Skit Video Generator · Open Skill 公開スキル
Skit Video Generator
AI Drama Skill: "The Thespian Engine" – Functional Description The "Thespian Engine" is an AI skill designed to simulate, script, and perform dramatic narratives with emotional depth, subtext, and character-driven conflict. The skill functions as an interactive dramaturgy engine that prioritizes psychological authenticity and narrative tension. #### 1. Core Competency: Character Archetyping & Psychological Profiling The skill utilizes a "Character Matrix" system. Instead of generic dialogue, it constructs personas based on: Backstory Anchors: Hidden history markers that dictate how a character reacts to specific triggers. The Subtext Layer: The skill generates "Internal vs. External" dialogue tags—what the character says versus what they are actually thinking, allowing for nuanced, non-linear human interactions. Emotional Arc Tracking: The system monitors a "Conflict Meter," adjusting the dialogue tone dynamically from subtle passive-aggression to explosive confrontation based on the progression of the scene. #### 2. Narrative Engine Specifications Beat Sheet Integration: The skill operates on structural beats (Setup, Inciting Incident, Rising Action, Climax, Resolution). Users can input a "Dramatic Goal," and the AI will automatically generate scene beats that increase the stakes. Staging & Directional Cues: Beyond spoken lines, the AI embeds stage directions regarding tone, body language, and environmental interaction, ensuring the "performance" feels visual and visceral. Conflict Velocity Control: A slider parameter that allows the user to control the "pacing of tension," ranging from "Slow Burn" (focus on dialogue and atmosphere) to "High Stakes" (focus on rapid escalation and plot twists). #### 3. User Interaction Workflow The "Rehearsal Space": A dedicated interface where the user acts as the Director. The user provides a scene prompt, and the AI generates a draft. The user can then "direct" the characters by providing feedback like, "Make him more defensive," or "She is lying about the money." Dynamic Response Branching: During a live session, the AI allows the user to step into a scene as a character. The engine interprets the user’s input and calculates the logical emotional response, ensuring the drama remains consistent with the established character profiles. Tone Matching Library: The skill includes a library of "Dramatic Styles" (e.g., Noir, Shakespearean, Modern Minimalist, Absurdist) which alters the vocabulary, rhythm, and structural approach of the dialogue generation. #### 4. Creative Utility Output The final output of this skill is not merely text, but a "Drama Manuscript" that includes: Dialogue Scripts: With embedded emotional tags. Character Motivation Summaries: Explaining the why behind every line delivered. Tension Graphs: Visual representations of how the emotional energy of the scene builds and releases. This skill serves as a collaborative partner for playwrights, novelists, or roleplay enthusiasts, acting as a tireless scene partner that maintains perfect character continuity and narrative integrity.
ShortDrama
8 1 1
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4127000/2026-09-27/yh9595kz.png
# Skit Video Generator ## When to use - The user gives a prompt/idea and wants short comedy-sketch or drama mini-scene videos for TikTok / Reels / Shorts. - The user wants a few skit script variations to pick from, consistent characters across skits, and auto voiceover + background music. ## Steps 1. Clarify the idea: genre (comedy sketch or drama mini-scene), tone, rough length, and any character details. If anything is unclear, ask before generating. 2. Use `write_todos` to list the pipeline steps and update the list as you progress. 3. Generate 3 skit script variations with `create_document` (`entity_type='scripts'`, `name='skit-script-variations'`). Present them in natural language and let the user pick one (or ask for a tweak to a variation). 4. Once a variation is picked, define the characters. Generate one character reference image per main character with `generate_image` (`model='t2i'`, `orientation='portrait'`, `entity_type='characters'`, `name='<CharacterName>'`). These become the canonical look reused across skits. 5. Generate one storyboard image per scene/beat of the chosen skit with `generate_image` (`model='i2i'`, `input_urls=<character reference images>`, `orientation='portrait'`, `entity_type='storyboard_images'`, `name='scene-<n>'`). 6. Generate one video clip per storyboard image with `generate_video` (`model='i2v'`, `input_urls=<storyboard image>`, `orientation='portrait'`, `entity_type='storyboard_videos'`, `name='clip-<n>'`). 7. Generate voiceover for the dialogue/narration with `generate_tts` (`model='tts.design'`, `voice_instruction=<tone matching the scene>`, `entity_type='voiceovers'`, `name='vo-<line>'`). One clip per line/beat. 8. Generate background music with `generate_music` (`model='music'`, `entity_type='music'`, `name='bgm'`). 9. Mix everything with `mix_clips`: concatenate the video clips in order, overlay the voiceover clips (each with a `delay` matching its beat) and the background music track. Pass `orientation='portrait'`, `entity_type='final_videos'`, `name='final-skit'`. 10. Present the final video in natural language. Offer next steps: pick a different variation, restyle a character, change the music/voice, or redo a single clip. ## Notes - Always vertical 9:16 (`orientation='portrait'`) for TikTok / Reels / Shorts. - Keep characters consistent by reusing the character reference images via `i2i` (and `ref2v` if a clip needs identity-preserving generation). - Video generation does not support NSFW/adult content on any channel — politely decline such requests. - Report failures honestly; never fabricate URLs. - Intermediate storyboard images/videos and clips will be visible on the canvas. - When a stage completes, call `set_guidance` to suggest the next actions.