What is Smart Composition? How AI video reframing works.
A practical guide to choosing between a crop, a tracked crop and a rebuilt composition.
Start with what must remain visible.
A video frame may contain a speaker, a demonstration, a scoreboard and captions at the same time. List which elements are editorially essential before choosing a portrait layout. A visually smooth crop is still a poor output if it removes the information the audience needs.
Three approaches, three trade-offs.
Center cropping uses a fixed window and is simple to automate. Subject tracking moves that window as a person or object moves. Smart Composition can rebuild the layout from multiple relevant regions, which is useful for multi-speaker studios, split screens and sports with persistent graphics.
How the workflow works.
Analyze the source, identify important regions, select a destination layout, recompose each scene and review the transitions. Keep framing stable enough to watch comfortably. Define fallback layouts for ambiguous scenes rather than forcing one crop through the entire program.
What to test.
Use examples with fast action, several faces, shot changes, lower thirds and full-screen graphics. Check whether the output preserves meaning, readability and continuity. Evaluate the actual target device, not only a large desktop preview.
Center crop, tracking crop, or recomposition?
A fixed center crop keeps one window from every frame. A tracking crop moves that window to follow a subject. Smart Composition can rearrange distinct regions into a new layout. For a studio panel, that can mean retaining multiple speakers rather than forcing the entire conversation into a single portrait crop.
A useful newsroom test
Use a representative broadcast with a presenter, supporting footage, lower thirds and a full-screen graphic. Compare whether names remain readable, the active speaker stays visible, and the supporting evidence survives the change of format. A good portrait output preserves meaning, not just faces.