A video intercom is only worth having if you can see the face

Usually. Two-wire digital bus platforms carry video, audio, power and control down a single pair, which is exactly why they dominate retrofit work in this city. Where the riser passes a proper ring-out, you can have a video door station and in-unit monitors or app access with nothing opened but the panel box.

Where it does not pass, the alternative that avoids a riser pull is a cloud panel with no in-unit hardware at all: one panel, one network drop, and video delivered to residents’ phones. That trades a capital cost for a permanent per-unit fee, and the right answer depends on how long the building intends to own the problem.

Because the camera is metering for the bright scene behind the visitor. A person standing in front of a glass entrance at midday is backlit by several stops, and a camera without true wide dynamic range cannot hold both the street and the face in one exposure.

True WDR captures multiple exposures at different settings and fuses them, which is a hardware capability. Digital WDR and backlight compensation are software approximations that help with moderate contrast and fail completely on a glass door in direct sun. If the existing panel has only digital WDR, the practical fixes are repositioning the camera so the glass is not directly behind the subject, adding front fill light, or replacing the door station with one that has real WDR.

The illuminator is being reflected back into the lens. In the order we find it: the dome or front cover is dirty, which is the most common cause by a distance; the dome bubble is not seated properly on its foam gasket, letting IR leak internally; a non-manufacturer bracket or housing is trapping IR inside the enclosure; the camera is tilted or yawed far enough that the illuminator bounces off its own body; or a white wall inside the IR cone is throwing the light back.

Fixes are cheap and mostly not hardware: clean the cover, re-seat the gasket, bring tilt under about 70° and yaw under about 50°, enable Smart IR so the LEDs throttle at close range, and where the entrance is glass, turn IR off entirely and light the area with white light instead.

Both work, and mixing them is normal. In-unit monitors are hardware the building owns and controls: they work without a resident’s phone, without push notifications and without anyone downloading anything, and they age at a predictable rate. An app costs nothing per unit in hardware, follows the resident when they are not home, and makes turnover a directory edit.

The practical answer in most buildings is monitors where residents want them and app access for everyone who prefers it, on a platform that supports both. What we push back on is app-only in a building where nobody has surveyed how many households will actually install it, because the units that do not are the units that will prop the front door.

Many can. Our default is that it does not, outside of the live call itself. New York is a one-party consent state for conversations you are part of, but an unattended microphone capturing conversations between other people in a lobby is a different legal question: eavesdropping under New York Penal Law §250.05 is a class E felony, and a lobby panel left recording all night is exactly the fact pattern that raises it.

Video recording at an entrance is ordinary and expected. Continuous audio recording of a common area is a decision for the owner or board with their own counsel, with signage and a retention policy, and it should be a deliberate written choice rather than a default we left enabled.

It depends entirely on the platform and the tier, and this is worth pinning down before you buy. Some hardware ships with a free cloud tier capped at a few dozen motion events; paid tiers commonly run around a week of high-definition footage. A panel integrated into a building video recorder inherits that recorder’s retention instead, which is usually far longer.

We put the real figure in the scope document, in days, at the resolution and frame rate you are actually going to run, not the headline number from the box. A building that plans on thirty days and discovers it had five has effectively bought nothing.

A wired system (analog bus, two-wire digital or local IP) keeps working, because the call never leaves the building. Residents at in-unit monitors answer the door exactly as before. That is the strongest argument for keeping hardware in the apartments.

A cloud, app-based system depends on the internet by design. Reputable platforms cache locally so existing credentials and codes keep opening the door, but calls to residents’ phones, remote unlock and any credential change stop until service returns. Get that behavior in writing from the vendor, put a UPS on the network closet, and make sure the building has a documented mechanical way in while it is down.