Pick a certified camera or enter resolution and field of view; the calculator shows how many pixels per metre you get at a distance, and how far the camera can detect, observe, recognise or identify a person on the DORI scale of IEC 62676-4:2014 (25, 62.5, 125 and 250 px/m) — the scale Indian datasheets still print.
For counting machine idle time and people at a station, Observe (62.5 px/m) is usually enough; reading a label or a face needs Recognise or Identify. A wider lens covers more floor but each metre gets fewer pixels. Certificate status for each preset is in the certified camera finder.
Want a camera plan for your floor? Mama will get back to you.
What you can see at a given distance depends on three numbers: the pixels across the frame, the horizontal angle of view and the distance. A 2 MP camera with a 1/2.8″ sensor and a 4 mm lens has an angle of view of about 70°, so at 20 m it sees a strip about 27.9 m wide and spreads its 1,920 pixels to about 69 pixels per metre. That reaches the DORI "Observe" level (62.5 px/m) but not "Recognise" (125 px/m). For Recognise at 20 m the same camera needs a lens of at least 7.3 mm, an angle of view of 42° or narrower.
Last checked: 16 Oct 2026.
What the DORI levels mean and why datasheet distances run long is explained in DORI (IEC 62676-4) Explained.
The calculator above takes the angle of view, because that is what an Indian datasheet prints. If you only know the lens, convert it with the sensor width. The formulas (rectilinear lens model; the angle-of-view formula is the standard one on Wikipedia's "Angle of view" page):
Horizontal angle of view = 2 × arctan(sensor width ÷ (2 × focal length))
Scene width at the distance (m) = 2 × distance × tan(angle of view ÷ 2)
Pixel density (px/m) = horizontal pixels ÷ scene width
Minimum focal length (mm) = required px/m × distance × sensor width ÷ horizontal pixels
Pixels across a face = px/m × 0.16 (a 16 cm face, the basis used in Axis's summary of IEC 62676-4:2014)
| Sensor type | Width used | Source |
|---|---|---|
| 1/2.7″ | 5.37 mm | nominal width, Wikipedia "Image sensor format" |
| 1/2.8″ | 5.57 mm | Sony IMX307, a 1/2.8″ 2 MP security sensor: 1,920 recording pixels × 2.9 µm (Sony flyer) |
| 1/3″ | 4.80 mm | nominal width, Wikipedia "Image sensor format" |
The "inch" label is a naming convention left over from video camera tubes, not a measurement, so two sensors with the same label can differ slightly in width. If the datasheet gives the pixel size, multiply it by the horizontal pixel count for an exact width.
The 2014 edition of IEC 62676-4 sets the pixel densities below. The values are confirmed by the foreword of the 2025 edition, which lists the old scale it replaces (public preview of EN IEC 62676-4:2025 at the Slovak standards office). The definitions are Axis's wording of the 2014 standard.
| Level | px/m | px across a 16 cm face | What it lets you do (Axis wording) |
|---|---|---|---|
| Detect | 25 | 4 | "determine whether or not any individual is present" |
| Observe | 62.5 | 10 | "determine how many people are present and see characteristic details" |
| Recognise | 125 | 20 | "determine whether an individual shown is the same as someone seen before" |
| Identify | 250 | 40 | "identify an individual" |
IEC 62676-4 edition 2.0 was published on 9 October 2025 (IEC webstore); its foreword says it "cancels and replaces the first edition published in 2014" and moves to a new seven-step scale (20, 40, 80, 125, 250, 500 and 1,500 px/m). Indian camera datasheets still print DORI distances on the 2014 scale, so the calculator uses it.
A factory wants to see who walks through a gate 20 m from a camera mounted on the security cabin. The camera is a 2 MP bullet with a 1/2.8″ sensor and a 4 mm lens. The distance and the camera are an assumption for illustration. No rupee inputs are needed: the answer is in metres and millimetres.
| Input or result | Value | Basis |
|---|---|---|
| Horizontal pixels | 1,920 | 2 MP stream |
| Angle of view | 70° | 2 × arctan(5.57 ÷ 8) for a 4 mm lens on a 5.57 mm sensor |
| Distance | 20 m | assumption: cabin to gate |
| Level wanted | Recognise, 125 px/m | assumption: "is this the same person we saw before?" |
| Scene width at 20 m | 27.9 m | 2 × 20 × tan 35° |
| Pixel density | 69 px/m | 1,920 ÷ 27.9 |
| Pixels across a face | 11 px | 69 × 0.16 |
| Level reached | Observe | 69 is above 62.5 but below 125 |
| Farthest Recognise distance | 11.0 m | 1,920 ÷ (125 × 2 × tan 35°) |
| Minimum focal length for Recognise at 20 m | 7.3 mm | 125 × 20 × 5.57 ÷ 1,920 = 7.25 |
The 4 mm camera would show that someone came through the gate and roughly what they looked like, but not reliably who. Three ways out: a varifocal lens zoomed to 7.3 mm or more (the gate then fills a 15.4 m wide strip), a second camera closer to the gate, or a higher-resolution camera.
Lens choice at 20 m for the same 2 MP, 1/2.8″ camera:
| Lens | Angle of view | Scene width at 20 m | px/m | Level reached |
|---|---|---|---|---|
| 2.8 mm | 90° | 39.8 m | 48 | Detect |
| 3.6 mm | 75° | 30.9 m | 62 | Detect (just short of Observe) |
| 4 mm | 70° | 27.9 m | 69 | Observe |
| 6 mm | 50° | 18.6 m | 103 | Observe |
| 8 mm | 38° | 13.9 m | 138 | Recognise |
| 12 mm | 26° | 9.3 m | 207 | Recognise |
Widest angle of view that still reaches each level at 20 m:
| Camera | Detect | Observe | Recognise | Identify |
|---|---|---|---|---|
| 2 MP (1,920 px) | 125° | 75° | 42° | 22° |
| 4 MP (2,560 px) | 137° | 91° | 54° | 29° |
| 8 MP (3,840 px) | 151° | 114° | 75° | 42° |
Doubling the horizontal pixels lets you keep a wider view for the same level.
A real check: the CP Plus CP-UNC-TA21L3C-Q is a 2 MP camera with a 1/2.8″ sensor and a 3.6 mm lens. Its datasheet gives a horizontal angle of view of 86° and a Recognise distance of 10.5 m.
Wide CCTV lenses have barrel distortion, so their real angle of view is wider than the rectilinear formula gives, and, as Wikipedia puts it, "calculations for lenses producing non-rectilinear images are much more complex". For 2.8–4 mm lenses, use the angle of view printed on the datasheet rather than one computed from the lens. For 6 mm and longer, the formula and the datasheet are much closer.
How do I calculate the focal length of a CCTV lens for a given distance? Multiply the pixel density you need (px/m) by the distance (m) and the sensor width (mm), then divide by the camera's horizontal pixels. For Recognise (125 px/m) at 20 m on a 2 MP camera with a 5.57 mm wide 1/2.8″ sensor: 125 × 20 × 5.57 ÷ 1,920 = 7.25, so at least a 7.3 mm lens.
How far can a 4 mm CCTV camera see a face? On a 2 MP, 1/2.8″ camera (about 70° angle of view), the formula gives Recognise (125 px/m) out to about 11 m and Identify (250 px/m) out to about 5.5 m. The wide-angle distortion of short lenses makes the real distances shorter, so check against the datasheet angle of view.
What is the difference between Detect, Observe, Recognise and Identify? They are pixel densities from IEC 62676-4:2014: 25, 62.5, 125 and 250 pixels per metre at the target. Detect tells you someone is there; Identify is enough to say who it is. The 2025 edition of the standard replaces them with a seven-step scale, but Indian datasheets still quote the 2014 levels.
Is a higher-resolution camera better than a longer lens? For the same level, doubling the horizontal pixels halves the focal length you need, so you keep a wider view. A 4 MP camera reaches Recognise at 20 m with a 5.5 mm lens where a 2 MP camera needs 7.3 mm. The trade-off is more storage and bandwidth per camera.
Tell Mama about your plant and Mama will come back with how many cameras you need, where they go, and the retention your state expects — sized to your floor, in ₹. Built for Indian factories.