In Germany, there is the term "Gebrauchsgrafik." It refers to visualizations of products or services. These works typically lack artistic value and are often only visually aligned with the zeitgeist. Think of your bog-standard low effort advertising or IKEA instructions, generative AI is more than suitable for that. Not everything is or has to be considered art.
Low effort advertising sure. Diagrams I've had no luck with personally. With current models it's been a very frustrating process to even try. I can generate stuff with maybe one or two key ideas, but instructions are ignored as you try to incorporate more details. There also aren't really any affordances yet for dimensions. Perhaps styling / editing from sketches fixes this, I would be interested to hear from anyone who has managed to get a useful, Ikea style diagram out of gen ai.
To me it feels we are a long way from AI being able to produce useful minimalist (i.e, omitting un-necessary detail) diagrams at better than a surface level of stylistic mimicry.
Producing useful functional diagrams entails some understanding of the physical components and assembly process, along with some model of how people think and what information they need at what step to minimise confusion.
Naturally a lot of this information is embedded in the design of the assembly itself and perhaps the big picture can all be "prompted in" but it's not a domain where "bullshitting it" works well.
We work with flux, custom loras and controlnet and it's working exceptionally well. Using only prompts as conditioning is also not the way to go for higher complex tasks. It's day and night compared to dall-e,imagen, sd1.5 or sdxl.
