Originally posted on Substack
The detail is everything -Vladimir Nabokov
As much as the last year has been at a slower pace of releasing this newsletter. I do want to get back into the habit of highlighting the wonderful world of generative and creative coding arts on a more regular basis so I hope you have been able to enjoy the work.
Few Thoughts on AI Art
You will find that much of this weeks highlights do look at the Stable Diffusion model that has been released in the past 2 weeks. It’s of course exciting the recognize the beauty that comes from these models and the interesting ways that artists will use these outputs to enhance their practices.
I know that there are rumors and clickbait about the end of the artist’s way.
However as any artist knows, deep inside there is always an almost “prophetic” urge to create. To think that there is such a thing as true creativity in a deep learning model “at least right at the moment;” well I suppose will depend on one’s definition of creativity.
I don’t think we have approached this boundary at the moment, only the future can tell.
Yet if there is a universe where artists do not continue to create, there is still an issue where models will only become stale and as it becomes the “norm” it will only become yet another tool in the arsenal. To create and be creative one yet has to think outside of the realm of the possible.
A mantra in machine learning is “the models are only as good as the inputs.” Therefore artists are still required to make anything from DALLE2, Midjourney, Imagen, or Stable Diffusion.
{{ youtube TVBbFEQyQFI }}
Just as a side note, if you haven’t had much experience in these Deep Learning AI tools, check out the wonderful revcatt talking about Midjourney.
Celebration of Creativity
Deep learning models are more of a celebration of creativity packed tightly together to create some of the most interesting synthesized outputs we have yet seen in the world.
Above you find a collection of outputs I’ve put together over the summer as I was playing with them. Of course there is a bit of a “deep learning” look to them. None the less there are a number of capital uses of this technology. Some of these uses are described further down!
Thanks for reading and sticking around!
Much love,
Chris Ried
Early Generative Art
Siemens 305 - CalComp Plotter Programmed in Fortran and Produced at Siemens, Munich
Anton Zöttl
Anton Zöttl studied electrical engineering at the Technische Universität München, Munich. From 1957 theoretical work in radar technology, especially in information processing in complex radar systems. From 1970 production of computer graphics using a Siemens computer and CalComp plotter. (ZKR)
🖌️ AI Art
Stable Diffusion
Stable Diffusion is a latent text-to-image diffusion model. Thanks to a generous compute donation from Stability AI and support from LAION, we were able to train a Latent Diffusion Model on 512x512 images from a subset of the LAION-5B database.
You can see that the image of the left has been merged with the hand drawn image of the right side to show how this new diffusion model takes an input drawing to generate some of the most interesting drawing based on the input text.
How to Draw Anything
Watching a vibrant Seattle sunset the other day, my imagination started running. The otherworldly hue of the sky evoked something from science fiction. The hazy orange-purple was mesmerizing.
One Week of Stable Diffusion
Stable Diffusion was released 1 week ago. In just one week, a Cambrian explosion of services, adaptations, user interfaces and integrations emerged - we list here our curation of them and the first time they showed up. Also check out “Multimodal AI art updates that you maybe missed with the Stable Diffusion launch”
I have to say that the interest and speed that the community has adopted and used these technologies have been something really neat to see. It’s always wonderful to continue to see innovation happen in Deep Learning as it continues to bring new tools and interests into the world of artistry.
MinImagen - Build Your Own Imagen Text-to-Image Model
DALL-E 2 was released earlier this year, taking the world by storm with its impressive text-to-image capabilities. With just an input description of a scene, DALL-E 2 outputs realistic and semantically plausible images of the scene, like those you can see below generated from the input caption “a bowl of soup that is a portal to another dimension as digital art”
{{ youtube DsUdvPz5VCo }}
Shoutout to Blender Innovations for the Substance Sampler technique for generating the tiling material, I was trying to do it manually in Substance Designer before seeing his video and it sped this process up a lot.
Here’s a quick little way of using Midjourney to generate new 3D textures and using them in Unreal Engine which is a lovely touch to the generative
🚤 Motion
{{ youtube JuKi35j3Dwk }}
Audience, one of their earlier pieces of work from 2008, uses motion tracking software and creative coding to create an almost uncomfortable, anthropomorphic experience. As a gallery visitor steps in front of rows of individually dancing mirrors, they instantly synchronize and lock onto the viewer. With 100 mirrors now looking right back at you, you then become the focal point of your own onlooking.
Its an oldie but good one.
🔖 Articles and Tutorials
Suture pattern formation in ammonites and the unknown rear mantle structure
Ammonite shells have complex patterns of suture lines that vary across species. The lines are formed at the intersection of the outer shell wall and the septa. The wavy septa can form if the rear mantle of the ammonite, which functions as the template, has a complex shape. Previous hypotheses assumed that the rear mantle is like a flexible membrane that can be folded by some physical force.
Grown, Printed, and Biologically Augmented: An Additively Manufactured Microfluidic Wearable, Functionally Templated for Synthetic Microbes
How can we design relationships between the most primitive and the most sophisticated life forms? Can we design wearables embedded with synthetic microorganisms that enhance and augment biological functionality?
This project has been one of my favorites since its been created back in 2016. Really Neri Oxman and her work using generative processes to and natural materials infused together to create both statement pieces and function works are incredibly inspiring. The following is the Paper that details some of the these processes in how the piece was built and how the idea of the living entwined with the new is always another push toward exploring boundaries.
📚Books
Surrealism Beyond Borders
This groundbreaking book challenges conventional narratives of Surrealism, tracing its impact and legacy from the 1920s to the late 1970s in places as diverse as Colombia, Czechoslovakia, Egypt, Japan, Mexico, the Philippines, Romania, Syria, Thailand, and Turkey. In doing so, it presents a more inclusive and accurate understanding of the fundamentally international character and lasting significance of the revolutionary artistic, literary, and philosophical movement. Vibrantly illustrated with more than 300 works of art by both well-known figures—including Dalí, Ernst, Kahlo, Magritte, and Miró—and numerous underrepresented artists, this expansive book pushes beyond the borders of history, geography, and nationality to provocatively redraw the map of the Surrealist movement, investigating how its visual languages, ideals, theories, and practices were framed or reframed in contexts far from its Parisian origins. Contributions from more than 40 distinguished international scholars explore themes such as the channels used to transmit ideas; artists’ responses to the challenges of political oppression, social unrest, and the effects of colonialism; and experiences of displacement and exile in the twentieth century.
A great introduction into better understanding the Surrealist movement and the themes that were explored through the art of some of the greats.
{{ youtube CBLMibOwqdw }}