How do I turn a panel discussion into short clips?
Upload the panel as a single file, inside the 3-hour and 2 GB limit, in MP4, MOV or WebM, then pick how many clips you want, from 1 to 20, and a target length between 15 and 90 seconds. shortshort transcribes the panel, keeps the passages that form complete sentences, crops each one to 9:16 on the person speaking and times the captions to the audio. You are charged one credit per minute of source whatever the number of clips, and each export comes back as a 1080 x 1920 MP4.
There are five people on stage. How does it know who to frame?
The vertical crop sits on the face of whoever is talking and hands over at the change of speaker, rather than hovering between them. A wide shot that holds all five, or a slide on the screen behind, is held whole with black bands above and below, since cropping it vertically would remove people from the picture. Any framing keyframe can be corrected by hand afterwards.
Can each panellist get their own clip to post?
Yes, and that is usually why you would ask for ten or more. The ceiling per video is 3 clips on the free tier, 5 on Starter, 10 on Creator and 20 on Studio, so a five-person panel on Creator gives every speaker two. Exported shorts join the public gallery by default; every paid plan lets you keep a panel out of it before you send the files to the speakers.
What happens when two panellists talk over each other?
Cross-talk is rarely picked, because the selection keeps passages that read as finished sentences and a sentence spoken over someone else is not one. If a passage does carry two voices, the framing can hold the whole row instead of jumping between faces, and the in and out points stay editable in the studio before you render again.
Our panel was filmed as one locked wide shot from the back of the room. Is that enough?
It works. The vertical frame is taken inside that wide image and lifts the zoom as far as x1.6 on the panellist speaking, which is what brings a distant row of chairs to a size worth watching on a phone. If the source is low resolution, or a zoom lands beside a face, each keyframe can be moved or pinned by hand.