Wide - considerably wider than intuition suggests, because visitors stand close and the useful information is often at the edges.

A person pressing a button stands within arm's length of the panel. A narrow lens at that distance frames a portion of a torso or a face cropped at the chin, which is unhelpful.

A wide angle captures the whole person, what they are carrying, and the area around them - including whether somebody else is standing to one side, which is exactly the thing an occupant wants to know.

The trade is distortion at the edges and less detail on any one part of the image, which is why resolution and field of view have to be considered together.

Mounting height interacts with this. A station mounted too high looks down at the top of a head; too low and it looks up. Where accessibility requirements set the button height, the camera position follows, and a station with an adjustable or wide-angle lens compensates.

Check the actual image from the intended position before finalising - a mock-up costs nothing and prevents an unsatisfactory result.

With wide dynamic range - and without it, a visitor against bright sky is a silhouette, which is the commonest complaint about door entry video.

An entrance frequently has the visitor between the camera and a bright background: open sky, a sunlit street, a glazed lobby. A camera exposing for the overall scene renders the face as a dark shape.

Wide dynamic range combines multiple exposures or uses sensors capable of capturing a wide brightness range, so both the bright background and the shaded face are rendered. It is the single most valuable feature in a door station camera and should be treated as a requirement rather than an upgrade.

Orientation helps. A station facing north avoids direct sun; one facing east or west will have the sun behind visitors for part of every day.

A canopy or recessed position reduces direct sun on the lens, which also prevents flare and protects the lens cover from ultraviolet degradation over the years.

Infrared for a discreet monochrome image, white light where colour matters - and in both cases the illumination must suit the very short distance involved.

Infrared is invisible, does not disturb residents or neighbours, and gives a clear monochrome image. It is the usual choice.

White light gives colour, which helps identify clothing and vehicles, and it also deters. It is more intrusive and can be objectionable at a residential entrance at night.

The common failing with either is overexposure. An illuminator sized for several metres, pointed at a face half a metre away, produces a white blur against a black background. Door stations need illumination matched to the close range, and good ones adjust the output automatically.

Reflection is the other problem: illumination bouncing off a glossy panel surface, a glass door or wet ground back into the lens washes out the image.

Test at night from the actual position rather than accepting the specification.

Many systems capture an image or a short clip per call, and some integrate with video recording - which brings the installation within data protection obligations.

Call snapshots are useful: a record of who called while nobody was in, and evidence after an incident.

Where images of identifiable people are recorded, data protection law applies. The requirements differ by jurisdiction but the common elements are a lawful basis for recording, signage informing people that recording takes place, a defined retention period, restricted access, and a means for individuals to exercise their rights.

In multi-occupancy buildings there is an additional question of who controls the recordings - the management company or the individual residents - and it should be settled in writing rather than assumed.

Cameras that overlook public space or a neighbour's property attract particular scrutiny, and the field of view should be set to avoid it where possible.

Establish the position before commissioning, since retrofitting compliance is harder than designing for it.

Most current systems offer it, and it is one of the main reasons people replace an older system - but it introduces dependencies worth understanding.

App integration forwards the call to a smartphone, so the door can be answered from anywhere and deliveries handled when nobody is home.

The dependencies are a working internet connection at the building, the manufacturer's cloud service, and push notification services on the phone. Any of those failing means calls do not arrive - and a call that silently fails to notify is worse than no app at all.

Most systems keep the indoor units working independently of the app, which is the arrangement to insist on.

Ongoing cost and service longevity matter: some manufacturers charge a subscription, and a cloud service withdrawn after a few years leaves the app function dead in hardware that is otherwise fine.

Ask about the service commitment and whether the system remains fully functional without it.

Enough to identify a face at the door, which is less than a general surveillance camera needs - dynamic range and illumination matter more.

The subject is close and fills much of the frame, so the pixels available per face are high even at modest resolutions. High definition is comfortably sufficient for identification at door distances.

What spends that resolution is a very wide field of view, so the two specifications should be considered together - a very wide lens at low resolution gives few pixels on the face.

Beyond a certain point, extra resolution is limited by the system rather than the camera: two-wire systems constrain the video they can carry, and the indoor monitor's own screen may be lower resolution than the camera.

Spend the budget on dynamic range and illumination first. A moderate-resolution image that is correctly exposed identifies a visitor; a high-resolution silhouette does not.

Robust housing, recessed lens, tamper-resistant fixings and a considered position - and the lens cover is the vulnerable part.

Stainless or die-cast housings with flush buttons resist casual damage. Plastic panels in exposed public positions do not.

The lens cover is what gets attacked or scratched, so a recessed lens behind a toughened cover, ideally set back so it cannot easily be reached or sprayed, is worthwhile.

Security fixings prevent removal, and a tamper alarm reporting to the system provides notice of interference.

Position contributes as much as the product. A station within view of other cameras, in a lit position, and not in a concealed recess, attracts less attention.

Where the risk is high, consider a separate protective enclosure or an alternative position - and keep a spare, since a damaged station in a shared entrance is an urgent problem for every resident rather than an inconvenience for one.