菜单
'Construct a production-ready prompt for Seedance 2.0 using the Director\'s Formula. Use when a user has a clear vision and needs to translate it into a genre-aware, intent-driven prompt. Covers genre routing, I2V gate, 30-100 word target, physics language, and anti-slop check.'
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use this skill when designing, creating, or modifying API endpoints; defining contracts between services; evaluating protocol choices (REST, GraphQL, gRPC, WebSockets); versioning or deprecating an API; designing error responses or pagination; or reviewing an existing API for consistency, security, or ergonomics. Signal phrases: "design the API", "create an endpoint", "REST", "GraphQL", "gRPC", "API versioning", "backward compatible", "breaking change", "error response", "pagination", "rate limiting", "API contract", "OpenAPI", "Protobuf", "webhook", "idempotency", "deprecation", "endpoint structure". Also activate when API responses expose database column names, collection endpoints return unbounded results, or breaking changes are being introduced to active versions without a migration plan. Do NOT shape the API contract around implementation convenience — design for the consumer and the future.
Use this skill when designing system boundaries, data flows, component interfaces, or making structural technology decisions. Activated when the user is initializing a new project, service, or major feature; designing data flow between components; breaking apart a legacy system; defining folder structures or project organization; evaluating core technology or framework choices; or when the current system fails to scale. Examples: "how should I structure this?", "should this be a microservice?", "design the architecture for X", "what's the best way to organize this codebase?", "how should data flow between these components?". Do NOT use for pure implementation (use coding skill) or infrastructure provisioning (use devops-infra skill).
Use this skill when analyzing a brand's positioning, auditing an existing website's brand communication, identifying mismatches between brand intent and visual/copy execution, or defining the brand strategy for a new business from scratch. Activated when the user asks to "audit the brand", "diagnose the brand's positioning", "align copy and visuals to the brand", "define the brand perception", "what phase is this brand in", or "how does the brand extend to visuals and animations". Do NOT use for writing actual copy (use copywriting), editing copy (use copy-editing), or configuring animations without brand direction (use cinematic-motion).
Use this skill when designing or implementing highly animated, immersive, and cinematic web user interfaces. Activated when the user asks to build animations, integrate GSAP, use ScrollTrigger, build Three.js R3F models, implement video-scrubbing, image mask scaling, modulo-based infinite sliders, bento grid lighting, or premium interactive web layouts. Do NOT use for standard, non-animated web interfaces or pure backend APIs.
Use this skill when building cinematic, animated, or immersive web experiences. Covers scroll-driven storytelling (GSAP, ScrollTrigger, SplitText, video scrubbing, parallax, mask reveals), 3D product showcases (R3F, Zustand, studio lighting, model switching, video textures), layout composition (bento grids, masonry, power lines, overlapping typography), and register-aware creative direction. Activated when the user asks to "build a cinematic landing page", "add scroll animations", "create a 3D product viewer", "scrub video on scroll", "build a bento grid with scroll reveals", "animate text on scroll", "add parallax effects", "create a mask reveal", "build an infinite slider", "implement model switching", "add studio lighting to 3D scene", "map video textures on 3D screens", or any task producing premium animated web surfaces. Also activated for FFmpeg keyframe encoding, GSAP ScrollTrigger configuration, SplitText typography, gltfjsx model conversion, Zustand state for 3D, or scroll-bound timeline orchestratio
Use this skill when writing, modifying, or generating code of any kind. Activated when the user asks to build features, implement functions, create components, write endpoints, set up forms, wire things together, or produce any working implementation. Trigger phrases: "build this", "implement", "write the code for", "create a function that", "add this endpoint", "make it work", "update this code to", "wire this together". Do NOT use for: pure architectural decisions before implementation starts (use architecture skill), root-cause debugging of existing broken code (use debugging skill).
| name | seedance-prompt |
| description | 'Construct a production-ready prompt for Seedance 2.0 using the Director\'s Formula. Use when a user has a clear vision and needs to translate it into a genre-aware, intent-driven prompt. Covers genre routing, I2V gate, 30-100 word target, physics language, and anti-slop check.' |
This skill translates a user's creative vision into a production-ready prompt for Seedance 2.0. It uses a genre-aware, intent-driven workflow based on community-proven best practices.
Seedance 2.0 is an AI director. Tell it WHAT you want and HOW it should FEEL, not every micro-detail of HOW to execute it. Trust the model.
All prompts should follow this structure. The default target length is 30–100 words.
[Subject] + [Action] + [Scene] + [Camera] + [Style] + [Constraints]
First, ask the user to identify their content category. This determines the prompt's skeleton, length, and emphasis.
| Genre | Lead With | Length Target | Key Sections |
|---|---|---|---|
| Product/E-commerce | Subject | 30-50 words | Subject, Camera, Style |
| Lifestyle/Social | Action | 40-60 words | Action, Scene, Style |
| Drama/Narrative | Scene | 60-100 words | Scene, Action, Camera |
| Music Video | Style | 50-80 words | Style, Action, Camera |
| Landscape/Travel | Scene | 30-60 words | Scene, Camera, Style |
| Commercial/Brand | Style | 40-70 words | Style, Scene, Action |
| Anime/Artistic | Style | 50-90 words | Style, Action, Scene |
| UGC (User-Generated) | Action | 20-40 words | Action, Scene, Camera (handheld) |
For detailed templates per genre, see [ref:genre-guides].
Ask if the user has reference images. This fundamentally changes the prompt.
For detailed I2V guidance, see [ref:i2v-guide].
Construct the prompt using the Subject + Action + Scene + Camera + Style + Constraints formula, tailored to the chosen genre.
Instead of vague mood words, use concrete physics and intensity descriptors.
tires smoke, gravel sprays, sweat flies off in slow motion, dust erupts violently, fat renders slowly, oil bubbles form.slowly, dramatically, violently, gently, frantically. The model cannot infer intensity from images; it must be stated.Scan the prompt for slop words that conceal a lack of specific information. This now includes action and camera slop.
beautiful, stunning, epic, amazingdynamic, energetic, cinematic camera movement, cool transitionReplace all slop with measurable, observable details.
Maintained by Emily (@iamemily2050)