dbs-video-navigation reads a local video or a timestamped SRT, divides it into chapters, topics or steps by what is being solved, and renders a black-and-white navigation strip with a blue segmented progress bar. The strip is a normal-speed MP4 that covers the range you choose, delivered with a project snapshot, a first-and-last subtitle timetable, placement notes and an acceptance record.
It checks FFmpeg, fonts and the source video first, looks at key frames for screen recordings and silent parts, renders into a new folder and inspects the frames at every chapter change. It does not merge the strip into your video.
What You Get
- A navigation MP4 with a segmented progress bar covering the chosen range
- A project snapshot that stores source, target canvas, frame rate, chapters and style
- A timetable with the first and last subtitle of every chapter, plus placement instructions
- An acceptance record with frame checks and the items left unverified
Common Use Cases
- Adding a chapter and progress strip to a long tutorial or lecture video
- Showing viewers which step of a screen recording they are on
- Reusing a saved style on a new video while keeping its own timeline
What Makes It Unique
It treats chapter structure as an editorial judgement made from the whole subtitle text and key frames, and it records visual evidence for sections with no dialogue. It will not invent chapters for videos that have no stable phases. Rendering uses exact frame arithmetic to avoid drift, and it never overwrites the source or an earlier output folder.
Under The Hood
The skill runs a bundled Python script that calls FFmpeg and ffprobe on your video to probe, extract audio and encode the strip, and on macOS a small Swift program draws the chapter text, while other systems use Pillow and font files you provide. It writes projects, images, logs and the MP4 into new folders and keeps style presets under your home folder. It has no speech model of its own; if you want cloud transcription it asks for your authorization first and uploads only the extracted audio you approve.