How to Add B-Roll to a Talking Head Video

The reason b-roll feels hard is that most advice describes what it is rather than what it is for. Here is what it is for: it buys you a cut.

A cut you do not have to justify

When you cut from your face to your face, the viewer notices. That is a jump cut, and it reads as an edit, which is fine but it is visible. When you cut from your face to a shot of the thing you are describing, the viewer does not notice at all, because the cut is motivated by what you just said.

That is the whole mechanism. B-roll hides edits, covers the moment you fumbled a sentence, and lets you compress twenty seconds of explanation into eight without the seams showing. Everything else people say about b-roll follows from this.

What b-roll is and how to use it covers the vocabulary; this is the placement.

Where it goes

Not "every fifteen seconds". B-roll goes where your words became visual and stayed there for a moment:

• You named a physical object
• You described a place
• You gave a number or a process worth showing
• You said something you are about to demonstrate

And it does not go where you are making an argument, expressing an opinion, or looking directly at the viewer to say something that matters. Those moments need your face. Cutting away from a sincere line to a stock shot of a laptop is how a video stops being yours.

How long to hold it

Two to four seconds, and cut back before the sentence finishes.

The mistake is holding b-roll until the topic ends. What happens is that the viewer forgets there is a person in the video, and the piece flattens into narration over footage. Coming back to your face slightly early keeps the shot feeling like an aside rather than a detour.

If a single b-roll shot is running past six seconds, it is not b-roll any more, it is the video, and it needs to earn that on its own terms.

Doing it automatically on a Mac

Zella has an Add B-Roll button in the editor sidebar. It fetches matching stock video clips from Pexels and drops them onto the timeline based on your captions, so the placement follows the words rather than a timer. Tap it again and every clip it added comes back off.

Two requirements, both real:

• Captions first. The matching runs off the caption text. No transcript, nothing to match.
• A free Pexels key. You bring your own, from the Pexels API. The clips come to you directly rather than through anyone else's server, and the Pexels licence covers commercial use.

Add B-Roll is a Mac button. On iPhone you place clips yourself from the media bin, which is more work but the same three rules above still decide where they go.

Keep the audio underneath

The single most common b-roll mistake is letting the stock clip bring its own sound.

Your voice should run continuously underneath. The picture changes, the audio does not. That is what makes it read as one continuous thought rather than two clips glued together. In practice this means muting the b-roll clip entirely, which is what the automatic placement does.

The exception is a clip where the sound IS the point, and in that case you are not cutting away, you are cutting TO something, which is a different edit.

Making it not look like a stock reel

Automatic placement gets you the timing right. What makes it look intentional is a small amount of taste applied afterwards:

Cut some of them. If eight clips got placed, four of them are probably better than eight. The ones to cut are the literal ones: you said "email" and got a shot of someone typing. Nobody needed that.

Match the grade. Stock footage is usually brighter and more saturated than a room recording. Pulling it toward your own footage is a thirty second job and it is the difference between "b-roll" and "a different video briefly". See how to colour grade a video.

Vary the shot size. Four consecutive wide shots read as filler. Wide, then close, then wide reads as coverage.

When to skip it entirely

A talking head video with no b-roll and a good hook outperforms a mediocre one with a lot of it. If you are adding b-roll because the video feels slow, the problem is usually pacing, and the fix is cutting the dead air rather than covering it up.

B-roll makes a good video smoother. It does not make a slow video fast.

Common questions

How long should a b-roll shot be? Two to four seconds, and cut back to your face before the sentence finishes. Past about six seconds it stops being an aside and becomes the video.

Should b-roll keep its own audio? Almost never. Mute it and let your voice run continuously underneath, so the picture changes and the audio does not. The exception is when the sound is the reason you cut to it.

Do I need a Pexels key to add b-roll? Only for the automatic fetch on Mac. A free key from the Pexels API is enough. You can always place your own footage without any key.