AI assisted industrial design
Accessing over 5 billion images and understanding natural language, the open-source text-to-image generator Stable Diffusion can combine those two into one coherent image. Due to its ability to produce absurd combinations, this tool is currently mostly used for weird images, portraits, and surreal art. Given that the underlying deep-learning technology is comparable to the human brain, the generated images can actually provide original new creations than just combinations of already existing images.

The images are created through trial and error of complicated word combinations, changing settings, text prompt and initial images, by which the algorithm is guided, and selecting the image that is similar to the vision in mind.

With a background in industrial design, I wanted to use this tool as inspiration and immediately began experimenting to create new products. These are my early creations, each of which has a brief descriptionThis project is an ongoing exploration.

It seems that right now, AI is like shooting arrows at a wall and then drawing the target around some of them. It's clear that AI needs a lot of human interventions and evaluation to be useful. My other text-based "Product Generator"-project demonstrates this more clearly. For me, there is a place in the brainstorming phase, but only under very specific conditions.

I am actively exploring this project and working on implementing AI into my ideation process. Currently, I use ChatGPT as ideation and responsive notepad for my personal product developments.
2024
Integration
before
before
--->
--->
after
after
The concept was to render a basic but interesting three-dimensional form. After that, turn it into a product by using Photoshop's new generative fill features. Trying to integrate this workflow into the process of developing a product.
Although I don't find the outcome to be inspiring for my own design work, I do think that the product's environment-based staging can be useful.
before
before
after
after
Fixing common render issues with generative inpainting tools (like Photoshop's generative fill) to get to an acceptable image.
2023
Ideation
Combining conveyor belt sushi (kaiten-zushi) and home Raclette: "Running Raclette"
In my experience, generative tools are not capable of novel ideas, only serve some inspiration.
Materials
Office chairs using recycled plastics - new software, new results.
Organic shaped elm wood chair.
2022
Materials
Office chairs using recycled plastics.
​​​​​​​
Anthropomorphism
A calm and a loud chair.

Combination
Elm chairs utilizing sustainable materials.

Public seating made from recycled cans and parachutes.
"Peaceful public seating" made from recycled military equipment

Ideation
Lamps that can be moved to change the mood.

A raclette heating appliance with an additional level which can be adjusted.

Repurposed phone booths, message boards, information, memorials.

A speaker made for listening to podcasts.

Transfer
A japandi style incense holder and burner.
I described the existing self-made woodwork incense burner as accurately as I could. 
2023 / 2022
2021
Variants
Trying out different industrial design styles on a cube with early publicly available image generation software.