What shipped, what's still rough.

The first public build. Installed from source, tested on Windows 11 with an NVIDIA GPU, best-effort on Linux.

Added

• Story view with four lenses: Script (cut and restore by dragging across words), Board (reorder scenes, arc strip by role), Timeline (Picture, B-roll, Graphics, Music and Captions lanes) and a read-only Source lens that shows the video as script.mdx.
• Local transcription with WhisperX large-v3 and word alignment, as an optional pack.
• Automatic story layout into 3 to 8 titled scenes.
• Word-anchored cues: drop B-roll, graphics, music or a sound effect onto a word.
• Motion engine shared by preview and export: Footage, Animation and Cameo shot modes, four layouts, camera moves, seven backdrops, duos and ensembles, and the Lens character.
• Co-pilot Ask bar (optional, your own Anthropic key) with four tools: plan scenes, cut filler words, suggest B-roll and set a shot. Every result is a proposal you accept or reject.
• Captions burned in a few words at a time with the active word highlighted.
• Export to 16:9, 4K, 9:16 and 1:1 at once, NVENC with a libx264 fallback, loudness normalised, with a scene cache for re-exports.

Known rough edges

• Transcription is English only.
• macOS is not supported. Linux is best-effort.
• Silence trimming, karaoke captions and face-tracked reframing exist in the pipeline but have no controls yet.
• Code pull requests are paused while the contributor agreement is finalised. Issues and discussions are open.