Diffusion

📄️ Text-to-Image Generation

Generate high-quality images from text descriptions using diffusion models in Transformer Lab.

Transformer Lab supports downloading and using LoRA (Low-Rank Adaptation) adaptors from Hugging Face to enhance your diffusion models with specialized styles, subjects, or techniques. These adaptors can dramatically change the output style while using the same base model.

📄️ Image-to-Image (img2img) Generation

Transform existing images using diffusion models with text guidance in Transformer Lab.

📄️ Image Inpainting

Overview

📄️ Using ControlNets

ControlNets are powerful tools that allow you to control the composition and structure of generated images by using reference images as guides. They enable precise control over poses, edges, depth, and other structural elements in your generated images.

📄️ Managing Image History

Transformer Lab automatically saves all your generated images from Text-to-Image, Image-to-Image, and Inpainting workflows, creating a comprehensive history that you can manage, organize, and even convert into datasets for future training.

📄️ Uploading Image Datasets

Downloading Datasets from Hugging Face