Communitygithub.com

seedance-storyboard-generator

Generador profesional de guiones y storyboards para vídeos de IA en la plataforma Seedance 2.0. Convierte artículos, historias o esquemas en guiones completos, storyboards y activos visuales para modelos como GPT-Image-2, Seedream y Nano Banana Pro.

¿Qué es seedance-storyboard-generator?

Seedance Storyboard Generator es una herramienta especializada para creadores que utilizan la plataforma Seedance 2.0, diseñada para automatizar todo el flujo de trabajo de producción de series de vídeo con IA. Acepta entradas como novelas completas, artículos extensos o simples esquemas narrativos, y produce un guion formateado profesionalmente con descripciones de tomas (precedidas por '△'), diálogos, voces en off, flashbacks y subtítulos. El sistema sigue un flujo de trabajo de cuatro etapas: 'Texto-recursos-vídeo-edición', que abarca la creación del guion, la construcción de activos (personajes, escenarios y utilería), la generación de vídeo, la optimización del storyboard y el diseño de sonido. Soporta la generación de indicaciones de activos para modelos de imagen como GPT-Image-2, Seedream y Nano Banana Pro, asegurando consistencia visual. Incluye funcionalidades avanzadas como la generación de vistas múltiples de personajes (frontal, lateral, trasera y primer plano), presets de estilo (como la animación tradicional china 'ShangMeiYing') y un formato de guion específico para series cortas con estructura de enganche rápido. Es compatible con cualquier plataforma que acepte indicaciones estructuradas, pero está optimizado para Seedance 2.0.

Compatible con~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/liangdabiao/dsh-storyboard-director/tree/HEAD/skills/seedance-storyboard-generator

Preguntar en tu IA favorita

Abre un nuevo chat con esta habilidad de agente ya precargada.

Documentación

Seedance Storyboard Generator

Professional AI Script and Storyboard Generation System for Creating Professional AI Video Series on the Seedance 2.0 Platform.

Core Workflow: Text-Assets-Video-Edit

A new-generation AI short drama production system, replacing the traditional "Text-Image-Video-Audio-Edit" workflow:

StageCore ContentPurpose
TextScript writing & storyboard breakdownProvides the story blueprint and construction blueprint
AssetsCharacter/scene/prop asset libraryEnsures final video stability and consistency
VideoMultimodal direct-to-video generationCharacter + scene + script-driven generation
EditPost-production editing & visual repairCompletes the final audiovisual integration

Workflow

Convert source material into a production-ready video script by following the steps in order:

1. Analyze Input

Determine input type:

  • Full Text: A complete novel/article that requires adaptation and division into episodes
  • Outline: A short story concept that requires full script development

Extract core elements:

  • Protagonist and key characters
  • Core conflict and narrative arc
  • Scene/world elements
  • Key plot points and emotional beats
  • Core gimmick/selling point

If the input is vague or incomplete, ask clarifying questions.

2. Confirm Production Parameters

Essential Questions to Ask:

  1. Visual Style: What visual style? (Realistic/Animation/Ink Wash/Sci-fi/Vintage/Cinematic/Other)
  2. Duration: Total run time? (Standard: Short drama 5 episodes × 15s ≈ 75s; Long drama 20 episodes × 15s ≈ 5 minutes)
  3. Target Platform: Aspect ratio? (16:9 Landscape / 9:16 Portrait / 2.35:1 Cinematic Widescreen)
  4. Tone: Overall emotional tone? (Epic/Warm/Suspenseful/Cheerful/Melancholic, etc.)
  5. Core Gimmick: What is the one-line selling point? (Last stand revenge / Satisfying revenge plot / Heartwarming healing / Suspense thriller, etc.)

Style Preset: Shanghai Animation Film Studio (Activated when the user specifically requests "Shanghai Animation Film Studio / Shanghai Animation Studio animation / Chinese style hand-drawn"; otherwise, follow the user's chosen style via the general workflow):

  • Shanghai Animation Film Studio = Shanghai Animation Film Studio retro hand-drawn animation, four sub-styles: Hand-drawn Ink Wash + Gouache (default) / Ink Wash Light Color / Paper-cut Style / Dunhuang Heavy Color, use only one for a single series
  • Once activated, must strictly follow references/Shanghai Animation Film Studio Style Guide.md throughout: Style blocks (§2, pure 2D flat colors with no shadows or gradients on the image side), Studio color palette declaration (§3.3, verbatim bilingual quoting), Modeling specifications (head-to-body ratio/outline lines/exclusive patterns, replacing the realistic rules of this skill's §4.2), Roaring-style hard constraints (§2.4, must be appended to all image generation prompts)
  • The recipe for the video side differs from the image side (§2.3): Retain hand-drawn texture, only prohibit 3D rendering/photorealism/volumetric light; Seedance first-line style line must be identical word for word per episode, add style-locking phrase at the end of the [Reference] line
  • End-to-end examples: example-projects/Lu Zhishen Drunkenly Makes Havoc at Mount Wutai Project/ (drunken comedy), example-projects/Rakshasa City Fantasy Story Project/ (fantasy mystery)
  • The animation style naturally avoids realistic human face review restrictions (no need for §5.3 circumvention treatment)

Record these parameters and ensure consistent application throughout the process.

3. Generate Complete Script Structure

Important: The output must strictly follow the format below and meet the quality standards of "Good Script.md".

3.0 Short Drama Script Writing Formula

Single Episode Structure Formula: Conflict First + Emotional Pull + Step-by-Step Payoff + Cliffhanger Ending

Each episode is 15 seconds, the pacing is extremely fast, there is no time for flat exposition; must get straight to the point.

AI Script Writing Prompt Template:

You are a professional short drama screenwriter. Please help me create a [genre] short drama script.
The story setting is as follows: [Write down the drama title, genre, number of episodes, protagonist, core desire, antagonist/obstacle, world/background, and any existing plot ideas].

Please follow the three-act structure to set hooks, present the core conflict directly, and grab the audience's attention. The pace must be fast, the information density high, the satisfying moments plentiful, and each episode must end with a new hook.

Requirement: The first episode must begin with a "Golden Three Seconds": use a strong image, strong dialogue, strong conflict, strong contrast, or strong suspense to immediately capture the audience's attention; do not introduce the background in a flat manner. The first episode must also introduce the world, the protagonist's core desire, and present the core conflict as soon as possible. Each episode must have a small reveal, resolving one conflict or completing one plot point, while leaving a small hook. Side stories can be added to enrich the plot but must serve the main storyline.

Please introduce a disruptive major reveal at the end of the second act, and reveal the truth and provide a relatively closed story ending in the third act.

First output: "One-line selling point, story synopsis, main character introductions, episode outline." After I confirm the episode outline, write the official script episode by episode using the standard script format.

Keep the official script format simple and clear: First write the episode number and the characters appearing in this episode, then write "Interior/Exterior, Location, Time" for each scene, followed by visual actions and character dialogues. Dialogue format is "Character Name: Dialogue". Aim for about 3 scenes per episode, and each episode must end with a hook.

3.1 Script Analysis Output Format

# [Title] - Script

I. Core Gimmick
[Two-word or short phrase core selling point, e.g., Last Stand, Classic Wuxia, Satisfying Revenge]

II. Story Synopsis
Story Background: [Era, location, initial situation]
Opening Conflict: [Event that breaks the status quo, triggers conflict]
Protagonist Portrait: [Who is the protagonist, core traits, current state]
Main Storyline: [Key plot progression → Climax → Ending]
Ending: [Final result, character transformation]

III. One-Line Selling Point
[Emotional marketing hook, e.g., Enough is enough! In the snowy night, Leopard Head Lin Chong draws his spear, blood splatters the Mountain God Temple!]

IV. Character Biographies
[For each main character include:]

**Character Name**
Visual Image: [Age, appearance details, clothing, key visual markers]
Identity Background: [Social role, background, relationship with protagonist]
Core Tags: [2-4 personality tags, e.g., Enduring, Tragic Hero, Slayer Incarnate]
Personality Traits: [Detailed personality description, character arc]
Signature Line: [Dialogue that best represents the character]

V. Script Outline
Early Stage (Setup): [Establish world, protagonist, status quo]
Middle Stage (Rising Action/Twist): [Conflict development, complication, turning point]
Late Stage (Climax): [Climax, confrontation, emotional peak]
Final Stage (Resolution): [Ending, aftermath, theme resolution]

VI. Script Body

3.2 Script Body Output Format (Per Episode)

Critical: Each episode must strictly use the following format:

Episode X
X-X [Day/Night] [Interior/Exterior] [Scene Name]
Props: [Key props list]
Characters: [Character list]

△ [Establishing Shot/Opening Shot] [Detailed shot description, including camera movement, composition, lighting, atmosphere]
△ [Shot 2 description, including specific camera movement: Push/Pan/Tilt/Track/Crane/Circle/Dolly Zoom/Steadicam/Handheld]
△ [Shot 3 description - continue beat by beat]
Character Name (Inner Monologue): [Inner monologue content]
△ [Reaction shot description]
Character Name (Dialogue/Action Description): [Dialogue content]
△ [Action/Sequence description]
【Subtitle: xxx】[Add when subtitles are needed]
△ [Scene continues]
△ [Flashback] [Flashback scene description]
[Flashback content, described with △ shots]
【Flashback End】
△ [Return to reality, reaction shot]
Character Name (Voice Over): [Voice-over from character not in the frame]
△ [Climactic action sequence]
△ [Establishing Shot] [Closing atmosphere shot]
【Subtitle: xxx】

Script Formatting Rules:

  1. Each shot must start with △ (Triangle + Space)

  2. Shot language must be specific:

    • Shot Size: Establishing/Long/Full/Medium/Close-up/Extreme Close-up
    • Camera Movement: Push/Pan/Tilt/Track/Crane/Circle/Dolly Zoom/Steadicam/Handheld
    • Combinations: Medium Shot Push In, Close-up Circle, Long Shot Top Angle, Fast Push In, Slow Pull Out
  3. Dialogue Markers:

    • Character Name (Inner Monologue) - Character's inner thoughts, visible on screen but mouth not moving
    • Character Name (Voice Over) - Character not currently in the frame, sound comes from off-screen
    • Character Name (Shock/Anger/Joy) - Dialogue with emotion
    • Character Name - Normal dialogue
  4. Special Elements:

    • [Establishing Shot] - Unpopulated atmosphere/establishing shot
    • [Flashback] / [Flashback End] - Flashback sequence
    • [Subtitle: xxx] - On-screen text/title
  5. Action Description Format:

    • Use → for continuous actions: Lin Chong spears Chabai through the chest → Blood sprays → Snow turns red
    • Use emotive language: Overflowing with murderous intent, Eyes burning with anger, Trembling all over
  6. Sensory Details:

    • Visual: Color, light and shadow, composition, atmosphere
    • Auditory: Sound descriptions integrated into the shot
    • Tactile: Cold, pain, texture

4. Create Asset Generation Plan

Categorize and number all visual assets:

CategoryPrefixExampleDescription
CharacterC01-C99C01 Lin Chong·Full Body FrontMultiple angles per character
SceneS01-S99S01 Cangzhou Hay Depot·SnowKey locations
PropP01-P99P01 SpearImportant items

4.1 Asset Construction Workflow

Workflow:

  1. Extract character information from the script/novel
  2. Generate structured character biographies using a large model
  3. Refine visual keywords from character biographies
  4. Organize visual keywords into AI image generation prompts
  5. Iterate to a stable image, then supplement with three-view images

4.2 Asset Prompt Generation Rules

AI-assisted Character Information Extraction Prompt Template:

Please first read and analyze the following short drama script:

[Paste the full script here]

---
You are a professional AI image designer and short drama visual asset analyst.

Based on the story script provided, fully decompose the visual assets needed for this drama, and separately output AI image generation prompts for [Character Assets], [Scene Assets], and [Prop Assets].

Overall image generation style requirements:
The style must match the visual style chosen by the user in step 2 (Realistic/Animation/Ink Wash/Sci-fi/Vintage/Cinematic, etc.). Characters, clothing, scenes, and props must all match the character's identity, historical background, regional environment, social class, and plot atmosphere of the script.

I. Character Asset Requirements

Please list all characters from the script that need images, and generate one Chinese AI image generation prompt for each character.

Character prompts must meet the following requirements:

  1. Must be full-body front view, pure white background
  2. Must explicitly write "standing full body", "front view", "complete from head to toe in frame"
  3. Must include a description of shoes to avoid generating half-body shots
  4. The character must not hold anything
  5. Hands naturally hanging down, body upright
  6. The character has no expression, neutral expression
  7. Must describe the character's height, build, age, hairstyle, hair color, clothing, shoes, face shape, eyes
  8. The character's appearance must match the character's race, historical background, identity and profession, hairstyle, and clothing setting
  9. Adult characters should ideally use a head-to-body ratio close to 2:8, looking good, natural, and photogenic
  10. Even villains should not be designed to look too ugly or extreme; while retaining character traits, try to make the appearance attractive
  11. Do not create child or elderly characters unless the script explicitly requires it

Character Prompt Format:

[Character Name]
Role Positioning:
AI Image Generation Prompt:

II. Scene Asset Requirements

Based on the important locations in the script, list all scenes that need images, and generate an AI image generation prompt for each scene.

Scene prompts must meet the following requirements:

  1. The scene must contain no characters
  2. The scene style must match the visual style chosen by the user in step 2, with a cinematic quality
  3. The scene must match the script's era, region, social class, atmosphere, and plot function
  4. Generate at least three prompts for each important scene:
    • Forward-shot scene image
    • Reverse-shot scene image
    • Side-wide scene image
  5. Forward and reverse shots should be designed based on the characters' likely positions and dialogue direction to facilitate storyboard references
  6. The side-wide image should show the complete spatial relationship for easy understanding of character positions and blocking
  7. Scenes must not contain text, watermarks, posters, character silhouettes, or irrelevant elements

Scene Prompt Format:

[Scene Name]
Scene Purpose:
Forward-Shot Scene Prompt:
Reverse-Shot Scene Prompt:
Side-Wide Scene Prompt:

III. Prop Asset Requirements

Please list all important props from the script, and generate one Chinese AI image generation prompt for each prop.

Prop prompts must meet the following requirements:

  1. Must have a pure white background
  2. Must not contain any characters
  3. Must not contain realistic environmental backgrounds
  4. The prop must be displayed alone, with the main subject clearly visible
  5. The prop's shape, material, sense of age, and wear must match the script's setting
  6. If the prop is related to plot clues, identity twists, suspense, or satisfying moments, reflect its importance in the prompt
  7. Must not contain text, watermarks, or irrelevant decorations

Prop Prompt Format:

[Prop Name]
Prop Purpose:
AI Image Generation Prompt:

Asset Generation Prompt Format:

### [Number] — [Name]

[Style Prefix], [Detailed visual description (Chinese)], [Technical specifications]

**Style Prefix Examples:**
- Chinese ink wash painting style mixed with anime cel rendering
- Cinematic realistic style, dramatic lighting
- 3D Pixar-style animation rendering
- Sci-fi cyberpunk aesthetic, neon lighting

**Character Differentiation:** Use a different color scheme and visual markers for each character to ensure recognition within the chosen art style.

Output Format: An organized list with unique IDs, suitable for copy-pasting into image generators.

4.3 Character Three-View and Four-View Assets

After finalizing the full-body front view of the character, generate the character's three-view images (front, side, back) + a facial close-up, i.e., the "Four Views".

Four Views Prompt:

Generate a full-body three-view image and a facial close-up. (The far left, occupying 1/3 of the space, is a large facial close-up; the right 2/3 contains the front, side, and rear views, 2:8 body proportion), pure white background. The style must be consistent with the visual style chosen by the user.

Note: The prompt must describe the art style, otherwise, the style of different images might be inconsistent.

4.4 Image Aspect Ratio Recommendations

Asset TypeRecommended RatioRecommended Resolution
Character Full-Body Front View9:162K
Character Three-View/Four-View16:9 or 21:92K
Scene Image16:92K
Prop Image1:11K/2K

4.5 Common Character Design Issues

  1. Inappropriate Costume/Style: Period dramas need to pay attention to Hanfu characteristics of different historical periods; costume idol dramas tend to be glamorous; fantasy xianxia should highlight ethereal dreaminess; wuxia should showcase free-spirited elegance
  2. Proportion Issues: Ensure full-body shots, avoid half-body shots, pay attention to body proportions, use the 2:8 body proportion
  3. Aesthetic Issues: Check thumbnail recognition, expression symmetry, color coordination
  4. Other Issues: Do not hold objects, do not include backgrounds, maintain a standing posture

It is forbidden to directly use actors' faces for video production. AI may reference the actor's styling, but must not directly use their likeness images.

5. Generate Seedance 2.0 Storyboard Script

5.1 Seedance 2.0 Storyboard Prompt Generation

For each episode, generate the following:

a. Material Upload List

| Upload Slot | Asset ID | Asset Description |
| Image 1 | C01 | Lin Chong full body front (Character consistency reference) |
| Image 2 | S01 | Hay depot scene reference |

b. Seedance Prompt (Timeline Format, Use Chinese)

[Style description], 15 seconds, 9:16 vertical, [overall atmosphere]

0-3 seconds: [Scene establishment] - [Camera movement], [detailed image description], [atmosphere building]

3-6 seconds: [Subject introduction/plot development] - [Camera movement], [specific action], [emotional expression]

6-9 seconds: [Development/conflict] - [Camera movement], [key action], [detail close-up]

9-12 seconds: [Climax/turning point] - [Camera movement], [emotional explosion], [visual impact]

12-15 seconds: [Ending/final] - [Camera movement], [final image], [lingering emotion]

[Sounds] Generate only sound effects, do not generate music. Do not generate any subtitles. [Specific sound effect description, e.g., footsteps, breaking sounds, gasps]

[Reference] @Image 1 [usage description], @Image 2 [usage description], @Image 3 [usage description]...

c. Ending Frame Description

  • Record the content of the last frame, ensuring continuity to the next episode
  • Includes: Subject, background, lighting, composition, atmosphere

Camera Movement Keywords: Push/Pan/Tilt/Track/Crane/Circle/Dolly Zoom/Steadicam/Handheld

Episode Transition (from Episode 2 onwards): Use Extend @Video 1 by 15s in the starting prompt and upload the previous episode as a video reference.

For more Seedance 2.0 usage tips and templates, please refer to references/seedance-manual.md.

5.2 Material Transition Techniques

  1. Frame Stitching: Take the last frame generated from the previous video as the first frame reference for the next video
  2. Video Extension: Upload the previous clip and ask the model to extend the video in the prompt

5.3 Solutions for Platform Real Human Face Review Restrictions (e.g., Jimeng)

For platforms like Jimeng that have review restrictions on realistic human photos, the following methods can be employed:

  • Local Face Repainting: Use an image model to repaint the face area, reducing the realism threshold
  • In-Platform Image Washing: Make minor modifications to the image within the platform to increase platform trust
  • Face Grid: Add a geometric topology wireframe grid over the face area
  • Face Close-up Occlusion: Use a small square to partially block the face area

5.4 Storyboard Optimization

Significance of Storyboard Images:

  1. Reduce video generation failure rate: Generate a storyboard image for complex extreme shots first, then run the image to generate video
  2. Remediation of flawed shots: Optimize through "frame capture → image editing → re-generate video from image"
  3. Establishing shots and emotional stills: Solve for exquisite static images that video models struggle to generate directly

Recommended Image Editing and Generation Tools:

  • Nano Banana Series
  • GPT Image 2
  • Seedream Series (Jimeng/Doubao image model)
  • LibTV (Aggregation Platform)

Standard Image Generation Prompt Formula:

[Art Style/Medium] + [Shot Size & Perspective] + [Subject Description] + [Environment/Scene] + [Lighting & Color Tone] + [Image Quality & Texture Modifiers]

Storyboard Script Prompt Template:

You are a senior film editor, visual director, and AI video storyboard designer. Based on the script I provide, decompose it into a storyboard script table suitable for AI video production, and in the last column of the table, directly generate the AI image generation prompt for each storyboard.

[Script Content]
{Paste the script here}

## Output Requirements
Output only one Markdown table, no additional explanations. Table columns:
| Storyboard No. | Shot Size | Camera Movement | Image Content | Dialogue | Estimated Duration (seconds) | Storyboard Image Prompt |

## Storyboard Rules
1. Convert literary descriptions from the script into shootable camera language. Do not use un-shootable descriptions like "thinks in mind", "realizes", "suddenly understands", "feels despair"
2. Each storyboard duration should be controlled at 2-5 seconds, with total storyboard duration per episode summing to 13-15 seconds
3. Shot sizes must vary, avoid consecutive identical shot sizes, include at least one shot size transition
4. Action descriptions must be specific and executable, use → to connect continuous actions
5. Dialogue must indicate the source of the sound: use "Character Name says" for on-screen dialogue, "Character Name inner monologue" for inner thoughts, "Character Name voice over" for off-screen sounds
6. Dialogue scenes should use shot/reverse shot blocking; insert reaction shots during long dialogues
7. At scene transitions or emotional turning points, use establishing shots with pre-dialogue to enhance cinematic feel
8. Only write explicit environmental/action sound effects; omit if no clear sound effect is needed
9. Do not generate music, do not generate subtitles

Output Format Description

This workflow produces several types of output. The relationship between them is as follows:

OutputStageFormatPurpose
Script BodyText△ Storyboard script format (Section 3.2)Story blueprint, the foundation for all subsequent work
Asset PromptsAssetsNumbered list format (Section 4.2)Generate character/scene/prop images to be used as reference materials for video generation
Seedance Storyboard PromptsVideoTimeline format (Section 5.1b)Directly paste into Seedance 2.0 to generate video clips
Storyboard Image PromptsEditTable format (Section 5.4)Optimize/repair flawed shots by first generating a storyboard image, then converting it to video

Flow Relationship: Script Body → Extract Assets → Generate Asset Images → Combine Script + Asset Images to Generate Seedance Storyboard Prompts → Generate Video → If flawed shots exist, use Storyboard Image Prompts to repair


Final Output Checklist

Upon completing the entire workflow, the following deliverables must be provided:

  1. Complete Script - Including core gimmick, story synopsis, character biographies, episode outline, script body (△ storyboard format)
  2. Character Asset Prompt List - Full-body front prompt for each character + four-view prompts
  3. Scene Asset Prompt List - Forward-shot/reverse-shot/side-wide prompts for each scene
  4. Prop Asset Prompt List - Individual display prompt for each prop
  5. Seedance 2.0 Storyboard Prompts - One timeline-format prompt per episode, including material upload list and end frame description
  6. Episode Transition Instructions - End frame description for each episode to ensure continuity for the next episode

Quality Self-Checklist

Before completing all outputs, check each item one by one:

Format Check

  • Each shot starts with △
  • Camera movement uses specific terms (Push/Pan/Tilt/Track/Crane/Circle/Dolly Zoom)
  • Dialogue is correctly marked (Inner Monologue/Voice Over/Emotion)
  • Includes sensory details (Visual + Auditory + Tactile)

Content Check

  • Each episode has a clear emotional arc
  • Transitions between episodes are natural (end frame → start frame)
  • Character personality is consistent and shows development
  • Core conflict is resolved
  • Theme is expressed

Technical Check

  • Each episode is controlled within 13-15 seconds (3-7 shots)
  • End frame description is detailed (Scene + Image + Lighting + Composition + Atmosphere)
  • Color planning is clear
  • Sound design is complete (Generate only sound effects, no music, no subtitles)
  • Asset prompt style matches the visual style chosen by the user
  • The [Sounds] section of Seedance prompts contains only sound effects, not background music

Detailed Usage Guide

  • Seedance 2.0 Platform Manual & Prompt Templates: references/seedance-manual.md
  • Shanghai Animation Film Studio Style Preset (Style blocks/Palette/Modeling specifications/Asset templates): references/Shanghai Animation Film Studio Style Guide.md (Must read when this preset is activated)

Skills relacionados