Two cameras, lined up by what you said

How the cameras are synced

The second camera is matched to the main one by its words. Both files are transcribed, and the matching words give one steady offset between them, say camera B started 4.2 seconds after camera A, for the whole recording. You don't need a clap, a slate or timecode.

That steady offset is also how a second camera is told apart from a second take. A second camera hears the same words at the same offset all the way through. A second take drifts, because nobody says a script at exactly the same pace twice. Keep each camera rolling from start to finish so one offset fits the whole file.

How angles are chosen

The switching follows a few plain rules:

• At jump cuts. Where a pause or filler was cut, switching camera hides the jump instead of letting your head pop sideways.
• At the start of a sentence after about 7 seconds on one angle, so a long stretch without cuts still gets a change.
• Never sooner than 2 seconds after the last switch, so it doesn't flicker.
• Never to an angle that isn't rolling at that moment or shows no face.

Sound follows the picture

Right now the sound comes from whichever camera is on screen. That has a practical effect: if camera B is across the room with its built-in mic, the audio gets roomier every time the edit cuts to B. Put both cameras close to the speaker, or give each one a decent mic, so the switch is invisible to the ear.

How to use it

1. 01Set up two cameras on the same speaker, for example a wide shot and a closer shot, and start both before anyone speaks.
2. 02Import both files into one Talking head project.
3. 03When the analysis finishes, the Cuts tab shows Cameras with the number it lined up. Leave "Switch angles" on.
4. 04Play it through. Pauses, fillers and retakes are cut as usual, with angle changes at the cuts.
5. 05Export. Every camera's footage stays on your device the whole time.

What it won't do yet

It doesn't pick the camera of whoever is speaking. In a two-person interview with one camera on each person, it still switches on the rules above, which means it can show the listener while the other person talks. For a conversation, a single wide shot of both people is the safer setup for now.

It doesn't keep one microphone across angles, as covered above.

It needs both cameras recording the same performance. Two takes filmed one after another are treated as takes and go to the best take picker instead.

When another approach is better

For an interview where the shot must follow the speaker, use a desktop editor with a multicam timeline, or film one wide camera here. For a single presenter, a second angle mostly earns its keep by hiding jump cuts, and the punch-in option does a similar job with one camera: it alternates a slight zoom so cuts within one take don't jump.