← All articlesRepurposing

Interviews & Panels: Extracting Shorts From Landscape Conversations

Interviews and panels are clip goldmines hidden in 16:9. Here is how to extract vertical shorts from landscape conversations without breaking the multi-speaker frame.

Repurposing 💬 2-5 speakers per frame

A good conversation is the densest clip source there is. Put two thoughtful people in a room — or four on a panel — give them a topic and an hour, and they will produce a string of quotable, surprising, emotionally charged moments that no scripted format ever matches. The back-and-forth generates tension, the follow-up questions dig out admissions, and the unguarded exchanges produce lines that travel. An interview or panel is, in effect, a moment-generation engine. The problem is that all of those moments are buried inside a long 16:9 file built for two or more people side by side — the exact shape that resists becoming a vertical short.

That tension is what this post is about. Conversations are the best raw material for shorts and the hardest to reframe, because the thing that makes them valuable — multiple people interacting — is also the thing that breaks a vertical crop. Master the extraction and a single landscape interview becomes a week of feed content; fumble it and you get clips with two half-faces and a slice of empty background. Here is how to pull shorts out of landscape conversations without losing the people, the exchange, or the substance that made the conversation worth recording.

2-5speakers per frame
10-20quotable moments per hour
1active speaker at a time

Why conversations generate so many clips

Monologue content is constrained by one person’s preparation; conversation is constrained by nothing. When two people interact, each response is shaped by the other’s question, so the material goes places neither would have reached alone. A sharp question pulls out a story the guest didn’t plan to tell. A disagreement on a panel exposes the real tension in a topic. A follow-up forces specificity where a prepared talk would have stayed vague. The result is a recording dense with self-contained moments — a strong claim, a vivid anecdote, a clean disagreement — each of which is already shaped like a short.

This density is why a single hour-long interview routinely yields ten to twenty viable clips while an hour-long monologue might yield five. The conversation is doing the work of generating hooks for you, continuously, as a byproduct of two people thinking against each other in real time. Your extraction job is not to invent moments but to recognize the ones the conversation already produced — the exchanges where something genuinely landed — and lift them out. The material is rich; the only question is whether you can reframe it without breaking it.

The multi-speaker reframing problem

Here is the core difficulty. A conversation is shot wide because it has to hold multiple people, usually seated or standing apart across the 16:9 frame. That layout is fundamentally hostile to a 9:16 crop, which can only show a narrow central column. Crop the middle of a two-shot and you get the gap between the speakers — background, with a sliver of each person at the edges. The single most valuable thing in any conversation clip, the face of the person currently talking, is precisely what a fixed crop is structurally unable to keep.

Worse, the active speaker keeps changing. In a real exchange, attention should be on the interviewer when they ask and the guest when they answer, swinging back and forth every few seconds. A static crop can’t follow that; it stays parked on one region while the conversation moves around it. To reframe a conversation faithfully, the vertical frame has to be speaker-aware — it has to know who is talking right now and follow them, then follow the next person when the turn passes. Doing that by hand means re-cutting the crop on every line, which is why so many clipped conversations look broken and why so many teams give up on clipping them at all.

Naive crop vs. speaker-tracked extraction

SituationNaive center-cropSpeaker-tracked
Two-person interviewBoth at the edgesActive face centered
Turn-takingCrop stays putFollows each speaker
Four-person panelMostly backgroundTracks whoever speaks
Reaction momentsMissedCaptured
Effort to do wellHand-keyframingAutomatic

The difference is whether the crop understands the conversation or just slices the middle of the frame. Speaker-tracking turns the hardest reframing case — multiple people, constant turn-taking — into something automatic and faithful, which is what makes conversations practical to clip at volume rather than one painstaking clip at a time.

A conversation-to-clips workflow

1Drop in the full conversationStart from the complete landscape interview or panel.
2Surface the exchanges that landedLet AI find the strong, self-contained moments.
3Reframe with speaker-trackingCrop to 9:16 and follow whoever is talking.
4Caption both sidesSubtitles make the back-and-forth legible on mute.
5Cut question-and-answer pairsKeep the setup line so the payoff makes sense.

The workflow leans on automation exactly where conversations are hardest — finding moments across a long exchange and following speakers through turn-taking. AI clipping with face-tracking handles both, so the multi-speaker problem that used to kill conversation clipping becomes a non-issue and you get to focus on choosing which exchanges to publish.

Captions and the rhythm of dialogue

Conversation clips have a specific captioning need: the viewer has to follow a back-and-forth, often between people who are no longer both on screen because the crop is following one at a time. Subtitles carry the dialogue’s thread when the visuals can’t show both faces, letting the viewer track who said what even as the frame swings. They also make the clip legible on mute, which matters doubly for conversations, where the meaning is entirely verbal. A captioned conversation clip reads cleanly even at a tight crop, because the words preserve the exchange the cropped frame can only partly show.

Where the clips come from in an hour

Clip yield across a one-hour interview
Sharp answersmany
Stories & anecdotesseveral
Disagreementssome
Throwaway lines that popunpredictable

The clips come from everywhere in a good conversation — the planned answers, the stories, the friction, and the off-hand lines nobody scripted. That spread is exactly why conversations out-yield monologues, and why you should clip generously across the whole recording rather than only the parts you expected to be good. The unscripted line that pops is often the one that travels furthest.

💡Keep the question in the clip. A great answer often makes no sense without the question that prompted it. When you cut a conversation clip, include the setup line so the payoff lands — a few seconds of context turns a confusing fragment into a complete, self-explaining short.
⚠️Don''t clip people out of context. Conversations contain nuance, qualifications, and changes of mind. Lifting a single line that misrepresents what a guest actually meant is unfair to them and risky for you. Clip the moment as it was meant, qualifiers included, rather than the cleaner-sounding version that distorts it.

The conversation is a content engine

Interviews and panels are the richest clip source you can record, because two or more people thinking against each other generate more genuine, quotable moments than any single speaker can. The only reason they stay under-clipped is the multi-speaker reframing problem — and speaker-tracking solves it. Record good conversations, let the system find the exchanges that landed and follow the active speaker into a vertical frame, caption the back-and-forth, and keep the questions with the answers. Do that and every hour-long landscape conversation becomes a week of shorts, each one a moment the conversation already produced for you.

Key takeaways

  • Conversations out-yield monologues because interaction generates moments continuously.
  • Multi-speaker 16:9 layouts are the hardest case for a vertical crop.
  • Speaker-tracking follows the active talker through turn-taking, making clips faithful.
  • Captions carry the back-and-forth when the crop can only show one face.
  • Keep the question with the answer, and clip the moment as it was actually meant.

More on landscape-to-shorts

Turn every conversation into a week of clips

Extract vertical shorts from landscape interviews and panels with speaker-tracking.

Turn landscape video into shorts →
Landscape to shortsRepurposingInterviews