Expert Interview #4: The Designer
Speaking about AI and creativity
Arnaud Imobersteg is a Swiss designer and illustrator. His Shanty Stack project was a winner in the professional category of Architizer's prestigious Vision Awards, ‘the world’s biggest awards program dedicated to the art of architectural representation’. The same project won the One Rendering Challenge, also from Architizer, in 2002. He’s the latest creator I’ve interviewed to get the perspective from creative experts in their field on how they are engaging with AI.
Explainable: I first came across your work with Shanty Stack, could you tell us a little about that and what tools you needed to make it?
Arnaud Imobersteg: The Shanty Stack image is a personal project made in 3D using Blender and relying a lot on the Favelas Kitbash3D assets. It went through a few iterations of quick rendering to plan what to develop further.
The image, once finally rendered, is post-processed using Photoshop to rework the lights, atmosphere, colors, contrast, add picture elements. This is my usual process with a highly polished “big” 3D environment. It takes about a full week, including the sketches, assets and references.
And 3D is a very precious tool when it comes to spatial design and environments because you have full control over basically everything, from the geometry and the “real space” this would imply. But it also works with the light, with materials and everything is mathematically calculated. But it needs some adjustments to make the image nice and to give it more taste and personality.
So the idea here was to question the giant slums that keep being built. Big metropolis, like Dharavi or Kowlooon. This extreme population density shows the need for a more proper urban development strategy and, possibly, a more holistic approach.
Explainable: A theme of this newsletter is that generative AI tools look impressive in all sorts of creative fields until you talk to an expert. With that in mind, what impresses you about generative AI tools?
Arnaud: My work is very related to new technologies or improvements, as in the 3D field. I have to be aware of the new tools, and techniques that are appearing. Not necessarily to adopt them, but I can't be fully closed to them. Or to stick with Blender from 2014, for example
I'll speak mainly about Midjourney, and a little on ChatGPT and [AI in] Photoshop, because those are the ones I know and use. It has been only a year since they arrived and it has already produced some big impacts, if not changes.
At first, like everybody I guess, I was shocked at how powerful and fast the Midjourney tool was. And it has already tremendously improved.
Explainable: And, conversely, where do they fall down? I'm thinking in terms of illustration, design, and architecture so feel free to address all, or just some, of those areas!
Arnaud: Suddenly, there were a lot of people pretending that art was not limited anymore, and exclusive to people with talent. That everybody can now create worlds from a click, without having to learn how to craft an image.
And on the other hand, there are people afraid of being replaced, or worried about their intellectual property and fearful they are going to be robbed or copied by the AI.
This might be the truth for most non-professional people, but it’s a trap and a false debate to me. The AI is basically Pandora's box, now it has just opened. Stopping its development would also probably mean stopping the development of informatics , which will likely not happen.
For sure there have been legal issues about the license on the objects it was trained from, but the all interest of Midjourney is that it creates a new image from a noise pattern.
And does it create art with deep meaning? Well, not at all, is just a probability thing that the words can influence the image, it doesn't even understand the meaning of words.
And, as people, do we create original content every single time? Or do we re-assemble pieces constantly without naming them properly or even without realizing where it gets inspired?
Explainable: Plenty of people in creative industries are concerned about how AI tools will affect their craft, as well as their earnings, is that a concern you share? Or could these tools improve things for creators?
Arnaud: So when it comes to using Midjourney in a professional environment, right now it is a bit tricky in terms of flexibility, rights, and quality, especially regarding a final product.
Compare that to 3D. Say the client wants to move things like so, or the camera, those are things that 3D allows quickly to change. And in the same manner, 3D rendering allows for precise selections of elements which means that you can still edit the result very easily once rendered. Because you have full control.
In the end is just a tool. If you rely on it, to produce all the work instead of you, then it becomes the final product. And, if so, you might just become a button presser hoping for a different result.
Photoshop AI is a very different thing to me. Not especially in terms of rights but mostly in terms of usage. You cannot ask it to design different demon creatures (like you could do in Midjourney) but you can ask for a tree, or a cloud, at a precise spot in your image. So it’s more like a smart content-aware filler and it can speed up some annoying processes (like removing people from a picture)
Explainable: Are there any generative AI tools that you particularly like using?
Arnaud: ChatGPT. I will give you an example. Recently, I wanted to improve my knowledge about the tropical rainforest to build an automated system in 3D that replicates it as closely and efficiently as possible.
ChatGPT allowed me to not only establish a list of plant species and their characteristics but also to evaluate the spacing between each tree and why.
This kind of information was not possible with Google, or not for a single person at least.
Explainable: What advice do you have for anyone starting in design?
I would highly recommend learning the fundamentals of design, art, how to draw, and the basics, in order to understand properly how things work. Then we can have some abilities to direct a tool well.
Expert Interview is a regular feature where I talk to creatives about their use of generative AI tools. Their wins, their frustrations, their tips, and their concerns. Want to talk about your own use of generative AI for a future issue? Get in touch contact@explainable.online
Expert Interview #3: The Creative Coder