exactly the setting you chose for your identification camera. Ask for the f-number at the focal length the camera will actually be set to.
Aperture is not the whole low-light story. Night performance combines aperture, pixel size, sensor sensitivity, noise-reduction processing, and the shutter speed the camera decides to run at. That last one is the trap: a camera can produce a bright, clean-looking night image by holding the shutter open longer, and the cost is that anything moving smears. A still night image is not evidence of a usable night image.
Aspect ratio: why the same megapixel count is not the same picture
Aspect ratio is the shape of the image — 16:9, 4:3, 1:1 and so on. It gets almost no attention and it changes what a given megapixel count buys you.
Megapixels are a total: pixels across, multiplied by pixels down. That total can be arranged in different shapes. A wide shape spends its budget on horizontal reach; a taller shape spends more of it vertically.
This matters because the thing you usually care about — a person — is taller than they are wide. In a corridor, an aisle, a stairwell or a narrow alley, a wide image spends a large share of its pixels on the two walls either side of the space, which contain nothing you will ever need. Some cameras address this with a corridor or vertical mode that rotates the image so the tall dimension runs along the tall space, recovering a share of an otherwise wasted budget without buying anything.
So when a sheet says a camera does a given number of megapixels, ask the follow-up: at what aspect ratio, and is that the shape of the space I am pointing it at? Watch also for the case where the maximum megapixel count is available at only one aspect ratio, so selecting a different shape reduces the total. That detail lives in the mode table, not the headline.
This is the part of the sheet most likely to mislead, and it does so without containing a single false statement.
Several headline figures are measured under conditions the manufacturer selects and states in a footnote:
- Minimum illumination, quoted in lux. The figure depends on the shutter speed used, what the vendor considered an acceptable image, and whether the infrared illuminator was on. A very low lux figure taken with a slow shutter and IR active describes a different situation than a higher figure taken under stricter conditions.
- Wide dynamic range, quoted in decibels. The test scene and method behind that number are the vendor's.
- Frame rate, frequently a maximum available at one particular resolution, codec and feature set. Turning on analytics can reduce it.
- IR range, quoted in metres — how far the illuminator throws light, not how far you can identify a person.
We are not aware of an openly available method that all vendors are obliged to use when stating minimum illumination, which is why we will not tell you that any two lux figures are comparable. If a vendor says their figures follow a standard method, ask which document, and read the footnote yourself.
For the application side — how much pixel density a given task needs — the international standard is IEC 62676-4, Edition 2.0. It is paywalled and we have not read it, so we will not tell you what is in it. What is public: Axis publishes pixel-density figures written against that edition, which is what gives the focal length arithmetic above a target to be measured against.
The general principle: compare within a vendor freely, and across vendors only with the footnotes side by side.
The questions a spec sheet cannot answer
A spec sheet describes a camera. It does not describe your building. These are the questions no datasheet can settle, which is what a site visit is for:
- What is the actual distance from the mounting position to the spot that matters? The arithmetic above needs this number as an input, and nobody can guess it from an address.
- What will be behind the subject at the worst hour of the day? A camera aimed at a glass entrance is fighting the sun at some point every day. No specification predicts that; standing there at that hour does.
- What is the mounting geometry? Height and angle decide whether the camera sees a face or the top of a head, and vendors publish geometry limits when they know the analytic depends on it. Axis, documenting its own plate-reading product, states that "the camera's mounting angle should not be larger than 30 degrees in any direction" and that "the image of the license plate should not tilt more than 5 degrees horizontally", alongside a requirement that "the width of the license plate equals at least 130 pixels for license plates with one row and at least 70 pixels for license plates with two rows". Those are Axis's numbers for Axis's analytic, not a universal rule — but they show the shape of the constraint, and no general camera datasheet tells you whether your wall can satisfy it.
- What is actually lighting that spot at 2am? Not the lighting design. The lit reality, including the lamp nobody has replaced.
- What is the cable path, and does it already exist? The largest cost variable on most projects is not the camera.
- What happens to the footage afterwards — how long it is kept, who can retrieve it, how quickly. A perfect image in a recorder nobody can operate has solved nothing.
The through-line is simple enough. A spec sheet tells you what the camera is capable of under conditions someone else chose. Only a measurement at your site tells you what it will do under yours. Ask for the second in writing, and treat the first as what it is — a starting point, printed by someone who wanted you to buy it.
