Architecture of a Live Multicamera Production.

A live multicamera production rarely fails because of camera quality. It fails because of the architecture: nobody drew where each signal comes from, who watches it or what happens if a link goes down. The setup works in rehearsal, then falls apart four hours into the live production as the switcher heats up and someone asks about camera three’s return.

Work in the opposite order to the usual approach: first the signal map, then the equipment list. If you know how many outputs you need and what they are for, the switcher, converters and task allocation almost choose themselves. Start with equipment, and you will end up with excess cabling, duplicated sources and last-minute decisions in front of the audience.

The Signal Chain of a Production

Almost any setup, from a three-camera concert to a ten-camera event, can be reduced to one line:

Camera → transport → switcher → program → encoder → platform

Every arrow is a technical decision and a point where something can degrade. Capture determines resolution, frame rate and latency. Transport determines how far you can go and the margin for error, with SDI, HDMI, fibre and video over IP all playing a part. The switcher mixes, applies keyers and generates every output the rest of the team will need. Program is the signal ready to go out. The encoder compresses it, and the platform delivers it to the viewer.

Draw that line on paper before setup, and write the format and cable count above each arrow. Most problems in a live production appear on an arrow nobody drew: a converter without a backup, a distribution system that ran out of channels, or a return that was never tested.

PGM, PVW and Aux Outputs

Program (PGM) is the signal on air at this instant, with transitions, titles and layers already composited. Everything the audience needs comes from it, so PGM also feeds the main recording.

Preview (PVW) is the same kind of signal without an audience: it shows the source you have prepared before cutting to it. It lets you verify framing, focus, exposure and titles before the source enters PGM. Working without PVW means guessing.

Aux outputs are the most frequently forgotten part. Each aux output is an independent copy you can assign to any switcher input, including a specific camera or multiview itself. They serve venue screens, a guest’s monitor, a press room or a sponsor’s feed. If your switcher has few aux outputs, your architecture has a low ceiling: this is the first number worth over-specifying.

How Many Outputs You Need

Count destinations before equipment: broadcast, recording, venue screens, guest monitors, press and sponsor confidence monitoring. Every destination other than the live broadcast needs its own aux or distribution path.

Clean Feed and Dirty Feed

A clean feed is the clean program signal: no titles, channel logo, keyer layers or burnt-in graphics. Its value lies in what it omits. Use it when the signal passes through another switcher that adds its own graphics, or when supplying press, a client or a screen where titles are unnecessary.

A dirty feed is the complete program, with all graphics and layers already composited. It is what the audience sees and what usually goes into the show archive.

Deciding which feed to send to each destination is an architectural choice. Sending dirty to a second switcher guarantees duplicated titles that cannot be removed. Sending clean to the final broadcast leaves the program without information. Define it by destination and write it on the signal map, because nobody will have time to reason it through during the live production.

ISO Feeds, Multiview and Recording

ISO feeds are individual outputs from each source: one camera per signal, unmixed. They are not needed for broadcasting, but allow you to rework the production afterwards, more calmly, once the live event has ended. Recording PGM plus each ISO is the difference between an archive and editable material.

Multiview is the supervision grid: every source, PGM and PVW on one screen, with audio and status indicators. It is the director’s working tool. A well-configured multiview reduces cutting mistakes more than any improvement in picture quality.

Reserve one aux output for multiview and another for recording, so changes to program configuration do not affect the record. If you record ISOs, use multiple drives rather than the same volume running the switcher software.

Return Feeds and Confidence Monitor

Return feeds travel in the opposite direction: from the switcher to cameras and workstations. A return usually carries PGM, PVW, intercom and sometimes another position’s signal. It lets the camera operator see what is being broadcast and make informed framing or exposure adjustments.

A confidence monitor is the screen showing that return at a particular position: sound booth, stage manager, presenter or guest area. Its role is to confirm that the correct signal arrives and coordination is on track, rather than to monitor picture quality. A wrongly assigned confidence monitor causes contradictory alerts and arguments during the live production.

Test each return before starting the broadcast. A missing return is discovered at the worst moment: the director asks for a shot change, and the camera does not respond because the operator never heard it.

Tally and Intercom

Tally is the status indication telling each camera whether it is in PGM, in PVW or off air. It travels through the same return or a dedicated channel, and appears as a red or green light on the camera monitor. Without tally, the operator does not know when the camera is on air or when it can move.

Intercom is the other half of the system: the voice channel connecting the director, camera operators, sound, lighting and stage manager. It is organised into channels or group lines and usually includes IFB to the presenter or host. With the equipment map in front of you, allocating positions is easier to understand in team roles in a multicamera production.

Design intercom with a dedicated camera channel and another for the rest of production, even if it initially seems excessive. Sharing one channel makes framing instructions overlap sound instructions, multiplying confusion exactly when time is shortest.

Redundancy and Single Points of Failure

Go through the signal map and mark every point whose failure stops the broadcast: a single converter, distribution without a second output, a network switch without a backup, or a multiview feeding both supervision and recording. Those are the single points of failure.

A sound architecture duplicates only a little, but duplicates the critical parts: local PGM recording, the switcher’s power supply and the signal path to the encoder. If the outgoing connection drops, the broadcast is lost but the show is not: the recording and ISOs continue saving material. That detail, decided beforehand, lets you publish the highlights the next day.

Common Architecture Mistakes

• Choosing the switcher by input count and forgetting how many aux outputs it supports.
• Combining supervision, recording and broadcasting software on the same device.
• Failing to put PVW on a visible multiview and cutting blind.
• Sending a dirty feed to a destination that already adds its own graphics.
• Improvising intercom and discovering the problem with the first urgent instruction.
• Failing to test returns, tally and confidence monitors before starting the broadcast.

Each mistake can be corrected on paper in five minutes, but costs a missed shot live. Architecture is cheap to fix beforehand and extremely expensive to fix during the production.

Next Step

With the signal map settled, the next job is the link feeding everything else: the cameras. Continue to cameras for live production: choosing, configuring and matching, which covers choosing bodies and lenses by position and matching different cameras so cuts are unobtrusive. If your budget also depends on available bandwidth, first review encoding, bitrate and multistream: the number of ISOs you choose to record and the quality of the program signal determine what the encoder can deliver.