Formats · Length 45-120s · 9:16 or 16:9 · intermediate
Talking head with b-roll video: how to shoot, edit, and prompt it
A talking head with b-roll is you talking to camera while clips of what you describe cover the parts where you mention them.
By Sulan Zhang, co-founder of Vyra · Updated 2026-09-24 · Example: @sulansart
- Platforms
- TikTok, Reels, Shorts, YouTube
- Length
- 45-120s
- Aspect
- 9:16 or 16:9
- Difficulty
- intermediate
- Gear
- phone, tripod, 10-20 minutes of supporting footage
Same as a plain talking head, but when you say "the brushes" the viewer sees the brushes.
At a glance
| Best for | Process explanations, reviews, "how I did this" |
| Hook style | Your claim, spoken over your best b-roll shot |
| Pacing | Face 4-8 seconds, b-roll 2-4 seconds, repeat |
| Caption style | 3-5 word phrases that stay on during b-roll |
| Music | Low bed, ducked under speech |
| Shot list | The talking head. 8-15 b-roll clips of 5-10 seconds. Hands, objects, screens. One wide. |
| Common mistakes | B-roll that doesn't match the words. Cutting away mid-word. Face gone for more than 6 seconds. |
Structure
- 0-3s: hook over the best b-roll clip
- 3-10s: cut to face, who you are and why this
- 10-70s: face and b-roll alternate on every new noun
- 70-85s: the payoff on your face, no b-roll
- 85-90s: one last b-roll shot or a still
How to shoot it
- Record the talking head first. Then write down every concrete thing you mentioned.
- Shoot 5-10 seconds of b-roll for each thing on the list.
- Move the camera a little on b-roll. A slow push or a slide.
- Match the light to the talking head so the grade stays even.
- Shoot vertical if the final is vertical. Cropped horizontal looks soft.
How to edit it
- Cut the talking head tight first, on its own.
- Put b-roll on the exact word it shows, starting a few frames early.
- Come back to your face before every new claim.
- Keep captions running over b-roll.
- Music 6-10 dB under speech.
- Straight cuts only.
Prompt it in Vyra
Cut my talking head to 75 seconds, then cover everything I mention with matching clips from the project. Every time I name a tool, place, or step, cut to the b-roll that shows it for 2-4 seconds and come back to my face. Captions stay on throughout.
Find the moments where I mention the "morning routine" and overlay the kitchen and desk clips there. Keep the audio from the talking head only.
Use the attached reference to set the rhythm of face vs b-roll. Add a soft lo-fi bed that ducks under my voice.
Variations
- Screen b-roll, screen recordings cover software mentions.
- Photo b-roll, stills with a slow zoom for old stories.
- Split frame, face on top, b-roll below.
FAQ
How much b-roll do I need? About a third of the runtime. For 90 seconds, 12-15 clips you can use for 2-4 seconds each.
Can the agent pick b-roll for me? Yes. Vyra transcribes your speech and analyzes every clip, so it can put the coffee pour on the sentence where you say "coffee".
Should b-roll have sound? Usually no. Keep the talking head audio and the music.