We’ll show you what’s possible with generative image AI today.
31.7.2024 16:00h @ fablab unibz
Discover our locally installed open source AI and use it for free. Understand how AI works and control every single pixel of your generated AI images.
Generate realistic images with comfyUI:
- Text to Image
- Image to Image
- Inpaint/outpaint
- Deep fake/face swap
- Image upscale
- Image to 3D
- Image to Video
Let’s explore how we can use AI in art, design, photography and architecture.
Download from GitHub – comfyanonymous/ComfyUI:
The most powerful and modular stable diffusion GUI, api and backend with a graph/nodes interface.
Examples
Image control when generating
Consistent images generation with „controlnet“
Open pose model to guide the AI
For photographers
Remove Background from images
For product design and architecture
Work with your 3d models
Get ready with our well selected video tutorials for you.
- OpenArt ComfyUI Academy
- ComfyUI – Getting Started : Episode 1 – Stable Diffusion AI Art generation
- ControlAltAI – YouTube
- ComfyUI: Advanced Understanding
- ComfyUI: Installation | Stable Diffusion | Deutsch | Englische Untertitel
deep dive….
- Stable Diffusion explained (in less than 10 minutes) (youtube.com)
- https://www.youtube.com/@latentvision
- https://www.youtube.com/@TheFutureThinker
- How AI ‚Understands‘ Images (CLIP) – Computerphile (youtube.com)
- Understanding Variational Autoencoders (VAEs) | Deep Learning (youtube.com)
- Why Does Diffusion Work Better than Auto-Regression? (youtube.com)
Paper:
- Deep Unsupervised Learning using Nonequilibrium Thermodynamics (arxiv.org)
- Improved Denoising Diffusion Probabilistic Models (arxiv.org)
- Denoising Diffusion Probabilistic Models (arxiv.org)
- Diffusion Models Beat GANs on Image Synthesis (arxiv.org)