Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Added _meta.json for metadata and references/camera-and-styles.md with camera/style notes. - Expanded modes to include "Text-only," multi-segment stitching (>15s videos), and scenario-specific strategies. - Added comprehensive section on IP/copyright avoidance and moderation-safe prompting. - Introduced detailed best practices for dialogue, sound separation, and reference image style matching. - Enhanced and clarified special-case instructions (e.g., multi-segment production, product ads, long tracking shots). - Included scenario-specific tables for quick reference to recommended techniques and modes.
What This Skill Does
Generates structured English prompts for Seedance 2.0 and Seedance 2.0 Fast video generation. Covers mode selection (text-only, first/last frame, all-reference), asset mapping for up to 12 mixed inputs, and timecoded beat planning. Includes strategies for multi-segment stitching, character replacement, beat sync, and IP-safe original character design.
The structured output format with explicit asset mapping and timecoded beats reduces prompt rejections and retakes compared to freeform descriptions.
When to use it
- Building a 10-second product ad from a single product photo
- Extending a video clip past the 15-second generation limit using segment chaining
- Swapping a character in a reference video while preserving the original camera work
- Syncing a music video's cuts and motion to an uploaded audio track
- Writing moderation-safe prompts for original characters without franchise references
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Create a 10-second product ad for a glass perfume bottle using my reference photo
- 1Declares mode as All-Reference with @image1 bound to the product as identity anchor
- 2Writes timecoded prompt: 0-3s studio reveal with hero lighting, 3-7s 360 degree rotation showing glass reflections, 7-10s close-up on label with soft bokeh background
- 3Specifies material rendering: glass reflections, clean gradient background, no hands in frame
- 4Adds negative constraints: no watermark, no on-screen text, no hands
- 5Sets generation settings: 10s duration, 16:9 aspect ratio
A complete Seedance 2.0 prompt block with mode declaration, asset mapping, timecoded final prompt, negative constraints, and generation settings, ready to paste into JiMeng
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.