EAVA 2025

Join us for our second screening and artist talk this Fall with @douglasrgoodwin.

Doug makes short films, photos, and interactive works based on his longstanding interest in machine learning, systems thinking, computational photography, artificial intelligence, and interactive media. He has a background in experimental theater and writing. He is a teacher and is an alum of the Experimental Writing program at CalArts.

Come hear from this talented artist and CalArts alumn as part of the Experimental Animation Visiting Artists series.

summary

This talk traces a practice rooted in cyanotype photography, generative coding, and speculative media archaeology. He describes rejecting opaque AI systems like Stable Diffusion in favor of “authentic synthesis”—a method of combining chemistry, code, and history in processes he can fully understand and shape.

Goodwin recounts the loss of his house and early cyanotype archive in the Eaton Canyon fire, which forced him to rebuild his process from scratch, developing custom software to convert video into negatives and overcome the limitations of scanners that cannot register cyanotype blue. His cyanotype film project connects to the Laschamps Excursion 42,000 years ago, when Earth’s magnetic field collapsed, people sheltered in caves, and left ochre handprints on the walls. Imagining a future recurrence, he designs hand-cranked Mutoscopes that hold 850 cyanotype frames each—cinema without electricity, meant to endure beyond the grid.

Alongside these analog processes, Goodwin demonstrates generative work in Processing, slime mold simulations, and constraint-based systems inspired by Georges Perec. He frames both cyanotype and code as ways of confronting instability, chance, and human error, rather than chasing AI’s polished perfection.

The talk moves between personal history, technical detail, and speculative storytelling, always returning to the principle of authentic synthesis: a refusal to outsource imagination to opaque models, and a commitment to carrying the process oneself—being, in his words, “your own mule.”

edited transcript

I appreciate your noticing the human part of my practice. Today I’ve decided to show only work that I’ve controlled closely. There’s just one piece here made in Stable Diffusion. If you’ve used AI tools like MidJourney or RunwayML, those are all built on Stable Diffusion.

I’ve been thinking a lot about what these models bring into my practice. They’re created by corporations and trained on our photographs, our families’ photographs, even fragments of medical records. All kinds of strange material shows up in them. That raises the question: how do we use them responsibly, and how do we maintain control of our own artistic output?

Most of the work I’ll show comes from a large cyanotype film project I started about a year and a half ago. I’ve produced around 1,400 cyanotype prints. Cyanotype is an iron-based process: you mix your own chemistry, coat paper in dim light, let it dry, and then expose it under the sun. I often check NOAA’s UV index to calibrate exposure times.

The prints you’re seeing today are not from that first stack. My house burned in the Eaton Canyon fire, and with it most of my early work. Starting over meant new printer, new frames, new chemistry—but also a chance to refine my technique. I wrote software that takes video, breaks it into frames, and arranges them nine-up on 8×10 sheets for printing. I created fiducial markers to align the scans myself, because commercial scanning software doesn’t recognize cyanotype blue. Historically, reprographic systems were designed to ignore blue lines on paper—so I built my own tools.

My approach grows from a mix of art and software. I was lead developer at LA Metro for ten years, building the real-time bus and train tracking system that still runs today. That experience taught me: if you don’t have a mule, you become your own mule. You figure out how to make the tools you need. That’s how I see art-making too.

Growing up, I didn’t have drawing paper. My parents were experimental psychologists, and I drew on discarded IBM punch cards from the lab. Failed programs became my sketchbooks. That blend of computing and creativity has stayed with me.

Cyanotype also connects to deep time. About 42,000 years ago, during the Laschamps Excursion, Earth’s magnetic field collapsed. Cosmic rays bombarded the planet, climates shifted, and people sheltered in caves. On the walls, they left ochre handprints—gestures of endurance, time capsules for the future.

The poles are moving again. Since 2009, magnetic north has been shifting 30–60 km per year. Some scientists believe we may be heading toward another excursion. If that happens, compasses and GPS will fail, grids may collapse, and our electronics will falter. Cyanotype, with its iron chemistry, becomes both archive and antidote: the same Prussian blue compound is stockpiled as a medicine for radiation poisoning.

I’m designing the cyanotype sequences for 19th-century devices called Mutoscopes—hand-cranked mechanical projectors that function like flipbooks. Each can hold 850 frames, about a minute of motion. They work without electricity.

I plan four Mutoscopes: glaciers collapsing into the sea (releasing their 800,000-year-old archives of dust and cosmic radiation), people moving through modern “caves” like parking structures, studies of wind and clouds, and speculative freeway scenes imagining life after a magnetic collapse.

I’ve also been experimenting with magnesium flash. Early photographers used these explosive lights. I built cages lined with cyanotype paper; when the flash goes off, UV light blasts through pinholes and projects an image. The resulting prints look like holographic fossils of the explosion itself.

Parallel to cyanotype, I work with generative systems in Processing, a Java-based platform for artists. I adapt models from scientific research—for example, slime mold simulations—and modify them to create forms that grow, connect, or collapse. When pushed past stability, they explode into unexpected structures. I value these instabilities. They differ from AI’s polished outputs: here, the rules are visible, adjustable, and the “errors” become part of the work.

This connects to Georges Perec, who once wrote a novel without the letter “e.” Constraint breeds invention. For me, constraint—whether chemical, algorithmic, or procedural—is a way to invite surprise.

I think of this approach as authentic synthesis. It’s not about mastery for its own sake. It’s about combining code, chemistry, history, and speculation in ways I can trace and understand.

What unsettles me about generative AI isn’t the images themselves—it’s the opacity. Too often artists accept the output without knowing how it was made or what data went into it. I want to participate in the process, to shape it directly. That’s the difference between being fooled by an image and being responsible for it.

If you remember one thing from this talk, let it be this: be your own mule. If the process won’t carry you, carry it yourself. Build the tools you need. Mix the chemistry, write the code, strike the match. Out of that labor comes not just images, but endurance.