B-Roll

Definition

B-roll is supplementary footage cut in over or between the main footage of a video. The main footage is the A-roll — typically the person speaking on camera — and the B-roll does three jobs at once: it shows what the speaker is describing, it hides the join where the A-roll was cut, and it varies the picture so a long take does not go static. In its most common form the B-roll replaces only the picture while the A-roll's audio keeps running underneath, so the viewer hears an uninterrupted sentence over a changed image. The name comes from 16mm film finishing, where shots were split across two reels — the A-roll and the B-roll — and printed in a checkerboard pattern so the physical splices, which were visible on stock that narrow, fell on black leader instead of on the picture.

Where the Name Came From

A cement splice needs an overlap, and on stock as narrow as 16mm that overlap intruded into the projected image — you could see every edit. 35mm was wide enough to hide the join in the frame line, so the problem belonged to the small gauge alone.

The fix was to stop splicing shots to each other. The negative cutter built two conformed rolls — odd-numbered shots on the A-roll, even-numbered on the B-roll, black leader filling every gap so wherever one roll carried picture the other carried black, interlocking like a checkerboard. The print stock is exposed twice, once against each roll, and every splice lands on leader instead of on an image. Overlapping the rolls at a join is also how the lab produced a dissolve, which is why the setup outlived the problem it was invented for. So the B is a reel number, not a grade.

The Two Operations People Both Call “Adding B-Roll”

This decides whether your video gets longer, and most explanations skip it. A cover swaps the picture for a window of time and leaves the audio and the runtime alone. A splice puts the clip in as its own scene, pausing the speech and extending the video. Both are B-roll. Only one is what you want when the speaker says “here is the app I built” and you need to show it without interrupting the sentence.

A second precision: a cutaway leaves the scene entirely, while an insert shot is a detail from inside the same scene at a different angle or focal length — the hands on the keyboard the speaker is sitting at. Editors use “B-roll” as the umbrella for both, but the difference matters when covering a cut: an insert preserves the sense of continuous time, while a true cutaway resets it and lets you remove far more of the A-roll underneath.

The Rule of When to Cut Away

1. Cut away when the words go concrete — on the word, not after it. If the speaker names something photographable, show it. Landing half a second late is the most common placement error and reads as sloppiness even to viewers who cannot say why. Word boundaries in normal speech are roughly 200–400ms apart, so this is a real target, not a vague one.
2. Cut away to hide a join you had to make. Every sentence deleted from a locked-off single-camera take leaves a jump cut. Covering the join is the original reason the technique exists.
3. Cut back before the point lands. Never cover a punchline, a reveal, or a moment of direct address to camera. When the content of the shot is the person's face, the face is the shot.
4. Do not cover what the words do not motivate. A generic city timelapse under a paragraph about hiring is called wallpaper, and it reads as filler because it is. If you cannot say in one clause why this image is over these words, cut it.
5. End it on purpose. Cut back on a beat in the narration, not when the clip runs out. A cover that ends because the source ended is visible; one that ends on a comma is not.

Density matters as much as placement. One to three purposeful cutaways per minute reads as edited; above that the piece turns into a montage and the speaker stops being present in their own video.

Sourcing and Shooting It

• Coverage ratio. Film at least three usable seconds for every second that reaches the screen. A five-minute program with 60–90 seconds of covers wants 4–6 minutes of usable B-roll — usually 15+ minutes rolled, because shots get lost to focus, shake and continuity.
• Hold each shot 10 seconds or more. You will use three. You need clean handles at both ends and a stretch in the middle where the camera has settled and nothing enters or leaves frame.
• Shoot sequences, not singles. The five-shot method — hands, face, wide, over-the-shoulder, one unusual angle — gives you shots that cut to each other. Five related shots of one activity beat fifteen unrelated pretty ones.
• Shoot MOS, deliberately. B-roll usually carries no usable sync sound because the A-roll is carrying the sentence underneath. That is the format, not a defect — but atmos, if you want it, is a separate recording.
• Shoot at double the timeline frame rate. 48 or 60fps against a 24 or 30fps program gives a genuine 50% slow-down with real frames. Slowing 30fps footage in post duplicates frames instead, and duplicated frames judder on motion.
• Match the A-roll or the cut announces itself. White balance, exposure and lens character carry across the join. Same-shoot footage matches by default, which is why self-shot B-roll beats stock even when the stock is prettier.

When you have none of it, do not invent it. A punch-in on the A-roll hides a join using only the footage you already have, which is why zooms and punch-ins and B-roll get planned together rather than as alternatives.

Why It Matters: One Take, Forty Cuts

You record a 12-minute talking-head straight to camera, one angle, tripod locked off, then strip the dead air, the filler words and three sentences you fluffed. That is easily 40 removals, and because the camera never moved, every one is a jump cut: the head teleports a few centimetres and the hands change position between consecutive frames.

There are four ways to deal with those joins: leave them (the native grammar of a lot of short-form), put a transition on each (which turns a talk into a slideshow within about six of them), punch in on alternating cuts, or cover them with B-roll. Do the arithmetic before choosing — covering all 40 at three seconds each needs two minutes of footage and means the viewer barely sees the speaker. The professional answer is the mix: cover the eight or ten joins that fall mid-sentence and are therefore most visible, punch in on the ones at a paragraph break, leave the rest as honest cuts. That triage, not the shooting, is where the editing time goes.

Common Mistakes and Misconceptions

How This Works in Valmera

Valmera is an agentic video editor: you describe the edit and an AI agent performs it. B-roll is where the two operations above become an explicit choice the agent has to make. A cover switches the picture for a window of program time while the speech, the music and the burned-in captions keep running — runtime unchanged. A splice inserts the clip as its own scene and adds time. The agent picks by whether the sentence should continue, and tells you which moments it covered and with what. The step-by-step version, including what to type, is on add b-roll to a video automatically.

Placement is word-accurate because the agent works from a word-level transcript of the current edit, not of the raw recording — so “show the dashboard when I mention it” resolves to the program time of that word after your cuts rather than an estimate. Sources: your own uploads, a pasted link, a stock search, an AI-generated still with a slow push on it, a short generated clip, or a recording of a live web page. Covers read best at 2–6 seconds. More in the docs on text and overlays and AI generation.

The real limits. A cover clip's own audio does not play — cover footage is silent by design, so it cannot bring in location atmos. Covers render above the footage but below the captions, so captions stay readable over B-roll; if you wanted the cutaway to obscure them, it will not. Nothing motion-tracks: an overlay holds its position and does not follow a moving subject underneath. Spliced clips are not captioned. And there is no true crossfade or dissolve between A-roll and B-roll — the change is a hard cut, the right default for this move but a genuine limit if you wanted a soft transition.

Related Terms

• Jump cut — the visible join B-roll is most often used to hide.
• J-cut and L-cut — offsetting the audio edit from the picture edit, which makes a cutaway land softly.
• Word-level timestamps — what lets a cover start on the word rather than near it.
• Silence detection — where the cuts that need covering come from.
• Shot detection — finding the usable shots inside a long clip of raw B-roll.
• Edit decision list — the record of which source ranges play when, covers included.
• Color grading — matching B-roll to the A-roll so the join stays quiet.
• Aspect ratio — why landscape B-roll in a 9:16 edit gets cropped or padded.
• All glossary terms

Frequently Asked Questions

Cover Your Cuts by Describing Them

Upload your footage, say which moments to cut away on, and the agent places the covers on the word. 50 free credits, no card.