S-Roll

S-Roll is a free Mac app that scores long videos, turns chat requests into clips, tracks subjects for vertical reframes and exports watermark-free MP4s, all processed locally on Apple silicon.

S-Roll is a Mac application that uses on-device machine learning to find and extract the most interesting moments from long video footage. Instead of uploading hours of gameplay, interviews or streams to a cloud service, the app downloads its models once and then runs the entire clipping workflow locally. The result is a private editing assistant that produces shareable short clips without a watermark, aimed at creators who already have raw footage and need a faster way to surface the parts worth posting. The workflow starts with a scored review. S-Roll analyses a video and presents a grid of candidate clips, each with a title, a short summary and a preview, so the first pass is browsing rather than scrubbing through a timeline. A chat panel handles more targeted requests: a user can describe an action, a moment or a topic in plain language and have the matching section turned into a clip. For vertical output, automatic subject tracking reframes the shot as the subject moves, keeping the action in frame without manual keyframing. Selected clips move into an editor where cuts can be trimmed, framing adjusted and captions or titles styled; a saved style can be reused as a brand preset before the final MP4 export. The tool fits between automatic highlight generators, which often feel like a black box, and full editors, which demand time and skill. S-Roll covers the discovery and rough-cut end of that range and leaves detailed finishing to dedicated software. It is built for Apple silicon Macs running macOS 26 or later, and the first launch requires roughly 6 GB of model downloads. Footage, transcripts, chats and exported clips stay on the machine; account details and update checks use the network. The app is free and exports carry no watermark.