ai-video-prompt

What This Skill Does

The ai-video-prompt skill is a specialized AI Video Prompt Builder designed to facilitate the creation of high-quality, long-form videos by leveraging a structured 'Keyframe Image + Multi-Segment Video' workflow. It simplifies the complex process of video generation by breaking down long videos (30-60 seconds) into manageable 5-second segments. This skill ensures visual consistency, logical narrative flow, and stable character/prop states across segments, which is critical for preventing common AI generation issues like character morphing, floating props, or illogical scene jumps. It is optimized for top-tier platforms such as Seedance (即梦AI), Kling AI, Runway, and Sora, outputting clear and precise Chinese prompts.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/atmosphere16happy/ai-video-prompt

Use Cases

• Professional Content Creation: Produce consistent 30s-60s narrative videos for social media or short films.
• Character Consistency: Ensure the same character remains recognizable across multiple camera angles and actions.
• Structured Storytelling: Plan complex narratives involving emotional arcs, such as the 'baby getting a vaccination' example, where prop states and expressions need to shift logically over time.
• Commercial Ad Production: Generate product-focused sequences where the object and its state must be clearly defined in every frame.

Example Prompts

1. "帮我规划一个30秒的视频：一个女孩在公园练瑜伽，从清晨到傍晚，需要包含6段视频，确保人物服装和背景光线的连贯性。"
2. "我需要生成一段60秒的短片，讲述一只猫咪从发现毛线球到玩耍再到睡觉的过程，请提供13张关键帧的Prompt描述。"
3. "请分析我的脚本：一个探险者在森林中寻找宝藏，要求每5秒为一个片段，请检查我的道具状态是否符合逻辑变化规范。"

Tips & Limitations

• Adhere to the Rule of 7/13: For a 30s video, strictly use 7 keyframes; for 60s, use 13. Remember that the end frame of segment N must serve as the start frame for segment N+1.
• Maintain Consistency: Always keep the core physical attributes of your subject (clothing, hair, face) identical across all keyframe prompts to avoid 'ghosting' or morphing.
• Logical State Changes: Props should not disappear magically. Map out your prop status in a table to ensure transitions (e.g., 'held' -> 'dropped' -> 'on ground') make sense for the viewer.
• Limitations: While this skill excels at planning and prompting, the final output quality is dependent on the generation platform's specific model capabilities regarding long-term temporal coherence.