B-roll for Talking-Head Videos: Show the Point

What should B-roll do?

B-roll is supplemental footage supporting the main recording, often called A-roll. In a talking-head video, your face and voice usually carry the explanation. A close-up, screen capture or other cutaway can show the thing you are describing while your voice continues.

Adobe’s filmmaking guide describes supporting footage as a way to establish context, connect shots and add meaning. For a short creator video, translate that into a concrete question: what becomes easier to understand when this image appears?

An atmosphere shot can still be useful. A quiet room, workshop or kitchen can establish where the story happens. The test is whether that context matters. If the speaker is explaining a spreadsheet formula, a stock city skyline probably does less work than the spreadsheet.

Keep the speaker on screen when the human response is the information: a reaction, a direct admission or a carefully qualified opinion. You do not need another visual for every sentence. Sometimes the clearest image is the person explaining the point.

Match a shot to a sentence

After selecting your main take, read the transcript once and mark only the sentences that call for a visual. Write the visual beside the sentence before searching for footage. That small constraint changes the search from “something about productivity” to “the exact calendar block being moved.”

These are authored teaching examples, not a filmed product test. For your own video, use the real object or interface when the claim is specific. A stock image can illustrate a general setting, but it cannot prove that your workflow, product or result happened.

Give each candidate a sentence-level purpose. “Shows the cable connection” is useful. “Looks cinematic” can be an aesthetic reason, but it does not explain the instructional value. If you cannot describe the purpose, try the passage with the speaker left on screen and compare the clarity.

Build a simple shot plan

Here is a constructed plan for a 45-second video about recording clearer speech at home. Start with the spoken explanation, then select four visual moments. The timing is a planning example, not a benchmark for how often every creator should cut.

The cutaways occupy 14 seconds: 3 + 4 + 4 + 3. Their share of the 45-second video is about 31.1%. That is a description of this plan, not an ideal B-roll ratio. If the comparison takes longer to understand, give it longer; if a face carries the point better, leave that passage on the speaker.

Keep your example honest. A comparison should use your actual before-and-after audio and explain what changed. If you have not recorded it yet, keep it as a planned shot rather than showing an invented result. If you use a waveform graphic, label it as an illustration unless it represents the real recordings.

The worksheet adds provenance and permission columns to the shot plan. That lets you see which ideas are ready, which need recording and which are waiting on the right to use an asset. A saved link to an attractive clip is not yet a finished editorial decision.

Record useful footage in one session

You can build a small useful media library without filming a separate movie every week. Record the objects and actions that recur in your explanations: the setup, the tool, the screen, the hands performing the task and the finished example.

• Record the context. A wider view establishes where the action happens.
• Record the action. Show the step from a view that makes it repeatable.
• Record the detail. Capture the small element the wider frame cannot explain.
• Leave editing room. Hold the shot before and after the action so you can choose a clean start and finish.
• Name the files usefully. Identify the object, action and version so you can find them again.

For a screen recording, enlarge the relevant control, close unrelated windows and use demonstration data. Check the result at the size it will appear on a phone. A desktop interface that looks clear on your monitor may become unreadable in a narrow vertical video.

Record enough time to show the action, but trim to the time the explanation needs. Do not accelerate an instructional step so much that someone cannot follow it. If a single shot tries to explain several actions, split it into separate views with clear labels.

Choose stock and generated visuals carefully

Use stock footage when it genuinely illustrates a general concept and you have the rights for your intended use. Search for an action and setting rather than an abstract category. “Hands placing a microphone on a desk” is a more useful request than “creator success.”

Check the selected provider’s current license, the asset source and any restrictions. For example, the Pexels license includes limits on uses such as implying endorsement by a person or brand. Save the asset URL and your rights notes with the project. The download being available does not establish every right you might need.

Generated footage needs the same editorial judgment. It can illustrate a hypothetical scene when clearly presented that way. It should not stand in for evidence of a customer result, an actual interface or an event you are reporting. Watch for incorrect controls, distorted objects and details that conflict with the narration.

In EditorOP, you can use your own uploaded supporting footage and review stock suggestions. Replace any suggestion that only matches the keyword but misses the meaning. Your footage is especially useful when the video explains your actual process rather than a general category.

Place B-roll and check the final video

Place the image around the sentence that explains it. Let viewers see the relevant detail while they hear what to look for. Keep the original voice continuous unless the supporting clip’s sound is part of the lesson. Unnecessary competing audio makes the explanation harder to follow.

Check the crop, captions and platform controls together. TikTok’s performance-ad guidance recommends vertical framing and respecting safe zones. Its advice comes from an advertising context; it does not establish an organic distribution rule. Use the destination app’s current preview to see whether its buttons or text obscure your important detail.

W3C’s caption guidance also says captions should not obscure relevant information. On a screen demonstration, move the explanation or caption away from the control being shown. If both must occupy the same space, change the layout or simplify the shot.

1. Watch once with sound: each cutaway should support the sentence beside it.
2. Watch once without sound: captions and visuals should still make the central point understandable.
3. Watch on a phone: inspect small details, line breaks and interface overlap.
4. Review provenance: every claimed result uses the actual evidence, and every asset has a recorded source.
5. Watch the export: confirm that the final file uses the intended image, crop, sound and timing.

If a supporting shot introduces a new question without answering the previous one, remove or replace it. The goal is a video people can follow. Once the explanation is clear, the retention guide can help you study what happens when real viewers watch it.