The question to answer is not "how many megapixels?" but "how many pixels land on the subject?". That depends on three things together: the camera's resolution, the lens focal length (which sets the field of view) and the distance to the subject.
Think in pixel density
Pixel density is the number of horizontal pixels per metre of scene at the subject's distance. Divide the image width in pixels by the width of the scene in metres. A 2560-pixel-wide image covering a 10 m wide entrance gives 256 pixels per metre. The same camera covering a 40 m wide car park gives only 64.
A common way to describe what you need is the observation level, often called DORI:
| Level | What you can tell | Typical pixel density |
|---|---|---|
| Detection | Something or someone is there | About 25 px/m |
| Observation | What a person is doing, clothing colour | About 60 px/m |
| Recognition | Whether it is a person you know | About 125 px/m |
| Identification | Who a stranger is, beyond reasonable doubt | About 250 px/m |
Many camera datasheets, including ours, list DORI distances for exactly this reason. Use them as a starting point and check against your own site.
Choosing the focal length
A short focal length (for example 2.8 mm) gives a wide view with less detail at distance. A long focal length (for example 12 mm) gives a narrow view with more detail far away. A vari-focal or motorised zoom lens lets you adjust between the two on site. If one camera has to do both jobs, it is usually better to add a second camera than to compromise both views.
Resolution trade-offs
- Bandwidth and storage rise with resolution. Efficient compression such as H.265 helps a lot.
- Low light: packing more pixels onto a small sensor can reduce night-time performance. Compare minimum illumination figures and look at sensor size.
- Frame rate: some cameras reduce frame rate at maximum resolution. Check the frame-rate table in the specifications.
- Display: an 8MP stream only shows its full detail when you zoom in or view it on a large screen.
Quick rules of thumb
- Doorways and counters where you need faces: aim for identification density at the point people pass.
- Perimeters and yards: detection or observation is often enough, paired with analytics that flag people and vehicles.
- Number plates need a narrow view, a fast shutter and good lighting, not just high resolution.
- When unsure, choose a motorised zoom lens so the view can be adjusted after mounting.
Want a second opinion on a layout? Send us the distances and widths and we will suggest suitable models.