Almost every incident anyone shows us happened after dark, and almost every disappointing system looks excellent at noon. Night performance is the real specification, and it is the one least examined before purchase.
The short answer is yes, CCTV works at night — but the gap between “records something” and “identifies someone” is much wider in darkness than in daylight, and that gap is where the money goes.
Night is the test that matters
A camera has two jobs, and darkness separates them ruthlessly. Detection is knowing that a person crossed the frame. Identification is being able to say who. In daylight, most cameras do both. At night, many do only the first, and the owner does not find out until the night they need the second.
So when comparing systems, the useful question is not “does it have night vision” — they all claim it — but “at what distance can it still identify a face after dark”.
How infrared actually works
Most cameras switch to infrared mode in low light: an IR filter moves aside, LEDs around the lens flood the scene with light the human eye cannot see, and the image becomes monochrome. Three consequences follow, and all three matter:
- You lose colour. Clothing colour, vehicle colour and skin tone all disappear. In a dispute or a police report, that is often the detail that mattered.
- Illumination falls off sharply with distance. IR obeys the same physics as any light source, so brightness drops roughly with the square of the distance. Doubling the distance needs about four times the light.
- Anything close and reflective blows out. A wall, a gate post or a nameplate near the lens reflects IR straight back and washes the whole frame, leaving the subject beyond it in darkness. This is the single most common cause of useless night footage.
Why the range on the box is optimistic
An IR range figure — 30m, 50m, 80m — describes the distance at which the LEDs put some light on a subject under favourable conditions. It is not the distance at which you can identify a face, and it assumes nothing in between reflecting light back.
As a working rule, treat useful identification distance as a fraction of the quoted IR range, and treat detection as most of it. A camera advertised at 30m will typically identify a face at a small part of that, and tell you “a person walked past” for most of it. Neither number is a lie; they answer different questions.
Rain, fog and dust all cut it further, because airborne particles scatter IR back into the lens — which is also why a camera that is fine in April can be poor in July.
When a light beats a better camera
Frequently. Adding modest white light to an entry point often does more for footage quality than upgrading the camera, and it costs less. Three reasons:
- The camera stays in colour mode, so you keep clothing and vehicle colour.
- Visible light is itself a deterrent in a way invisible IR is not.
- A lit approach avoids the IR-reflection problem entirely.
The trade-off is that a light has to be positioned so it illuminates the subject rather than shining toward the lens, and it needs to be acceptable to neighbours and residents. On a society gate or a house porch this is usually easy. On a plant perimeter it needs planning.
Where added light is not an option, the honest answer is a camera specified for low light rather than one specified for megapixels — larger sensor, better lens aperture. Those cost more and they are worth it exactly where light cannot be added.
How to test night performance before you pay
Do not accept a demonstration in daylight. Ask for one of these instead:
- A night sample from a site like yours, at the distance you care about — not a brochure image.
- A live view after dark from the proposed position during the survey, with somebody standing where a visitor would stand. Can you recognise them on the phone screen? That is your answer.
- A recorded clip rather than a live view. Recording compresses, and compression hurts noisy night images far more than clean daylight ones. Live view can look better than what gets stored.
Any installer who will not show you a night clip is asking you to buy the half of the specification that never gets tested.
Common questions
Do colour night vision cameras work?
Yes, when there is some ambient light to work with - a street lamp, a porch light, spill from a window. They use a larger sensor and a wider aperture rather than infrared, so they keep colour. In genuinely pitch-dark conditions with no ambient light at all, they fall back to infrared or to a built-in white light like anything else.
Why does my camera look foggy only at night?
Usually condensation inside or on the dome, or a spider web across the lens catching the infrared. Both are invisible in daylight and both scatter IR straight back into the lens. Cleaning the dome and moving the camera away from an adjacent light that attracts insects fixes most cases.
Is a higher megapixel camera better at night?
Often the opposite. Packing more pixels onto the same size sensor makes each pixel smaller, and smaller pixels gather less light. A well-chosen 2MP camera can outperform a 5MP one after dark on the same scene. Sensor size, lens aperture and lighting matter more than the megapixel figure.