“You don’t learn to walk by following rules. You learn by doing, and by falling over.” — Richard Branson

The first of many.

I’m sure of it. Absolutely and utterly sure of it. As sure as the universal constant that all humans poop. The common bond that unites us and Our shit.

Time blindness.

It’s a thing.

And along with all that…

Hyper focus.

It’s a thing too.

And wow. I’ve been captivated, enraptured, and just all around freaking fucking fascinated.

So fascinated the last week and a half of my time went by at the speed of light.

Stable diffusion.

Wow. This machine learning model is fucking just. Wow. Amazing. This technology has me feeling the most excited, and motivated to fuck around and find out than anything in the past many years.

Those amazing big brain folks that fed a neural network a dataset of millions and millions and millions of images of all types… and mushed it together with an enormous dataset that associates visual details with words.

So you can feed it prompts such as:

“Black hole with a multicolor tesla coil shock with Ryan gosling holding it"

And it will produce an image such as:

Or you can feed an existing image, write a prompt of what you want it to be, and it will change it.

For example, it turned this:

Into these:

Those big brained folks that made it possible for me to do that.

Wow. Just fucking wow. They’re just as amazing as their model.

And for the past 13 days I’ve messed around with messing around with many multiples of words. Words that are tokens and tokens into prompts. Prompts that get processed by neural network model which has condensed the visual essence of much of humanity into billions of bits into algorithms that can

Some inspired by some from ArtHub.ai, some inspired by the details of how an artist produces art then learning the words to describe what they did, and some inspired by the myriad and more experiences my sensors have sensed and my brain has processed.

Theoretically.

I think some of theses pieces of so special, so amazing, so unique, and so much more than that, that with the exponential explosion of content being produced by humans, and now with machine learning models being able to produce art from the essence of millions of visual depictions of human passion and emotion, I firmly believe that curation will become a very lucrative career.

When I talk to new people about this kind of new technology I usually say something like, “This will generate images that are universally unique and one of a kind.

My guess is that someone is going to create a machine learning model that will be able to reverse engineer artificially created images back into the tokens that formed them.

In 13 days I’ve managed to produce:

~/code/sdiffusion-v1
❯ ls images | wc -l
6096

~/code/sdiffusion-v1
❯ du -hs images
4.5G    images

That is six thousand ninety-six images. That is 4.5 gigabytes of pictures. That is a shitload of pictures, of art, of never before seen creations, to go through.

And so much of it is just amazing.

Pieces of art that are universally unique generated through distillation of our collective creative consciousness.

Well… that’s not completely accurate. Mostly. But noooot completely. When you RTFM, you’ll eventually read something that sounds a bit like: “Seed: A number between 0 and 4,294,967,295 that is used as starting point for the image generation. If the same seed is used with the same prompt and the same settings (except steps, which may vary), the same image will be generated. If left empty, a random seed will be used. "
This means that someone could recreate the piece, but there will be an obscenely, itsy bitty, teensy weensy, infinitesimal of a chance that something biological could reproduce it. One of my prompts I wrote was 210 letters. So… there is a 1 in 901,943,131,950 chance that someone will be able to recreate it. This math is wildly over simplified and inaccurate though, but still - the chances are small.
And, according to the CDC, “The odds of being struck by lightning in a given year are less than one in a million, and almost 90% of all lightning strike victims survive.” And according to random internet sites I just skimmed through,there’s about a 1 in 292,200,000 chance that you’ll win the lottery.

Then I realized that even though people can purchase ingredients and make their meals at home, they go to restaurants.

That even though people can plant seeds and grow their own produce, they go to grocery stores.

That even though people can produce art and information in unfathomable ways and methods, they go to the people that produce or reproduce it.

And that’s when it dawned on me that I can sell this stuff… and more. And as I thought more and more about how this might work as a business model, an avalanche of ideas buried me so deep beneath them that I’m still working my way through them so I can get some air.

And that’s when I decided I’m going to buy a nice printer, generate some, curate them, crop ’em, and create some prints that I have a a feeling of a hunch hunch that I’ll be able to sell at Saturday Market.

And as I share my thoughts with people, and some of these test prints, I feel more and more as if this kind of business model just might work,  and can be grown through providing novel and unique experiences and content.

So I’m fucking around and finding out.

In the last 13 days, I’ve:

  • Generated gobs and gobs of pretty good art.
  • Decided I finally have a use for a domain I bought in 2020, microdosing.to.
  • Setup a website for microdosing using Hugo, and Cloudflare that uses lazy loading to provide a stream to scroll through of those thousands of images.
  • Registered “microdosing llc” with the Oregon Secretary of State.
  • Obtained an EIN, and business checking and savings accounts.
  • Setup Square, Stripe, and BlueOwl to play around with how I might sell this stuff.
  • Setup BlueOwl and bought one of my own digital pieces. For $3.
  • Purchased a Canon PIXMA PRO-200.
  • Purchased a bunch of paper which cost more than I ever thought I’d ever spend on paper ever.
  • Realized that Stable Diffusion produces images that are a max of 768 x 768 pixels, which meant in order to make prints look nice I’d need to figure out a way to upscale them while retaining their fantastic qualities, so I…
  • Purchased Topaz Labs Gigapixel AI… and Topaz Labs was having a sale, so I was able to get 4 pieces of software, that normally sells for $460, for only $159!
  • Scaled, cropped, and printed 31 pictures.
  • Bought binders, sheet protectors, and put the images in them.
  • Shared the images with random people at random places, new and not, and received universally positive feedback.
  • And more.
Coincidentally enough, I initially bought microdosing.to with the intent of finding pixel art artists and selling their stuff. Guess I’m doing something similar sans artists… but I do have thoughts about that that’ll be written and shared sometime in the future.

So, yeah.

This has caught my attention.

So I’m going to mess around with my prompt wording to make basically universally unique content, then sell single pieces to single persons.

Single as in a singular individual.

And more than just prints.

But.

I’ve already written more than I thought I would. And I didn’t even really wanna write this because I really wanna keep fucking around to find out, but I told myself over and over that my goal was 3 posts a month.

And December is tomorrow.