Skills / nano triple
nano triple
3 images, one prompt, instant A/B/C comparison. Style presets, iterative refinement, image editing, aspect ratios. Powered by Gemini Nano Banana Pro. OpenClaw skill.
Installation
Kompatibilitaet
Beschreibung
Nano Triple
3 images, one prompt, instant A/B/C comparison. Pick the winner or refine any version.
Install
clawhub install nano-triple
Or copy SKILL.md to ~/.openclaw/skills/nano-triple/SKILL.md
What's New in v3.0.0
- Nano Banana 2 - New default model. Faster, cheaper ($0.067/img), 4K output, ultra-wide ratios
- Imagen 4 - Text-to-image from $0.02/img (Fast) to $0.06/img (Ultra), plus upscaling at $0.003/img
- 14 aspect ratios - Including ultra-wide 1:8 and 8:1 (Nano Banana 2 exclusive)
- Mask-free inpainting - Describe changes in text, no mask needed
- Layout-aware outpainting - Extend images with intelligent scene continuation
- Character consistency - Up to 14 reference images for consistent characters/objects
- Google Image Search grounding - Generation informed by real search results
- Configurable thinking - Control reasoning depth before generating
- 0.5K preview mode - 512px drafts for ultra-fast iteration
- Batch API - 50% cost reduction with 24hr turnaround for large jobs
Models
| Model | Cost | Max Res | Editing | Best for | |-------|------|---------|---------|----------| | Nano Banana 2 (default) | $0.067/img | 4K | Yes | Fast iteration, most tasks | | Nano Banana Pro | $0.134/img | 4K | Yes | Studio-quality output | | Imagen 4 Fast | $0.02/img | 2K | No | Cheapest text-to-image | | Imagen 4 Ultra | $0.06/img | 2K | No | Highest quality text-to-image |
Features
- 3 parallel variations - Same prompt, three distinct takes from the model's natural randomness
- 4 models - Nano Banana 2, Nano Banana Pro, Imagen 4 Fast, Imagen 4 Ultra
- Style presets - Photorealistic, illustration, watercolor, minimalist, gradient, retro, neon, sketch, anime, pixel-art, oil-painting, 3d-render
- 14 aspect ratios - 1:1, 1:4, 1:8, 2:3, 3:2, 3:4, 4:1, 4:3, 4:5, 5:4, 8:1, 9:16, 16:9, 21:9
- 4 resolutions - 0.5K (preview), 1K, 2K, 4K
- Iterative refinement - "I like B but make it more colorful" generates 3 new variations
- Image editing - Edit existing images with natural language
- Mask-free inpainting - Change parts of an image by describing what to change
- Outpainting - Extend images beyond their borders
- Character consistency - Maintain consistent characters across generations with reference images
- Upscaling - Imagen 4 upscaling at $0.003/image
- Batch API - 50% cost reduction for large jobs
Requirements
GEMINI_API_KEY- Get one at Google AI Studio
License
MIT
Aehnliche Skills
skyll
A tool for autonomous agents like OpenClaw to discover and learn skills autonomously
meihua yishu
梅花易數 - 適用於 Claude、ChatGPT、Gemini、DeepSeek 等 AI/LLM 的占卜技能
Apple Hig Designer
A Agent Skill for designing professional interfaces following Apple Human Interface Guidelines 一项用于设计苹果前端交互界面的代理技能
thesis figure skill
🎓 Claude Skill: Paste your paper text, auto-generate publication-ready LaTeX/TikZ diagrams. 粘贴论文文案,自动生成学术级配图。
ultimate seo geo
The definitive SEO + GEO skill for Claude. Full site audits with scored findings, AI search optimization (Google AI Overviews, ChatGPT, Perplexity), schema generation, E-E-A-T assessment, and 20 diagnostic scripts. Three modes: Audit → Plan → Execute.
TeslamateCyberUI
一个现代化的 Tesla 数据可视化面板,连接 TeslaMate 数据库,采用赛博朋克风格设计。