Formats · Length 30-90s · 9:16 · beginner
Talking head video: how to shoot, edit, and prompt it
A talking head is one person talking to camera, cut tight, usually with captions and no b-roll.
By Sulan Zhang, co-founder of Vyra · Updated 2026-09-24 · Example: @sulansart
- Platforms
- TikTok, Reels, Shorts, YouTube
- Length
- 30-90s
- Aspect
- 9:16
- Difficulty
- beginner
- Gear
- phone, window light, a quiet room
It's the format I use most. You need a phone, a window, and something to say.
At a glance
| Best for | Opinions, advice, explanations, updates |
| Hook style | Say your claim in the first 2 seconds |
| Pacing | One idea every 5-8 seconds, no dead air |
| Caption style | 3-5 word phrases under your mouth |
| Music | None, or a quiet bed |
| Shot list | One medium close-up, shoulders in frame. A 15-20% punch-in for cuts. 5 seconds of room tone. |
| Common mistakes | Leaving pauses in. Framing too wide. Reading notes below the lens. |
Structure
- 0-2s: the claim
- 2-8s: why it matters to them
- 8-40s: 2-4 points, one clean sentence block each
- 40-55s: the thing they can do today
- 55-60s: short close, no "like and subscribe"
How to shoot it
- Face a window, camera at eye level, about an arm's length away.
- Frame from mid-chest up. Both shoulders in.
- Record 2-3 takes of each point. Don't restart the whole video.
- Talk a little faster than feels normal. Cuts can fix slow. They can't fix flat.
- Leave 2 seconds of quiet before and after each take.
- Lock focus and exposure so the cuts don't flicker.
How to edit it
- Cut every pause over 0.3 seconds and every false start.
- Switch between full frame and a 15-20% punch-in at sentence breaks. It hides the jump cuts.
- Captions in 3-5 word phrases. One style, no bouncing.
- Pick one line and make it a big caption. Just one.
- Keep the room audio. Add music only if the delivery is calm.
- End on the last useful word. No fade.
Prompt it in Vyra
Cut this 6-minute talking head down to the best 60 seconds for Reels. Remove every pause, "um", repeated take and the parts where I look at my notes. Keep my strongest take of each point. Add 3-5 word captions in white under my mouth. No music.
Alternate between the full frame and a slight zoom-in at sentence breaks so the jump cuts look intentional. Make the line "you are not charging enough" a big bold caption on its own.
Match the caption style and cut rhythm of the reference video I attached. Keep my footage and my order of points.
Variations
- Punch-in talking head, two framings from one take, cut on every idea.
- Text-hook talking head, a bold title covers the first 2 seconds instead of a spoken hook.
- Walking talking head, handheld, outside, more energy.
FAQ
How long should it be? 30-60 seconds on TikTok and Reels, up to 90 on Shorts. Cut until nothing is skippable.
Do I need two cameras? No. One 4K take gives you enough room to punch in 20% for a second angle.
Should every word be captioned? Yes for short-form. Most people watch muted.