Date: 30 Mai; 6, 13, 20 June;
Time: 2-5 pm
BITZ unibz fablab now offers a course to learn ComfyUI,
a powerful free and open-source software for AI-powered image, video, 3D-model, sound, voice, text… generation.
The most powerful application for generative AI, open-source and node-based!
ComfyUI gives you more control and creative freedom than any other tool!
- Visual control: Unlike many other AI- tools, ComfyUI allows for a modular, visual workflow – ideal for designers and artists who want to directly influence image composition and style.
- Open Source = Open Mind: You’ll not only learn how to use a tool, but also understand how AI image generators work – with no black box.
- Future skill: AI is becoming increasingly important in the creative industry – from concept art and fashion to media and product design. With ComfyUI, you’ll gain both technical know-how and creative freedom.
What to expect
- Introduction to AI image generation with Stable Diffusion
- Hands-on work with ComfyUI
- Creative mini-projects & experiments
- No prior knowledge required
Sign up now!
Course is free and open for everyone!
Spaces are limited!
Registration mail to: Stephan.Pircher@unibz.it
Let’s explore together how art and technology can open up new paths to creativity.
We look forward to seeing you at Bitz!
—————————————————–
Learn more about ComfyUI:
Official website: comfy.org
ComfyUI Beginner tutorial: https://openart.ai/workflows/academy
Showreel AI 2024: https://youtu.be/Y5Isghixqq4?si=Of2W8eAjlbypuR1e
This is a list of some possible ComfyUI Workflows
- Creative Workflows
- Text-to-Image (TTI): Generate an image from a text prompt
- Image-to-Image (ITI): Use an image as a template for variations
- Inpainting: Selectively replace specific areas of an image with new content
- Outpainting: Expand an image beyond its original borders
- ControlNet Variations: Achieve precise control through additional image information
- Pose-to-Image: Human pose determines the content of the image
- Depth-to-Image: Depth maps influence the structure of the image
- Scribble-to-Image: Use a sketch as the foundation for the image
- Canny-to-Image: Use edge detection images as a control element
- Style Transfer: Apply the style of one image to another
- QR-Code Diffusion: Creatively and legibly embed QR codes into images
- LoRA Mixing: Combine multiple style models
- Model Merging: Merge two or more AI models
- Prompt Matrix: Automatically combine and compare multiple prompts
- Upscaling: Upscale images with AI-powered sharpening
- AnimateDiff: Combine individual images into animations
- Prompt Morphing: Smoothly blend between prompts
- GIF Creation: Generate animated sequences from created images
- Technical & Experimental Workflows
- GLIGEN: Place objects precisely using bounding boxes
- T2I Adapter: Alternative to ControlNet with lower data requirements
- Noisy Latent Composition: Combine multiple image subjects into one
- Semantic Segmentation: Analyze images into semantic zones
- Normal Map Generation: Create structural information for 3D applications
- Depth Map Generation: Generate depth information for VR/3D output
- Audio-to-Image: Use sound or music to influence image generation
- Terminal Node (Shell, Python): Run arbitrary scripts directly
- Prompt Comparison Automation: Automatically compare text prompts
- Batch Generation from JSON/CSV: Data-driven image production
- Language, Text & Voice
- Prompt Engineering via LLM: AI creates or improves prompts
- Captioning with CLIP + LLM: Automatically describe images
- TTS (e.g., Bark, ElevenLabs): Convert text to natural speech
- STT (e.g., Whisper): Convert speech to text
- Voice Cloning (RVC): Imitate voices from a few samples
- Voice-to-Image: Generate images from spoken text
- 3D & AR/VR Workflows
- TTI + Depth2Mesh: Turn text descriptions into 3D models
- Normal2Geometry: Convert normal maps into geometry
- NeRF: Reconstruct 3D models from 2D images
- Volumetric Rendering: Create 3D light volumes for realistic depth
- SMPL Avatars: Human avatars with movement/poses
- Multi-View Consistency: Generate consistent images from different viewpoints
- 360° Rendering: Create spherical panoramas or VR images
- ComfyUI + Unity Integration: Control interactive 3D scenes
- Utility & Automation
- Scheduled Generation: Automate generation at set intervals
- Random Style Switcher: Apply styles randomly
- File-Watcher Nodes: Respond to file changes in the system
- EXIF/Metadata Injection: Automatically insert image metadata
- Image Comparison (MSE, SSIM): Measure differences between images
- Template-based UI Mockups: Automatically populate design templates
- High-End / Advanced Workflows
- Dreambooth Training / LoRA Training: Train your own models from images
- Real-Time Video2Video: Transform live camera images with AI
- Stable Video Diffusion (SV3D): Create moving scenes from images and prompts
- Arduino Control via ComfyUI: Use sensor data to control AI output
- Robot Integration: Connect generative AI with the physical world
Multi-User Queue Server: Multi-user management with task queues