Stable Diffusion
by Stability AI
Open-source text-to-image AI model that powers a generation of creative tools
Visit Product
489 upvotes
676 views
About
Stable Diffusion is an open-source latent diffusion model for AI image generation developed by Stability AI, with foundational research from CompVis at LMU Munich and Runway. Unlike proprietary image generators, Stable Diffusion's open weights allow anyone to run it locally, fine-tune it on custom datasets, and build commercial products without licensing fees — making it the foundation of the open AI art ecosystem.
The model generates high-quality images from text descriptions (text-to-image), transforms existing images based on text prompts (image-to-image), fills in missing parts of images (inpainting), and extends images beyond their original borders (outpainting). The vibrant open-source community has created thousands of fine-tuned variants, LoRA adaptations, and specialized checkpoints for everything from photorealistic portraits to anime art.
Stable Diffusion powers dozens of popular applications including Automatic1111 WebUI, ComfyUI, Invoke AI, and is the backbone of commercial products from Canva, Adobe Firefly, and many others. Its architecture has influenced virtually every subsequent image generation model and represents a pivotal moment in democratizing AI creative tools.
The model generates high-quality images from text descriptions (text-to-image), transforms existing images based on text prompts (image-to-image), fills in missing parts of images (inpainting), and extends images beyond their original borders (outpainting). The vibrant open-source community has created thousands of fine-tuned variants, LoRA adaptations, and specialized checkpoints for everything from photorealistic portraits to anime art.
Stable Diffusion powers dozens of popular applications including Automatic1111 WebUI, ComfyUI, Invoke AI, and is the backbone of commercial products from Canva, Adobe Firefly, and many others. Its architecture has influenced virtually every subsequent image generation model and represents a pivotal moment in democratizing AI creative tools.
Product Features
- Open-source weights available for local and commercial use
- Text-to-image, image-to-image, inpainting, and outpainting
- Fine-tuning on custom datasets with DreamBooth and LoRA
- Thousands of community fine-tuned models on Hugging Face and CivitAI
- SDXL (Stable Diffusion XL) for higher resolution and quality
- ControlNet for precise control over composition and pose
- Runs locally on consumer GPU hardware (6GB VRAM minimum)
- API access via Stability AI platform
- Supports multiple samplers and schedulers for image quality
- Active community ecosystem with extensions and tools
- Text-to-image, image-to-image, inpainting, and outpainting
- Fine-tuning on custom datasets with DreamBooth and LoRA
- Thousands of community fine-tuned models on Hugging Face and CivitAI
- SDXL (Stable Diffusion XL) for higher resolution and quality
- ControlNet for precise control over composition and pose
- Runs locally on consumer GPU hardware (6GB VRAM minimum)
- API access via Stability AI platform
- Supports multiple samplers and schedulers for image quality
- Active community ecosystem with extensions and tools
About the Publisher
Stability AI was founded in 2020 by Emad Mostaque and has been headquartered in London. The company raised $101 million in funding in 2022 and was valued at $1 billion. Stability AI released Stable Diffusion as open-source in August 2022, a decision that fundamentally changed the AI image generation landscape and made advanced image AI accessible to researchers, developers, and hobbyists worldwide. Emad Mostaque stepped down as CEO in 2024, and the company has undergone significant restructuring.