Resolution isn't absolute, it's spread across the scene
“4 MP” or “8 MP” on the box tells you how many pixels the whole frame has, not how many land on the person you're actually watching. The same camera gives a completely different result depending on how many metres of scene that frame covers — a narrower scene means more pixels per metre, a wider one means fewer, regardless of the stated resolution.
Rule of thumb
The same rule used for choosing an entrance camera applies to judging distance: detection needs roughly 25 pixels per metre of scene width, recognition about 125, identification about 250. This is a rough ratio, not a guarantee — lens quality, light and compression all shift the final result.
Pixels per metre of scene width
- Detection
- ~25 px/m
- Recognition
- ~125 px/m
- Identification
- ~250 px/m
Example: the same camera at 20, 40 and 80 metres
Take a 4 MP camera with a horizontal resolution of roughly 2688 pixels. Across a 20 m scene that's about 134 pixels per metre — above the threshold for recognition. Across a 40 m scene, the same pixel count is spread over double the width, about 67 px/m — below recognition, above detection. Across an 80 m scene it drops to about 34 px/m — barely above the detection threshold.
This is an illustrative calculation, not a field measurement — it shows the principle, not a guaranteed result for every camera and every lens.
Illustrative example — 4 MP, ~2688 px frame width
- 20 m scene
- ≈ 134 px/m — recognition
- 40 m scene
- ≈ 67 px/m — between detection and recognition
- 80 m scene
- ≈ 34 px/m — detection threshold
The lens changes the outcome more than the megapixel count
A narrower lens (a longer focal length) concentrates the same pixel count onto a smaller field of view — effectively zooming the resolution in on the target. The same 4 MP camera with a telephoto lens instead of a wide one can turn detection into identification at the same distance, with no change to the sensor or the camera's price.
Which is why “how many megapixels do I need” has no real answer without also asking “with which lens.”
Recommendation
Before choosing a camera, set the job for each position (detection, recognition or identification) and the actual scene width at that position — measured, not estimated. Only then do resolution and lens make sense as a pair rather than as separate specs. The exact figure gets set on site, tape measure in hand.