How to stack a speaker above the thing they are talking about

Last updated 2026-09-05 · 8 min read

What is the short version?

When somebody is talking about a film, a game, a product or a document, the viewer wants both things at once — the person and the thing itself. Stacking gives you that inside a vertical frame without ever cutting away from either one of them.

1. Go to Vizard Agent with the interview and the material.
2. Say the speaker goes on top and the subject underneath.
3. Ask how the heights should divide before it cuts.

What do you need before you start?

Both sources, plus a view on which of them is actually the point. The layout Vizard Agent builds follows directly from that: a piece about the film gives the film more room, and a piece about the person gives the person more.

• The interview. The talking head.
• The material. Clips, screens, or stills.
• Which one leads. It sets the split.
• A header, if you want one. A title and a logo.
• The platform. It decides the safe area.

What do you type into Vizard Agent?

Ask for measured heights rather than a split. "Half and half" is easy to say and almost never right — a talking head cropped to exactly half a vertical frame loses the top of the head, and film clips squeezed into the other half lose their own shape.

Prompt

Variants worth knowing:

• "Work out the heights." Not a fixed split.
• "Header with title and logo." A third band, above both.
• "Keep the clips' proportions." Stops the lower panel distorting.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real set of stacked social cutdowns from a director interview. Steps six and seven are the ones almost everybody skips, and they are the reason most stacked layouts look very slightly wrong without anyone being able to say why.

1. Looks at the supplied stills and the interview's opening frame.
2. Reads the transcript in parts and picks the passages to cut.
3. Builds a layout preview from the stills before touching the video.
4. Re-extracts the interview windows with the speaker's face centred.
5. Rebuilds the surrounding chrome and extracts the stacked panels.
6. Measures the rendered header, the gap and the lower panel heights.
7. Checks the lower panel's height against the speaker's crop.
8. Rebuilds the header with larger type and more inset.
9. Checks how the title and logo sit inside that header.
10. Cleans the transcript, builds the captions and measures loudness.
11. Reviews a whole cutdown as a frame grid in the new layout.
12. Bleeps a word and removes it from the captions at the same time.
13. Re-extracts a clip so the interviewer's full question is included.
14. Re-cuts another so a specific line is actually in the audio.
15. Maps the trailer music's arc for the longer piece.
16. Reuses the established crop when the source is a different resolution.
17. Uploads the finished cutdowns.

Step six is the whole craft. A stacked layout is three heights that have to add up — header, speaker, subject — and treating them as fractions of the frame produces a speaker whose forehead is cut and a panel whose contents are letterboxed inside their own box. Measuring the rendered result and adjusting is the only reliable way to land it.

Step three is the cheap version of the same idea. Building the layout from stills, before any video is cut, settles the proportions while changing them costs nothing.

What does the result look like?

A vertical cutdown with a title band at the top, the speaker below it framed properly, and the material they are discussing filling the lower panel at its own aspect ratio — with captions that do not collide with either.

Both halves stay legible on a phone, which is the whole test. A stacked layout that requires pinching to see the lower panel has failed at the one thing it was for.

When does this not work well?

Stacking spends vertical space on structure itself, so both panels end up considerably shorter than either would be alone. Where one of them genuinely needs the whole frame to read, a cutaway serves you far better than asking Vizard Agent to hold a permanent split.

• Detail-heavy material. A dense screen will not read at half height.
• A speaker who moves a lot. The panel crop has to follow them.
• Very long stretches. A static layout tires; cut away occasionally.
• Three things at once. Two panels is the practical limit.
• Material in the same shape as the frame. It gains nothing from stacking.

How do you fix a result that came back wrong?

Say which of the three bands is wrong rather than describing the layout. Vizard Agent keeps the measured heights, the speaker's crop and the header layout as three separate settings, so changing the split will not disturb the framing inside either panel.

• "His head is cut off." Speaker panel re-measured and re-cropped.
• "The clips look squashed." Lower panel restored to its own ratio.
• "The title is too small." Header rebuilt with more inset.
• "Captions cover the film." Moved into the gap between panels.

How does Vizard Agent compare to doing it yourself?

By hand this is a template, and a template is exactly right until the source changes shape — a different camera, a 4K master, a clip that was shot wide. Then the fractions that worked last time quietly stop working.

Common questions

Should the speaker be on top? Usually, but not always. Vizard Agent puts whichever leads on top.

Is it always half and half? No, and that is the point. Vizard Agent measures rather than splitting.

Can I have a header too? Yes. Vizard Agent builds it as a third band with your title and logo.

Where do captions go? In the gap, usually. Vizard Agent checks they cover neither panel.

Can the lower panel be a screen recording? Yes. Vizard Agent takes clips, screens or stills.

What if the speaker moves? Vizard Agent follows them within the panel rather than fixing the crop.

Will the clips be distorted? No. Vizard Agent keeps their proportions inside the panel.

Can it switch to full screen sometimes? Yes, and Vizard Agent usually should for a strong moment.

Does it work for a horizontal video? Side by side works better there. Tell Vizard Agent which you want.

What about a 4K source? Vizard Agent reuses the established crop and re-derives it for the new size.

Can it bleep a word in this layout? Yes, and remove it from the captions in the same pass.

How long can a stacked section run? A minute or so before it tires. Ask Vizard Agent to break it up.

Can I get a full-screen version too? Yes. Vizard Agent renders one from the same cutdowns.

Why measure instead of using fractions? Because three bands have to add up to the frame exactly, and a fraction that is close leaves either a cropped forehead or a letterboxed panel.