Guide8 min read

What Is B-Roll, and How Do You Make It With AI?

B-roll is the cutaway footage that covers a talking track. Learn the five jobs it does, how to turn a script into a shot list, and how to generate it with AI.

By VdoBloom Team

B-roll is the cutaway footage that plays while a narrator talks: the hands on the product, the street outside, the coffee being poured. A-roll is the main talking footage. You can generate B-roll with AI by turning your script into a shot list of 3 to 8 second cutaways, prompting one clip per shot, and laying them over the talking track in the Timeline Editor. A 5-second Seedance 2 Fast clip at 480p costs 24 credits, so a full set of cutaways can be drafted before you pay for finished quality.

Updated October 2026.

A-roll versus B-roll in two sentences

A-roll is the footage that carries the story: a person speaking to camera, an interview, a presenter walking through a demonstration. B-roll is everything cut in around it to show what the words describe, cover an edit, or set a scene.

Five jobs B-roll does

B-roll is not decoration. Each cutaway should do one of five things, and naming the job makes the shot easier to prompt.

  • Cover a cut. When the talking footage jumps, a cutaway hides the seam.
  • Show what you describe. If the narrator says “the lid clicks shut”, the viewer wants to see the lid. The cutaway turns a claim into evidence.
  • Set a place. A wide shot of a street, a workshop or a kitchen tells the viewer where the story happens without a line of explanation.
  • Add pace. A sequence of short cutaways changes the rhythm of a long monologue and gives the eye something new to track.
  • Hide a jump cut. When a pause is removed, the picture jumps. Placing B-roll over the join makes the edit invisible.

Name the job for every shot; one that does none is slowing the video down.

Turn your script into a shot list of 3 to 8 second cutaways

Read the script one line at a time and ask what the viewer should see while that line is spoken. Write one shot per idea, not one shot per sentence: a 10-second paragraph may need two cutaways, and a single vivid sentence may deserve four seconds on its own.

A practical format is three fields: the line it covers, the subject, and the camera. For example: covers “the grip on a wet street”, subject “close-up of a shoe stepping into a puddle”, camera “low angle, slow motion”.

Keep each cutaway between 3 and 8 seconds. Shorter shots feel restless, and longer ones stop being cutaways. The set should also vary: alternate wide, medium and close shots so the sequence does not look like the same frame repeated.

Prompt patterns for establishing, detail and insert shots

Three shot types cover most of a B-roll list. Each has a reusable prompt pattern.

  • Establishing shot: [place] + [time of day] + [weather or light] + [camera move]. Example prompt to try: “A quiet residential street at golden hour, long shadows, a slow push forward, no people.”
  • Detail shot: [object] + [specific texture or action] + [lens feel]. Example prompt to try: “Macro close-up of a hand tightening a metal clasp on a canvas bag, shallow depth of field, soft window light.”
  • Insert shot: [action that supports the line] + [framing]. Example prompt to try: “Overhead shot of a notebook being opened on a wooden desk, a pen placed beside it, natural light.”
  • Transition shot: [movement through frame] + [direction]. Example prompt to try: “A train passing left to right across a city bridge, motion blur, overcast sky.”
  • Product-in-use shot: [person + product + action] + [setting]. Example prompt to try: “A cyclist tightening a water bottle into a frame mount, close on the hands, morning light, shallow focus.”

These are prompts to try, not outputs. Each one names a subject, an action and a camera behaviour; that structure is what keeps a generated clip from drifting into a generic montage. For more on matching a style across shots, see our guide to turning a blog post into a video.

Match look, lens and aspect across clips, and draft cheap before finishing

B-roll only works if the shots feel like they belong to the same video. Reuse the same language across prompts: the same light, the same colour palette, the same lens feel. If one shot is warm and shallow and the next is cold and wide, the cut reads as a mistake rather than a style.

Draft the whole set at a low resolution, watch the sequence, then finish only the shots that survive. The price difference is large enough to matter.

Setting5-second clipUse it for
Seedance 2 Fast, 480p24 creditsDrafting the whole shot list
Seedance 2 Fast, 720p50 creditsDrafts that may make the final cut
Seedance 2, 720p82 creditsThe shots you will actually use
Kling 3.0, standard5.6 credits per secondDrafting when you want a different look
Kling 3.0, pro7.2 credits per secondFinishing a shot that needs more detail

Ten draft shots at 480p is 240 credits, less than three finished clips. That is the right order for B-roll: decide what the sequence needs before spending on the shots themselves. Run the whole list as a batch; our guide to AI video batch generation covers that workflow.

Lay the clips over your talking track in the Timeline Editor

Open the Timeline Editor and put the talking footage on the video track. The editor is a sequence editor with one video track and two audio tracks, so B-roll is laid in as clips in the sequence rather than stacked on top of the talking video. You cut to the B-roll clip while your voice track continues underneath, then cut back.

The workflow is:

  1. Place the talking clip on the video track and the voice audio on an audio track.
  2. Split the talking clip where the first cutaway should start.
  3. Insert the B-roll clip on the video track at that point, trimmed to the length you need.
  4. Split again at the end of the B-roll and return to the talking footage.
  5. Repeat for each shot in the list, then add music and render.

A project holds up to 30 clips and 180 seconds, enough for about a dozen cutaways. For a faceless video where there is no talking footage at all, see faceless YouTube channels and the Faceless Channel Maker writes the voice-over script and generates cinematic B-roll from a topic, so the whole sequence starts in one place. For generating individual shots, use text to video, or image to video when you have a product photo or a frame to animate.

Licensing note for generated footage

Generated footage is not automatically cleared for every use. The terms of the model that produced the clip govern how the output can be used, and those terms can differ between models. Before you put a generated shot in a paid ad, read the terms for that model and check whether your use is covered.

If a real person, a real logo or a recognisable location appears in the footage, there is a second question about rights that has nothing to do with the model. Do not generate a likeness of a real person without their consent, and do not place a real brand in an ad without permission. Our notes on AI video commercial rights cover the questions to ask. This is not legal advice.

Render costs and generation costs are shown before you confirm, and a failed generation is refunded. Credit packs are one-time purchases that never expire, so a B-roll batch does not lock you into a monthly plan. Prices are on the pricing page.

Frequently asked questions

What is the difference between A-roll and B-roll?

A-roll is the main footage that carries the story, usually a person talking. B-roll is the cutaway footage that shows what is described, covers an edit or sets a scene.

How long should a B-roll clip be?

Usually 3 to 8 seconds. Shorter feels restless; longer stops functioning as a cutaway.

How many cutaways does a video need?

Roughly one per spoken idea, not one per sentence. A 60-second video often needs 8 to 15 shots, depending on pace.

Can I put B-roll on top of my talking video?

You cut to the B-roll clip in the sequence while the voice track continues. The editor has one video track, so it does not composite video over video.

How much does a B-roll shot cost?

A 5-second Seedance 2 Fast clip at 480p is 24 credits, at 720p it is 50, and a finished Seedance 2 clip at 720p is 82. The price is shown before you generate.

Open text to video with the first shot from your list, or start in the Faceless Channel Maker if you want a script and a B-roll sequence from one topic. New accounts start with 10 free credits and no card required, so you can draft a couple of cutaways and see how they sit under your voice before committing to the full list.

Ready to try it?

Create your first AI video in minutes — no credit card required.

Start Creating Free →