How natalia-3nitilab/vita-myth-studio works
vita-myth-studio produces narrated, long-form folklore videos with a persistent 3D VRM performer rather than AI-generated frames. The main workflow runs from a story script through narration and timing, story-beat backdrops, sentence-aware shot planning, and deterministic scene rendering. Supporting branches cover hook and episode packaging, character motion and appearance, and the production dashboard for reviewing episodes, renders, and backgrounds. The README documents several capabilities whose source was not sampled; those are represented at their documented module or directory level, without inferring internal wiring beyond the described workflow.
Main entry points: render.mjs Episode renderer · orpheus.txt Story script · index.html Review dashboard · tts.py Speech synthesis
Components 20
Story preparation
- Story script
content/orpheus.txt - Script generation
episode/build/script.py - Hook variations
episode/build/hooks.py - Episode packaging
episode/build/package.py
Narration and settings
- Speech synthesis
episode/build/tts.py - Mouth timing
episode/build/envelope.py
Scene and character
- Backdrop generation
episode/build/backdrops.py - Motion retargeting
stage/retarget.mjs - Character looks
stage/looks.mjs - Scene dressing
stage/props.mjs - 3D stage
stage/scene.html
Episode rendering
- Shot planning
episode/build/plan.mjs - Episode renderer
episode/build/render.mjs - Rendered episode
episode/build/render.mjs
Production desk
- Review dashboard
dashboard/index.html - Dashboard storage
dashboard/index.html
Other components
- Episode creator
- Vertex AI
- Kokoro TTS
- FFmpeg
Connections 28
- Episode creator → Story script: writes
- Episode creator → Script generation: requests script
- Script generation → Vertex AI: generates with
- Script generation → Story script: writes script
- Hook variations → Story script: reads script
- Episode packaging → Story script: packages metadata
- Episode creator → Hook variations: requests hooks
- Episode creator → Episode packaging: requests options
- Speech synthesis → Story script: reads script
- Speech synthesis → Kokoro TTS: synthesizes with
- Speech synthesis → Mouth timing: provides narration
- Backdrop generation → Story script: reads story beats
- Backdrop generation → Vertex AI: generates locations
- Motion retargeting → 3D stage: provides animation
- Character looks → 3D stage: provides appearance
- Scene dressing → 3D stage: provides set pieces
- Shot planning → Story script: reads sentences
- Shot planning → Backdrop generation: uses locations
- Episode renderer → Shot planning: gets shot plan
- Episode renderer → Speech synthesis: uses narration
- Episode renderer → Mouth timing: uses mouth timing
- Episode renderer → 3D stage: drives frames
- Episode renderer → FFmpeg: assembles video
- Episode renderer → Rendered episode: writes output
- Review dashboard → Dashboard storage: reads and writes
- Episode creator → Review dashboard: reviews episodes
- Review dashboard → Rendered episode: reviews render
- Review dashboard → Backdrop generation: reviews locations