AV Networking and ITGUIDE · LESSON 42

Video over IP

01 — Definition

Video over IP moves pictures across the same kind of Ethernet network as everything else, and the whole subject turns on a single trade-off: compression against bandwidth. Uncompressed video keeps full quality with minimal latency, but it is enormous — a single uncompressed 1080p stream runs at around 3 gigabits per second. That is why professional uncompressed IP video, standardised as SMPTE ST 2110, demands 10-gigabit-and-up infrastructure and careful engineering. Compression shrinks that dramatically: visually lossless schemes like JPEG-XS and NDI keep latency low while using only a fraction of the bandwidth, and codecs like H.264 and H.265 squeeze it further still, at the cost of adding more latency.

Video over IP
Photo : Lambtron, CC BY-SA 4.0, via Wikimedia Commons
02 — The guide

The discipline this asks of you is matching bandwidth to the medium honestly. A handful of NDI streams fit comfortably onto gigabit, but the same number of uncompressed feeds will not — and oversubscribing the link drops packets, which in video terms means tearing, freezing, and blocky artifacts rather than a graceful fade.

So the planning comes first, before the gear. Count the streams, multiply by their real bitrate, add some headroom, and then choose the network class — 1G, 10G, 25G — to suit the total. It is the available bandwidth that defines what is actually possible, not the badge on the equipment.

03 — The rule

Bandwidth defines what's possible — not the gear.