Aiming a camera for the right level of detail comes down to one number: how many pixels land on your target. The European standard EN 62676-4 codifies this as DORI — Detect, Observe, Recognize, Identify — and ties each goal to a minimum pixel density on the subject: roughly 25 pixels per meter to detect a person is present, 62 to observe behavior, 125 to recognize someone you already know, and 250 to identify a stranger from footage alone. Aiming is the act of choosing a lens, mounting height, and angle so that your target sits inside the part of the camera field of view where pixel density meets the task. Get that geometry right and a modest camera produces court-usable footage; get it wrong and an expensive multi-sensor unit gives you blurry shapes. Here is the step-by-step method we use on every design.
Start With the Task, Not the Camera
Before anyone touches a lens calculator, write down what each camera must accomplish at a specific spot. "Identify a face at the lobby turnstile." "Recognize a known employee at the loading dock." "Detect movement across the back fence line." Those verbs map directly to DORI tiers, and the tier sets the pixel-density target you have to hit.
The mistake we see most often is buying resolution first and discovering coverage second. A 4K sensor spread across a 60-foot-wide parking aisle delivers fewer pixels on a face than a 1080p camera covering a single doorway. Detail is a function of pixels on the target, not megapixels on the spec sheet. Decide the job, then engineer the optics to serve it.
Translate DORI Into Pixels on Target
Convert each goal into a working number. For identification, target at least 250 px/m across the subject; for recognition, ~125 px/m; for observation, ~62 px/m; for detection, ~25 px/m. A standing adult is about 1.7 m tall and roughly 0.5 m wide at the shoulders, so a face — the thing that actually identifies someone — occupies a small slice of the frame.
Practically: take your camera's horizontal pixel count and divide by the width of the scene at the distance where the target will be. A 1920-pixel-wide image covering a 7.7 m scene yields about 250 px/m — enough to identify across that plane. Cover 15 m with the same camera and you drop to ~125 px/m, which is recognition territory, not identification. This single division is the most useful calculation in surveillance design, and it is why two cameras with identical resolution can produce wildly different evidence.
Choose Lens, Distance, and Mounting Height Together
These three variables move as a set:
- Lens focal length sets the angle of view. A wider lens captures more scene but spreads pixels thin; a longer lens concentrates pixels for distant identification but narrows coverage. Use a manufacturer lens calculator or field-of-view chart to find the focal length that puts your target zone at the required px/m.
- Distance is fixed by where people actually pass. Place the camera so the required DORI zone falls within usable range, accepting that areas beyond it degrade to lower tiers.
- Mounting height controls angle. Mount too high and you photograph the tops of heads — useless for identification. For face capture, keep the vertical angle shallow, generally under about 15 degrees of downtilt, and mount at a height that frames faces, not scalps. Higher mounts resist tampering but trade away identification; that is a deliberate choice, not an accident.
A varifocal or motorized-zoom lens buys you tolerance here, letting installers fine-tune the camera field of view on site instead of swapping hardware when reality differs from the drawing.
Aim for Choke Points, Not Open Space
You almost never need identification across an entire area — you need it where everyone is forced to pass through a narrow, predictable lane: doorways, turnstiles, gates, stairwells, register lanes, elevator lobbies. Aim a tight, high-density camera at the choke point for identification, and use a separate wider camera for situational awareness and detection across the open space.
This pairing — one "evidence" camera and one "context" camera — is far more reliable than asking a single fisheye or panoramic unit to do both. Panoramic sensors are excellent for detection and overview, but their per-meter pixel density at the edges rarely reaches identification. Know what each device is for and stop expecting one camera to cover every tier.
Validate in Daylight and Darkness
A design that hits 250 px/m on paper can still fail in practice. Walk-test every critical camera: have a person stand at the target distance and confirm you can genuinely recognize or identify them on the recorded stream, not the live preview. Then test the conditions that break footage:
- Backlight and sun position. A camera aimed at a glass entrance can be blinded at certain hours; wide dynamic range helps, but angle matters more.
- Low light and IR. Identification at night needs adequate, even illumination. Infrared washes out at distance and can flatten facial detail, so verify after dark, not just at noon.
- Motion blur. Fast movement at a low shutter speed smears the exact frame you need. Confirm faces stay crisp on people in motion, not standing still.
Record these walk-tests as commissioning evidence. They are also exactly what an auditor or investigator will want when footage is challenged.
Build Compliance Into the Aiming Plan
Aiming is a design decision, and design is where compliance is cheapest to get right. Every camera you mount becomes part of a system that may be audited against NDAA Section 889 and TAA requirements — and provenance does not depend on where the lens points, but on whose hardware you installed. There is no point engineering a flawless identification zone on a camera whose country of origin or covered-entity status can't be confirmed.
We work vendor-neutral, so the camera that best serves a given DORI target gets specified on its merits — then checked against the prohibited-source rules before it ever reaches a purchase order. The same field-of-view documentation that proves your lobby camera can identify a face also feeds the as-built record that proves the system is compliant, supported across its lifecycle, and ready for the next audit. Aiming, procurement, and provenance are one continuous workflow, not three disconnected steps.
Get the Geometry Right the First Time
DORI is simple math, but translating it into mounted hardware across a real building — with backlight, varifocal trade-offs, choke points, and compliance constraints — is where designs succeed or quietly fail. If you want a coverage plan that hits identification where it matters and stays Section 889-clean end to end, request a quote and we'll engineer the field-of-view design with you.