概述
Stable Diffusion 是由 Stability AI 發布的深度學習文字轉圖像模型。與許多專有 AI 藝術生成器不同，Stable Diffusion 是開源的，用戶可以在本地硬體上運行，也可以透過各種雲端介面運行。憑藉其靈活性和龐大的社區插件及模型生態系統，它已成為 AI 藝術運動的基石。
主要能力
- 文字轉圖像: 將複雜的描述性提示轉化為高解析度的數位藝術作品、照片或插圖。
- 圖像到圖像： Use an existing image as a structural guide to generate a new version with different styles or elements.
- 補繪與刪減： Edit specific parts of an image or extend the canvas beyond its original borders while maintaining visual consistency.
- Custom Model Training: 透過 LoRa 和 ControlNet 等技術，使用者可以訓練人工智慧識別特定人物、藝術風格或精確姿勢。
最適合
穩定擴散 is ideal for digital artists, game developers, and hobbyists who require granular control over their output. It is particularly suited for those who prefer an offline environment for privacy or those who want to experiment with advanced parameters and custom checkpoints.
限制和定價
Because it is open-source, the software itself is free to download. However, running it locally requires a powerful GPU (typically NVIDIA with high VRAM). For those without hardware, various third-party hosting services offer paid subscriptions or credit-based pricing. Users should be aware that the learning curve is steeper than simpler tools like Midjourney.
免責聲明：功能、版本和第三方定價可能會有所變更。請造訪 Stability AI 官方網站查看最新詳情。
資訊可能不完整或過時；請在官方網站上確認詳細資訊。
