Zum Inhalt springen
Kitbay
Skill

stable-diffusion-image-generation

MITAktualisiert am 11. Okt. 202614.431GitHub-Sterne

A reference skill from Orchestra Research's AI research skills collection for generating images with Stable Diffusion through Hugging Face's Diffusers library in Python.

Was der Skill vermittelt

A reference skill from Orchestra Research's AI research skills collection for generating images with Stable Diffusion through Hugging Face's Diffusers library in Python. It shows the agent how to install the dependencies and write code for text-to-image with SD and SDXL pipelines, image-to-image, inpainting, ControlNet conditioning on edges, poses or depth, and LoRA style adapters. It explains pipelines, schedulers, guidance settings, seeds for repeatable output and negative prompts, and covers memory-saving options such as CPU offloading, attention slicing and loading models at lower precision. The models run locally in the user's own Python environment with PyTorch.

Funktionen

  • Write Diffusers code for text-to-image and image-to-image
  • Set up inpainting, ControlNet and LoRA pipelines
  • Reduce memory use with offloading and attention slicing