image-gen
由 wells1137 提供
Image & Video Generation简介
Generate images using multiple AI models — Midjourney (via Legnext.ai), Flux, SDXL, Nano Banana (Gemini)
描述
该技能的功能
Routes image generation requests to one of several AI models — Midjourney, Flux (Pro/Dev/Schnell), Ideogram, Recraft, and Nano Banana (Gemini) — based on the request type. A decision tree selects the model by evaluating whether the request is a single image or a series, and whether it needs artistic quality, photorealism, text rendering, vector style, or fast iteration.
Instead of requiring the user to know which model fits their request, the skill applies structured dispatch logic that matches each image type to the model best suited for it.
何时使用
- Generating a multi-frame storyboard with consistent characters
- Creating photorealistic product photos for e-commerce listings
- Designing flat-style app icons and brand assets
- Producing posters and signs with accurate embedded text
- Drafting quick concept visuals for rapid client feedback
示例工作流
User asks: Generate a 3-frame storyboard of a girl discovering a magical forest
- 1
Identifies the request as a multi-frame series requiring character continuity
- 2
Selects Nano Banana (Gemini) for context-aware generation
- 3
Generates frame 1 with a detailed character description and no reference images
- 4
Generates frame 2 using frame 1's URL as a reference image to maintain visual consistency
- 5
Generates frame 3 using frames 1 and 2 as references
Three storyboard frames with a consistent character appearance and scene style across all panels
前置要求
- FAL_KEY environment variable (fal.ai API key)
- LEGNEXT_KEY environment variable (Legnext.ai API key)
安装
clawhub install wells1137/image-gen运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。