How to train your imagegenerator

Workshop Description

Providers of AI tools promise unprecedented productivity gains through generative AI. However, the resulting business models, which are almost uncreative, such as copying existing social media platforms and feeding them with only generated video material, show that the application examples provided by the companies are anything else than innovative. The proprietary development of image generation engines is currently also taking place from a very technically oriented perspective, skipping an important basis for discussion about image aesthetics in general. In this course, we want to ask ourselves precisely this question: what distinguishes a “good” image from a “bad” image? We will explore which areas of application can still be developed and, most importantly, how such image generation tools can be usefully integrated into the design process. To this end, we will take a closer look at the open source image generation tool InvokeAI, which we will use to test different image generation models and fine-tune them ourselves in a training process.

Educational Goal

• How do AI image generators work?
• What roles can we take as designers in generative AI processes?
• How can we gain new insights into our creative processes by shifting perspectives – mediated by AI?
• How can AI hallucinations or misunderstandings enrich the design process?

What you need for this workshop

Laptop

Requirements

Basic knowledge of prompting with stable-diffusion models could be helpful, but is not necessary.
Also available for International students

Where is it located?

Computer or Laptop needed