Does the crop switch between two people in a conversation?
Not by voice: there is no automatic switching between speakers. The crop is placed on the face on screen and moves with it, and scene detection resets it when the recording itself cuts to another shot. Wide shots are detected and kept full-frame, with space above and below. If a keyframe lands on the wrong person, drag the frame to the face you want or lock that keyframe in the studio, then render again.
Can I output a square or 16:9 version?
No. shortshort renders one format: 1080 x 1920 vertical MP4 at 30 fps, H.264. There is no 1:1, 4:5 or 16:9 export and no 4K or 60 fps option. The Studio plan can export a short's edit plan as JSON for rendering on your own server, but the studio itself produces nothing other than the 9:16 file.
What happens on a stage recording filmed from far away?
The crop zooms in on the speaker up to x1.6 and moves with them across the stage. Because the zoom stops at x1.6, a very wide or low-resolution source may leave the speaker small; you can then adjust or lock a keyframe on the framing you prefer.
Why is the zoom limited to x1.6?
The zoom stops at x1.6 because a crop magnifies the source pixels, and the bound keeps the picture sharp on an ordinary recording. The range x1.0 to x1.6 is the one the AI works within on every plan, and every keyframe it writes can be changed by hand.
Is speaker framing part of the free plan?
Yes. Tracking, full-frame detection for slides and the keyframe editor are in the shortshort studio on every plan, including Free, which allows up to 3 shorts per video with a watermark on the export. Manual framing changes and the MP4 export do not even need an account; the AI analysis does.
Does it add B-roll or cut to a second camera?
No. shortshort works with the single file you upload and only crops and zooms inside it. It does not insert B-roll, stock footage, music or animated titles, and it cannot pull in a second camera angle that is not in the recording.