What Resolution Do You Actually Need? Why 8MP Doesn't Mean You'll Recognise the Face

Every camera spec leads with resolution, so that is what people compare. Then the incident happens, the footage comes back, and the intruder is an anonymous shape in a 4K recording. That is not a faulty camera. It is the most common and most expensive misunderstanding in CCTV: resolution is a property of the sensor, but identification is a property of the scene. What matters is not how many pixels the camera has — it is how many of them land on the thing you need to recognise. Last updated: October 2026. Pixels on target is the only spec that matters Spread 8 megapixels across a gate five metres away and a face fills a large part of the frame. Spread the same 8 megapixels across a 60-metre yard and that face is a handful of pixels wide. Same camera, same resolution, completely different outcome. The industry thinks about this as pixels per metre — how many horizontal pixels cover each metre of the scene at the distance you care about. The useful mental model is four bands: Monitor — enough to see that something is happening and how many people are involved. Low pixel density is fine. Detect — enough to know a person is there and trigger an alert. Recognise — enough to tell whether it is someone you know, if you already know them. Identify — enough to identify a stranger from the footage alone, which is what police and insurers need. Those four requirements differ by roughly an order of magnitude, and most disappointing installations were specified for one band while being expected to deliver another. The rule of thumb that saves the money Work backwards from the job, not forwards from the spec sheet. Decide, per camera, which of the four bands you need. Then work out how wide a scene that camera has to cover to do it. Narrow the field of view until the pixel density is high enough — and if it cannot be narrowed enough, you need a second camera, not a bigger sensor. That is the crucial point. Doubling resolution buys you a modest improvement in linear detail; halving the width of the scene buys you double. Framing beats megapixels, which is why our camera siting and mounting guide does more for footage quality than a resolution upgrade ever will. The two-camera pattern that actually works Most sites that get usable footage use cameras in pairs rather than trying to make one camera do everything. One overview camera, wide, low pixel density, tells the story: someone came over the fence at 02:14, went to the container, left with something. One identification camera, tightly framed on the pinch point everyone has to pass — the gate, the door, the only gap in the hedge — at head height, gets the face or the plate. Trying to do both with a single wide camera reliably gives you neither. If the pinch point is a vehicle entrance, a purpose-built plate-reading camera is the right tool for the identification half — see the ANPR guide. What quietly destroys resolution you paid for Night. Almost all serious incidents happen in the dark, and a sensor's daylight resolution tells you nothing about what it produces at 2am. A camera's low-light performance and its illumination matter more than its megapixel count — which is the whole argument for looking at thermal where detection at range is the goal. Compression. Footage is compressed to keep file sizes sane, and heavy compression smears exactly the fine detail that identification depends on. An 8MP stream squeezed into a low bitrate can carry less usable detail than a 4MP one given room to breathe. Motion blur. A moving person at a slow shutter speed is a smear at any resolution. Frame rate and shutter speed are part of the identification question, not separate from it. Bandwidth and storage. Higher resolution is more data — which off-grid means more power and, on a 4G site, more SIM cost. See which SIM data plan for that side of it. So what should you buy? For most UK sites: 4MP is the sweet spot for general coverage, and it is better spent on more cameras, tighter framing and better night performance than on jumping to 8MP for the same number of poorly framed views. Reserve the high-resolution cameras for wide scenes where you genuinely cannot get closer, and accept that even then you may need a second, tighter camera for identification. And be honest at spec stage about which band you need. "Identify a stranger across a 40-metre yard from one camera" is a requirement that no resolution on the market satisfies — knowing that before you buy is worth more than any spec sheet. Tell us what you need to see, how far away it is and whether it matters at night, and we will spec the framing as well as the camera — use the Solution Builder or request a quote. Trade pricing, UK stock, and we answer the phone ourselves.