Reverse Studio: take an explainer video apart, stage by stage

Reverse Studio is the analysis side of an explainer-video pipeline: instead of generating a video, it takes a finished one apart. A sidebar lists the pipeline stages (frame extraction, scene-cut detection, transcript, shot classification, entity detection, motion pairing, animation taxonomy, SFX isolation and counting, music-bed extraction, cross-modal sync and a query layer) with a status and count for each. The screens show an overview of one source video, the entity-and-motion pairs found per keyframe, a taxonomy of the animation moves used, isolated sound effects and how often each recurs, a timeline lining up frames, speech, SFX and music, and a query screen for asking questions of the results.

The problem and who it is for

The problem it addresses: to write rules for generating explainer animation, SFX and music, you first need ground truth from videos that already work. It is for the person building or tuning the forward pipeline.

Stack and status

Stack observed in the source: React 18 with in-browser JSX, a CSS file with light and dark themes, and a plain-JavaScript data file. Status: a UI prototype; the pipeline is represented by mock results for one mock video, and two screens are marked coming soon. It is one of the explainer-video tools listed on the homepage.

Live demo

Prototype status: UI demo · the channel, the video, every stage count, entity, SFX and metric on these screens is mock data from a bundled data file

Pick a stage in the sidebar to switch screens; the Tweaks panel is hidden unless the page is embedded in an editor. The demo needs JavaScript and a desktop-width window.