Three camera feeds do not make a production until someone can decide what is on air, hear the right sound, and recover when a field path disappears. This runbook treats 5G as the IP contribution path, then puts scene preparation, Program decisions, audio and configured output publishing in Callaba AV Mixer. The field team still owns the camera, uplink and local power; test the actual carrier and venue before relying on any path.
Give each camera a job before anyone goes live
The useful three-camera layout is editorial, not numerical. Assign Camera 1 to the wide safety picture, Camera 2 to the presenter or primary action, and Camera 3 to the reaction, detail, or roaming angle. A source with no decision role is usually the one that becomes confusing when a connection changes halfway through the event.
Name the feeds by role and location, then write down which scene can survive without each one. The producer needs that answer before the first transition, because a 5G path may reconnect later than a camera operator expects. The worksheet below is deliberately small enough to use on a call sheet.
| Source | Normal scene role | Audio decision | If absent |
|---|---|---|---|
| Camera 1: wide | Opening, reset, safe cover | Usually keep presenter or ambience by the planned Program mapping | Use Camera 2 close-up or a prepared two-up; do not cut to a black layer |
| Camera 2: primary | Presenter or main action | Confirm its microphone is intended before it reaches Program | Return to wide and hold the story with narration |
| Camera 3: detail | Reaction, cutaway, roaming detail | Normally do not make it the only intelligible speech path | Continue the main scene; tell the field team it is off-air |
For each camera, prove the contribution path with its real venue connection. Seeing a picture in a local phone preview is not the same as confirming the selected source is available to the operator. Use Multiview alongside the mixer when the producer must keep an eye on incoming feeds while preparing the next scene.
Build scenes, then separate Preview from Program
Create a wide scene, two single-camera scenes, and one recovery scene that contains only sources you have agreed are safe. Add sources as scene layers; keep the recovery scene free of the detail camera if it is the most mobile feed. This turns a fault into a known editorial choice instead of a frantic rebuild.
Put the opening scene in Preview while Program remains protected. Check framing, visible layers and the intended audio mapping there. When the new scene is ready, take it to Program with a CUT for an immediate editorial change or a FADE when the rundown calls for a transition. The distinction matters: Preview is a workbench; Program is what the audience receives.
Do one controlled rehearsal: wide to Camera 2, Camera 2 to Camera 3, then back to wide. After every TAKE, inspect Program rather than relying on the next item loaded in Preview. This catches source assignment mistakes that a source thumbnail alone cannot explain.
- Expected result: Program changes only after the operator takes the prepared scene.
- Producer check: the wide scene remains immediately available after every transition.
- Boundary: source transport and 5G quality are outside the mixer; a scene cannot repair a feed that never arrives.
Walk through the mixer flow
Use the public mixer demonstration to identify Preview, Program and scene flow. It is a view-only reference, not a place to create sources, issue invitations or publish an event.
- Locate the separate Preview and Program areas before rehearsing a TAKE.
- Identify where a prepared scene is moved to Program with CUT or FADE.
- Use an authenticated operator environment, not the public demo, for sources, audio, phone sessions and outputs.
- Verify the actual receiver after publishing a real rehearsal signal.
Watch the public mixer here; it is a view-only workspace.
Open Live mixer in a new tab ↗Loading public workspace…
Choose Program audio deliberately
Do not let a camera change silently become an audio change. Before the show, decide whether Program should carry a fixed presenter microphone, a mixed commentary path, or audio that follows a planned picture change. Apply the intended audio-mixer state to Program and listen on the operator monitor before publication begins.
A simple field rule helps: the camera used for detail should not be the only path carrying essential speech. If Camera 3 is a crowd or roaming shot, its natural sound can support the cut without being the explanation the audience must hear. If its microphone is the reporter's only mic, treat the camera and audio feed as a paired availability risk and have a return scene ready.
Record what the producer heard during the rehearsal: clean speech, expected ambience, no accidental duplicate microphone, and no audible change when a non-audio camera is taken. Those observations make the later incident decision faster than a vague instruction to “check audio.”
Start Program, then verify each configured destination
Program and publication are separate operational states. Start the configured output and verify the receiving endpoint with a short real signal: the correct scene, intended audio, and a current image. A green status in one control surface is useful, but it is not a substitute for a receiver-side confidence check.
That separation is helpful during a delivery issue. An operator can change the publish state of one configured output without replacing the mixer configuration, while Program continues to be produced. Decide in advance who owns that action and how they will tell the producer that the audience path, rather than the camera path, is the problem.
The public AV Mixer example is appropriate for observing the interface and its flow; treat it as view-only. Do not use a public demo as evidence that you can create sources, issue phone invitations, change an output, or operate a live production. Those are authenticated operator actions in the documented workflow.
Rehearse one lost-source decision
During rehearsal, have the Camera 3 team interrupt its contribution path on purpose. The producer should recognise the missing picture, keep a safe scene in Program, assign one person to the field check, and verify the returning feed before using it again.
- Keep the safe wide or primary scene in Program; do not take the failed source merely because it has reappeared in a tile.
- Use Multiview and the source context to establish whether the fault is limited to Camera 3 or affects the contribution path more broadly.
- Ask the field owner to confirm camera power, local network state and the contribution application or encoder without sharing credentials over comms.
- When the picture returns, load it in Preview, check framing and its intended audio relationship, then decide whether the editorial moment justifies a TAKE.
If a camera has two prepared SRT contributions, configure that pair before adding it to the mixer. On an empty, unfiltered SRT servers list, Create From Template → SRT Main/Backup Failover → Create this preset creates the source servers, routed output and Multiview. Use the failover output as the mixer source. In an authenticated operator's Multiview, open FINAL MIX → Open input menu → Source route to select backup or return to main, then verify the mixer input and Program. A public Multiview viewer cannot change Source route. On an existing setup, use its configured failover output; there is no need to recreate the servers.
Route selection changes which prepared SRT contribution supplies that input. CUT or FADE changes the scene in Program. Rehearse both decisions separately, and check the intended audio after each. If the drill produces a black picture or wrong-audio Program, change the source assignment or recovery scene before show day.
Next steps
Run this sequence first with the actual cameras, actual 5G venue conditions and intended receiver. Keep the completed source-assignment table with the event notes; it becomes the handoff record for the next producer. For the control model behind the exercise, review the AV Mixer workflow and the API reference before automating any part of it.