Lore Machine Meets WeavingWithAI
Jerome Herr's Expedition Into the Latent Space
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 birthplace 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 firstname.lastname@example.org.
WeavingWithAI, aka xeronimo, aka Jerome Herr is an artist and coder exploring the the newfound frontier of AI animation. You can explore his world on Twitter, Objkt, Foundation and TEIA.
Lore Machine: How do you use the term 'weave'?
WeavingWithAI: I just liked the anachronistic image it produced in my head. At first it only referred to 'weaving with code' back when I created art with code. Nowadays I'm weaving with words and artificial intelligence. Another reason is that I was a big fan of 'Loom' at the time. Younger readers may have to google that name.
I've been seeing the term 'latent space' popping up more…
Latent space is the space of the ‘possible’. It is an ocean, containing all possible images within. AI artists swim through this ocean, using prompts as their guide. These prompts lead them to different areas within the space, revealing images that no human has ever seen before. By experimenting with prompts, artists can keep discovering new areas, hopefully resulting in images that meet their initial vision. Of course, that vision can change or be adapted over the course of this expedition.
You've been in the creative tech space for a minute, as a coder and more recently an artist. What brought you to generative art?
My first reaction was: why the differentiation between coder and artist!? Creative coders are artists too! They’re just writing the actual code that produces the images or animations, while the AI artists are writing prompts that help the AI create images.
I always had this creative urge but it took time to find a medium that fit. I drew a lot as a child. Later on, I dove into photography. I even tried sculpture. It was only when I stumbled on Fun Programming in 2014 or 2015 that everything clicked. I went down the rabbit hole and created a lot over the years! My enthusiasm got me hired by Monash University in Australia as a moderator on their forum for an online course about creative coding. The culmination of all of this was minting some of my pieces as NFTs, first on HEN/TEIA and then on fxhash. At the time, I noticed more and more AI art in my Twitter feed from artists like Ganbrood. Once tools like Midjourney and Dall-E became available I switched camps and went full throttle on AI art.
Can you outline the difference between Generative Art and AI Art for the folks at home?
Generative art, at least for me and many others, is art generated by code that you have written yourself. AI art is art created with text prompts guiding image generators like Stable Diffusion, Midjourney or Dall-E.
The popular conception of AI art has evolved over the course of time. We're just now getting glimpses of the animation that can be done. Can you unpack your animation process?
I create my animations using the Deforum notebook on a Google Colab. It's way less intuitive than using apps like Midjourney, PlaygroundAI or Dall-E. There are more variables at play, which makes it more challenging but also potentially more impressive and satisfying.
Technically speaking, I am using five slightly different prompts and the AI is creating images based on those. I can also define whether the 'camera' should do any translations, rotations or zooms. This is all done by manually typing numbers. It’s a cumbersome process. That's why I keep it simple for now: only an x and/or y translation, and a very light zoom.
Even with these minimalist settings the results can be impressive. One of my first videos - ‘Memento Mori’ - has been viewed 777k times.
What’s the story behind ‘Dark Night of the Soul’?
‘Dark Night of the Soul’ is a combination of enigmatic concepts (which challenge the AI quite a bit), artists (Norman Rockwell and Edward Hoppers amongst others) and a minimalist black and white 'ink' look (which is unlike most other AI animations right now).
I play around with the phrasing of the prompt and seeds until I get the visuals and the 'story' I’m looking for. HOW the animation ultimately looks, based on the prompt and seed, is sort of out of my hands. I’m just indicating the general direction and curating the output.
Do you see AI replacing aspects of the film production process?
I’m not really from the industry, so I don't have much to say about this. But I'm sure that generative AI will enter the production process in a massive way. It will be a tool for inspiration and to create text, images and even video. The progress in AI-related technology is so fast these days that one can't exclude anything…
What do you see as the role of collaboration in the AI art and animation space?
I'm not a huge fan of collaborations since I prefer doing things at my own rhythm and in my own way. I did a collaboration with David Mrugala in my coding days called ‘Decadence’. Decadence started as a simple grid of random values that became a series of generative images with a shared aesthetic that values excessive indulgence in visual pleasure. It was a success!
I'm always open to suggestions though and there might even be an AI collaboration coming soon-ish. Can't say more at this point. I do belong to a collective called MAIF. Some of the members have collaborated on pieces but it's not the norm. It's more about having a network/support group of artist friends.
Any folks in the space that are blowing your mind?
This one’s hard… Ganbrood (obviously), Mario Klingemann, Phoebus, Artificial Bob, and Claire Silver.
What's next for you?
I will keep exploring the latent space!