Cameras: the field decides, not the resolution

The security article has already ranked this line: a camera answers "what happened", after the fact, and it comes fourth. We are not reopening that ranking.

This page handles what the ranking does not say: why a correctly mounted installation, with recent equipment, still produces unusable images on the day they are needed.

The answer nearly always comes down to three things, and none of them is the camera’s definition. The field covered, the light at the moment that matters, and the recorder — which is this page’s critical device long before the camera is.

We add a fourth subject specific to this country and absent from most documentation: the physical place the camera sits in, the dust, heat and water it will take there, and the rating that genuinely corresponds to that place.

Nearly every installation is signed off during the day, and nearly every event that justifies a camera happens at night. It is this page’s most expensive mismatch and it is entirely avoidable.

What changes at night is not only the amount of light. The camera switches to night mode, the image goes monochrome, the infrared illumination comes on, and motion sensitivity changes completely. It is a different camera.

Three faults appear only in that mode. The infrared reflects off glass, a nearby pale wall or a gutter, and washes out the whole image. Insects drawn to the infrared cross in front of the lens and trigger recordings all night. And a light source in the field — a streetlamp, a shop sign — saturates the sensor and makes everything else black.

The check is simple and nobody runs it: come back at ten in the evening, on an ordinary day, and look at each camera. What can be fixed that evening — an angle, a shade, a ten-centimetre move — cannot be once the installer has gone.

The line to write into the quotation is this: handover done at night, or a second night visit included in the price. It is a line that costs almost nothing and that we have never seen offered unprompted.

Quotations, comparisons and conversations are about the cameras. The device that genuinely decides the installation’s worth is the recorder, and it gets treated as an accessory.

Three reasons. It holds the images, so it is what gets carried away during a theft — and an installation with its recorder sitting in plain view in the stockroom was designed to prove nothing.

It is what fills up, and its behaviour when full is never checked: either it overwrites the oldest, which is correct, or it stops, which is much less so and happens with badly made configurations.

And it is what nobody has ever tried to extract a sequence from. A third party’s request for a recording — insurer, authority, lawyer — comes in a specific format, and an installation from which nobody has ever exported a minute discovers the problem at the worst moment.

The three corrections fit in a sentence each. The recorder in a locked room, not in the room it films. A copy of the important sequences elsewhere, like any data in the pillar. And an extraction tested once, in front of you, at handover.

A network camera is powered by its cable, which removes the need for a socket at height and accounts for most of the installation saving. It is also what makes it dependent on equipment described elsewhere.

The page on switches gives the figure that matters: the power available per port depends on the generation of the standard, and a switch’s overall budget is almost always below the sum of its ports. We do not repeat it here, we draw its consequence.

The consequence is a symptom to recognise, because it does not look like a power problem: the cameras work while there are few of them, then one restarts on its own, always around the same time of day, when the others are drawing most.

The cameras that draw most are the ones heating their window, the ones that pan, and the ones with powerful infrared illumination — which is to say precisely the outdoor cameras, in winter, at night. The worst moment coincides with the moment they are useful.

So the purchasing rule is to add up the real consumption stated by the manufacturer rather than counting cameras, and to keep a margin. It is the same rule as for ports, and it fails the same way: nobody checks it because everything works on installation day.

What we will refuse: installing a camera that covers a surface rather than a passing point, because that is the installation that produces eight cameras and no recognition, and because it sells better.

We will refuse to sign off an installation without a night check and without an extraction tested in front of you. An installation accepted in daylight is one tested in the conditions where it will not be used.

We will refuse to place a recorder in the room it films, and to open its remote viewing while its factory password has not been changed and recorded in your access inventory.

What we do: the passing points drawn before any quotation; framing checked with somebody standing in the field; the angle chosen against backlight before proposing a more expensive camera; the protection rating chosen by the real place, washing included; retention decided and the disk sized for it; consumption added up rather than cameras counted; and section 9’s five checks done before the installer leaves.

And what you can do tonight without us: walk past your cameras at ten in the evening and look at the live images. Then ask to see a recording from three weeks ago. Those two acts take ten minutes and will tell you what you actually bought.