Universal Story-to-Video Director · skill pública de Open Skill

Universal Story-to-Video Director

Universal Story-to-Video Director es una Open Skill pública en Digen. Ábrela para ver qué crea este flujo.

@SDramavox26

Story

14 2 4

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4224455/2026-10-10/y9akk63f.png

# Universal Story-to-Video Director

Turn any user-provided story or screenplay into a cinematic, scene-by-scene AI video: matching visuals, character-specific voices, natural dialogue, realistic expressions, and consistent characters.

## When to use
- The user provides any complete script, story, or screenplay and wants it turned into a cinematic scene-by-scene AI video.
- The user wants to restyle an existing story (change the visual style while preserving characters, plot, dialogue, and scene continuity).
- The user wants to advance or change one stage of an ongoing project (rewrite the breakdown, next clip, restyle a character, re-voice a line, remix the cut).

## Reference files
- `references/pipeline.md`: the full stage-by-stage workflow. Read this first when starting a project.
- `references/voice-direction.md`: how to design distinct, consistent character voices. Read when generating voiceover.
- `references/genre-templates.md`: per-genre defaults for visual style, camera, pacing, expressions, sound, and voice (horror, romance, comedy, action, mystery, thriller, family, fantasy/sci-fi). Read when the story's genre is clear, to shape direction.

## Steps (overview)
1. Read `references/pipeline.md` for the full workflow before starting.
2. Ingest the story. Ask for visual style / aspect ratio / duration only if not inferable; otherwise use sensible defaults. Default visual style = the user's stored preference (`memory_query` first; if none is stored, default to realistic cinematic drama). Default aspect ratio = landscape 16:9; natural duration per dialogue. If the story's genre is clear, read `references/genre-templates.md` and apply its defaults.
3. Call `write_todos` to list the stages and update it as you progress.
4. Produce the scene-by-scene breakdown as a document (`entity_type='scripts'`).
5. Build character reference sheets (`entity_type='characters'`).
6. Generate scene images (`entity_type='storyboard_images'`).
7. Generate scene clips (`entity_type='storyboard_videos'`).
8. Generate character voiceover (`entity_type='voiceovers'`) and background music (`entity_type='music'`).
9. Mix the clips in story order with burned-in subtitles (`entity_type='final_videos'`).
10. QC with `recognize_video` / `recognize_image`; fix and re-render as needed.
11. Summarize the result in natural language (no URL dumps) and call `set_guidance` for next steps.

## Canvas conventions (fixed per stage)
| Stage | Tool | `entity_type` |
|---|---|---|
| Scene breakdown | `create_document` | `scripts` |
| Character reference sheets | `generate_image` | `characters` |
| Scene images | `generate_image` | `storyboard_images` |
| Scene clips | `generate_video` | `storyboard_videos` |
| Voiceover | `generate_tts` | `voiceovers` |
| Music | `generate_music` | `music` |
| Final cut | `mix_clips` | `final_videos` |

Use a short, descriptive `name` on each asset (character name, scene/shot title, final cut title).

## Notes
- The user's script is the source of truth. Never replace it with a generic summary, never omit meaningful dialogue or reactions, never change character identities between scenes, never use the wrong speaker's voice.
- Video generation does not support NSFW/adult content on any channel — politely decline such requests for video.
- Intermediate stage products (breakdowns, sheets, clips) will be visible on the canvas; that's expected.
- Report failures honestly, never fabricate URLs, and use the closest supported workflow when a requested feature is unavailable.
- Remember the user's preferred visual style / aspect ratio across sessions with `memory_store` (kind `preference`); query it before starting.