Meet AI Artist Ben Scott-Pye
Meet AI Artist Ben Scott-Pye
Turns Out Ganimation and Ceramics Are Pretty Similar
Lore Machine explores the intersection of artificial intelligence and storytelling. We believe (read: hope to god) that AI isn’t the end of creativity, but the birth of an expansive new design surface. If you’re out there making cool stuff, we’d love to hear from you. Hit us up at info@lore-machine.xyz.
I found this AI animation by Ben Scott-Pye a few weeks ago called ‘Arcade’. It totally blew my mind. So I asked Ben to break his process down.
Lore Machine: What’s your background?
Ben Scott-Pye: I have a Contemporary Crafts degree, but I love everything. Choosing which path to go down was very difficult.
I spent most of my degree trying to learn processes that were different from what everyone else was doing. Stuff like folded paper, cast paper pulp, casting in plaster, 3d modeling and laser cutting things out of acrylic.
I'm pretty sure I have ADHD. I just get excited about new processes and media, buy all the stuff to learn it, and then abandon it in a few months for something new. The AI art thing seems to have stuck though, probably because the process is so fast. You can quickly get things out of your head and visualized in front of you.
My end goal is always to find some interesting way of combining new contemporary media with more traditional handmade processes.
There’s been so much progress in such a short space of time with this AI stuff that it's hard to stay on top of what's possible while working a menial day job... but that's the goal!
You refer to yourself as a craftshuman…
Calling myself a craftshuman might be a stretch because I've never stuck anything out long enough to master it before I get distracted by a new shiny thing. Ultimately I want to craft beautiful things that take a lot of time and effort that I can be proud of. The word "artist" is probably more accurate.
There’s a lot of folks who look at AI art as sort of cheating, like it doesn’t require as much artistic dexterity as other forms.
A counterpoint to criticisms of AI art is that a lot of processes in other materials aren’t fully under the control of the artist either.
There are plenty of ceramicists who make generic looking pots and vases but experiment with glazes that do wild unpredictable things. The ceramicist will spend a long time experimenting with different ratios of chemicals or combinations of glazes until they get the unique result they’re looking for. Then it's just about applying that formula to all of the pots and getting these crazy wild intricate-looking outcomes. None of these outcomes were really skillfully crafted, but they look beautiful. To anybody who didn't know how it was done it looks like magic.
Look up crystalline glazes, crackle glazes, foam glazes or crater glazes. Copper electroforming also has a similar kind of unpredictability to it, even more so when you get into applying patinas.
I'm bringing these up because I’ve experimented with these mediums and processes, particularly electroforming and patinas. Exploring those processes doesn’t feel that different than my current experiments with AI animation.
That said, I do see a ton of very unimaginative AI art. There seems to be a handful of popular artist names that tons of people plop into their prompts. They all generate a very similar look. These folks mostly seem to be trying to make a quick buck with NFTs. This really helps the case of AI image generation critics.
I was initially drawn to your piece ‘Arcade’. People are pretty accustomed to seeing still AI art. Can you describe how you put that piece together?
I had an idea for a look I wanted. I dug through tons of stable diffusion results where different artists’ names had been used. Eventually, I found one in the ballpark of what I was looking for. Then it was just about refining. As an example, say I was using an artist whose name creates a painterly 70s retro vibe but I feel the shapes and forms are overly-detailed. I then introduce other artists’ names that are more simplified or cartoonish or stylized. Often there's a process of rearranging the order of the prompt. Other words and phrases referencing processes, mediums and styles are useful too. Sometimes I spend hours trying to reach the look that I have in my head.
I usually use the Deforum Diffusion colab notebook to animate. It's fantastic and the people working on it have so many exciting things planned. My animations are usually made of multiple parts. I take an image from the end and use it as the init for the next frame. That gives me fine control.
The ‘Arcade’ animation was produced in three sections. The journey to the arcade machine, the zoom into the screen and then the 2d isometric animation. I tried a lot of different stuff before deciding to make the animation inside the screen a 2d isometric animation. I threw out so many versions before I was satisfied.
‘Arcade’ was for #SeptembAIr which is kind of like Inktober but for AI-generated art. I decided to use it as a chance to push myself to explore a lot of different styles and approaches. I often try to think of novel ways to use the tools. I had been thinking about this one in particular all day at work while doing boring mindless manual labour. It's a nice way to get through the day, developing projects in my head that are ready for when I get home.
How much digital massaging and time goes into your AI animations, like Peace or Quasar?
There isn't much in the way of post-production other than trimming clips, reversing them and putting royalty free sounds in the right place. I try to find music that fits the vibe and pacing of the animation I’ve created. I really should teach myself about color grading and things like that, but I only have so much time. When SeptembAIr is over I’ll be spending more time on single projects.
Any advice for folks just starting their journey?
I have put a ton of time into playing with these tools. I guess it depends what people want to get out of it. Creatively, I’d suggest avoiding doing what other people are doing. Don't use the same artist name combinations you see everywhere. Experiment a lot. Put things together in ways that haven’t been done yet. Play with the models a lot till you start to get a feel for how they work. Prompting in Dall-E works very differently to Midjourney or Stable Diffusion or Disco Diffusion. Don't be afraid to try unusual things out and fail.
$1 million to make whatever you want…
That would be the dream, wouldn't it? It's hard to say! I love so many different mediums and processes. I can picture a huge space with paintings and drawings all over the place. There’d be big vats of solution for electroforming copper over large 3d-printed pieces based on AI-generated designs and a big kiln for firing enamels on to them. There’d be large 3d printers in there on the scale of the Elegoo Jupiter. Probably a full wood/metal workshop with a CNC Router and laser cutter. I want to make large sculptures combining all kinds of processes and definitely involving AI image generation.
Are there any folks in the AI art scene you’re really in to?
@Infinite__Vibes on Twitter, definitely. They don't get enough love and I admire their willingness to just get stuck in and experiment in weird and wonderful ways.
@Squaremusher, also on Twitter does a lot of crazy interesting stuff that I really vibe with.
Can I say Holly Herndon? She's so great, especially her latest album.
Also I would mention @rachaelahz and @weaving.light who have really helped me out a lot and their stuff is neat.
@GanWeaving definitely has an eye for good art which shows in their work.
You're looking into a crystal ball that tells the future of generative art, what do you see?
We already have story/text adventure generators that do a decent job and are going to get better. Meta just showed off some of the results from their video generator which looks a little bit janky but is mind-blowing at the same time. I could see AIs generating full bespoke VR video games - experiences where you’re completely immersed in the world and can interact with all of the characters. If nothing wipes out civilization in the next 100 years it's going to happen.