lllyasviel/sd-forge-layerdiffuse
Sd Forge Layerdiffuse
该项目是 Forge 界面的一款 Stable Diffusion WebUI 扩展，实现了潜在层扩散模型。它作为一个 AI 透明图像生成器，旨在生成带有 Alpha 通道的图像，从而实现前景元素与背景的自动分离。
该软件支持生成可合成的兼容前景和背景层，以便进行后期编辑。它支持基于提供的背景生成透明前景、生成适合现有透明前景的背景，以及同时生成前景、背景和混合图像。
Features
- Latent Transparency Generation - Creates images with latent transparency to automatically produce separate foreground and background layers.
- AI Transparent Image Generators - Implements a system that produces images with alpha channels to separate foreground elements from backgrounds automatically.
- Layered Diffusion Models - Implements a specialized latent layer diffusion model for multi-layer image composition and transparency.
- Generative Layer Compositing - Simultaneously produces foreground, background, and blended images through a generative layer compositing process.
- Foreground Layer Generation - Produces transparent foreground layers designed to blend naturally with provided background images.
- Generative Foreground Isolation - Creates transparent foreground layers designed to blend seamlessly with provided background images.
- Conditional Backgrounds - Generates compatible background layers using an existing transparent foreground as a visual condition.
- WebUI Extensions - Ships as an extension for the Forge web interface to add layer-based generation and blending.
- Stable Diffusion Tools - Provides a specialized tool for generating transparent layers within the Stable Diffusion ecosystem.
