AI workshop: dreamIMG ✩₊˚.⋆☾⋆⁺₊✧ | BA & MA – WiSe 2023/24



Tuesdays 10.00 a.m.
Room 202
Students from the Comfort project

Kick off: 17 Okt.l 10.00h
Language: German & English

LB Olivia Heung
• Prof Hans (crit)

dreamIMG ✩₊˚.⋆☾⋆⁺₊✧

It can write, it can draw, it can speak, it can definitely play chess. Artificial intelligence seems startlingly intelligent. Reports of its smartness are vast and wide, with each model released promising to be more efficient, more powerful, more real, more imaginative than we can imagine! But after a while, some vague messages and a few foggy jpegs exchanges later, we get the impression that something is amiss. These nonsensical collages seem more reminiscent of dreams than visions for the future. The mismatch between our expectation for perfection and the reality of its low-res output suggests that AI is more like us than we think – clumsy and possibly a little less perfect than we’d like to admit.

For this workshop, we will inhabit this blundering space and explore AI or ANN (Artificial Neural Networks) as toolboxes for dreaming. What does it mean to design with hazy intent? To use a fuzzy machine that doesn’t produce crisp results but rather blurry artifacts embalmed in a lossy jpg? Through the process of text prompting, image generation, and video interpolation, we will speculate together whether AI in its current state is a machine better suited for imprecision and highly processed approximating; a tool for introspection that doesn’t exactly think for us but instead acts like a sieve or dreamcatcher we pour our imagination through. One that unearths hidden feelings, memories, favorite things we wouldn’t explore otherwise and samples them in a visual scene.

We will introduce typical AI tools for text and image generation such as ChatGPT, DALL-E, Midjourney, and Stable Diffusion as well as touch on other generators for speech and sound. We will keep track of our process through a ‘dream log’ and also dabble in AI imaging and video tools in Photoshop, Aftereffects, and RunwayML to animate them. At the end of the session students will produce a ‘dream’- a fleeting AI generated 5 second video of their exploration.

1. A Dream Log – a record of your inspirations, images, moods, interventions, the ingredients for your dream
2. A Dream – 5 second video incorporating your image generations and AI assisted video editing


TUES 17 OCT 10.00h:
A quick introduction within Prof. Hans first class session.

TUE 31 OKT 10:00-16:30h, Workshop 1: dreamCatching, SEE BRIEF
Presentation on AI Concepts and its current application in art and design.
Concepts: Mind Mapping, Prompt Writing, Basics of Text-Text/Image-Image/Text-Image Generation
Tools: DALL-E, Midjourney, Stable Diffusion, ChatGPT
Goal: Produce at least 5 AI generated images and one AI generated text of a descriptive scene
*Before class, collect at least 10 images or clips of text that will inspire your object of comfort. These images can be taken from your field trip, your favorite inspirations, a mood or style, a poem you like etc.

TUE 7 NOV 10:00-16:00hWorkshop 2: dreamIMG
Concepts: AI assisted Image Editing, Outpainting, Other AI generators beyond Image and Text
Tools: DALL-E, Photoshop AI tools, Google Colab, huggingface
*Please have the most updated version of Photoshop on your computer.

TUE 14 NOV 10:00-16:00Workshop 3: dreamClipping
Concepts: AI Video Editing, Aftereffects
Tools: Runway ML, Aftereffects
Goal: Produce a 5 second video using your AI assisted images from the previous workshops.
*Please have the latest version of Aftereffects installed on your computer.

DEC-JAN – individual scheduled consultations

12  FEB: Final Review