LibTV Image to Video Prompts

Writing video generation prompts that actually work requires balancing temporal physics, camera movement, and audio synchronization. This skill transforms reference images, storyboards, or rough ideas into professional-grade Chinese video prompts optimized for natural language workflows like LibTV. It eliminates the guesswork of duration budgeting and multi-stage action sequencing.

What it does

• Multi-stage Action Sequencing breaks down complex movements into logical phases from trigger to contact feedback to final state.
• Dynamic Camera Direction defines executable paths, focal shifts, and parallax movement while maintaining spatial consistency.
• Audio & Dialogue Budgeting maps dialogue length to video duration to ensure realistic pacing and lip-sync potential.
• Continuity Enforcement locks character counts, lighting directions, and physical material properties across frames.
• Atmospheric Layering builds environmental "texture" using ambient sound, lighting physics, and subtle micro-expressions.

How it works

1. Visual Analysis uses internal tools to verify subjects, lighting, and initial poses from the uploaded reference image.
2. Action Budgeting calculates whether the requested dialogue and movement fit within the target duration (e.g., 5s vs 10s).
3. Prompt Construction generates a structured output covering visual facts, performance, timeline, camera, and sound.

Frameworks & tools

Specifically optimized for LibTV and natural language video generation models. Works with standard image-to-video (I2V) workflows requiring high-precision Chinese prompting.

Why this beats prompting it yourself

Generic prompts often ignore the "cause and effect" of physics or try to squeeze too much dialogue into short clips. This skill calculates realistic timing for speech and movement, preventing the "sped-up" look common in AI video while ensuring camera paths are physically navigable.

Use cases

• Generating a "one-shot" continuous movement prompt for a character performing a complex task like cooking.
• Creating precise prompts for a 7-second dialogue clip with specific timing for three lines of speech.
• Optimizing "bold movement" prompts that require high parallax without breaking spatial logic or clipping through objects.
• Maintaining character and lighting consistency when transitioning from a static starting frame to a high-motion sequence.

Known limitations

Does not perform actual video rendering or media generation. Chinese language output is the default. Cannot guarantee exact lip-sync or model-specific parameter fulfillment without model-side support.