A professional-grade prompt engineering engine designed specifically for ByteDance's Seedance 2.0 AI video diffusion model.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install seedance
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install seedance using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Seedance 2.0 Prompt Generator?
The Seedance 2.0 Prompt Generator is a specialized tool within the Openclaw Skills ecosystem that enables users to synthesize production-ready video prompts. It utilizes a rigorous prompt architecture consisting of six layers—Subject, Action, Camera, Style, Audio, and Constraints—to minimize model drift and maximize visual fidelity. This skill is particularly effective for creators needing to maintain character consistency or execute complex camera movements that standard text-to-video tools often struggle with.
By prioritizing descriptive visual language over proprietary brand names, the skill ensures high success rates against content moderation filters while delivering cinematic results. Whether you are working with text-to-video (T2V) or image-to-video (I2V) workflows, this skill provides the technical framework necessary for high-end AI cinematography.
Seedance 2.0 Prompt Generator Use Cases
- Generating professional cinematic scenes with specific lens requirements (e.g., 24mm wide vs 85mm telephoto).
- Creating multi-shot narratives that maintain character identity across different scenes using @Image reference tags.
- Developing high-speed action sequences that require physics-accurate motion and environmental anchoring.
- Translating complex IP concepts into descriptive prompts that bypass restrictive content moderation filters.
- Designing synchronized audio-visual experiences by leveraging Seedance's dual-channel stereo generation.
How Seedance 2.0 Prompt Generator Works
- The user provides a scene concept, image, or video reference to the AI agent.
- The skill identifies the optimal workflow: Text-to-Video (T2V), Image-to-Video (I2V), or Multi-Reference (R2V).
- It constructs a layered prompt following the strict Subject → Action → Camera → Style → Audio → Constraints hierarchy.
- Visual descriptions are refined using the Translation Layer to replace banned brand terms with high-detail physical attributes.
- The skill appends 3-5 targeted negative prompts to prevent common artifacts like warping, extra limbs, or branding.
- A final technical recommendation is provided, including aspect ratio and test duration (5s/10s).
Seedance 2.0 Prompt Generator Setup
To utilize this prompt generator within your Openclaw Skills workflow, ensure your agent has access to the Seedance skill definition. You can trigger the generation using natural language commands.
# Example trigger for a cinematic prompt
"Create a Seedance 2.0 prompt for a wooden creature walking through a bioluminescent forest @Image1"
Access the resulting prompts through Jimeng AI, Doubao, or the Volcano Engine experience center to generate the final video assets.
Seedance 2.0 Prompt Generator Data Schema & Taxonomy
The skill structures data based on a precise architectural taxonomy to ensure model compatibility.
| Component | Type | Description |
|---|---|---|
| Subject | String | Physical details, clothing, and distinguishing features. |
| Action | List | Chronological beat-by-beat verb phrases in present tense. |
| Camera | Object | Shot size, movement, angle, and lens focal length. |
| Reference Tags | Array | @Image[1-9], @Video[1-3], @Audio[1-3] pointers. |
| Constraints | List | Physics guardrails, timing, and negative prompt bans. |
Seedance 2.0 Prompt Generator Advanced Features
- Continuous Shot Lock: A technique that forces temporal coherence by treating the generation as a single fluid motion path.
- Physics-Motivated Camera: Replaces generic movement terms with physically grounded verbs like slingshot, dive, or dart.
- Environmental Anchoring: Strategically places static geometry in the prompt to create clean parallax and reduce subject blur.
- Scale Progression Arc: Automates the transition from Macro to Micro and back to Macro for cinematic payoff.
- Sensory Render Instructions: Swaps vague mood words for computable physical effects like heat haze or grit snapping.
SKILL.md
Loading
Related Openclaw Skills
Janitor
A background maintenance agent that manages memory consolidation, archives stale data, and ensures workspace organization.
MML (Metaverse Markup Language)
A powerful tool for building interactive 3D scenes and metaverse experiences using the Metaverse Markup Language (MML).
Font Interceptor
Extract and download TTF or OTF font files from any website automatically using MSCHF Font Interceptor.
Clickbait Engine
A high-performance engagement engine designed to transform content into viral, click-worthy titles and hooks using proven social media psychological triggers.
Janitor
An automated maintenance agent designed to keep your AI workspace organized, healthy, and synchronized.
YouTube Downloader
A high-performance utility for AI agents to download and register high-quality YouTube videos as local assets.
Featured*
