Ideas

An SVG Sketch of a mountain, generated using only paths.
A wacky text mind melter, based on what I used for the Rep Cap layered text
A canvas piece that generates actual texture
Applying texture in a lot of different geometrically generated layers
Mixing a black & white textured print with a plotter-drawn outline, black and gold ink
Photo manipulation on a pixel level
Many many layers. Order from chaos.
Erosion, river beds
Start with boids/particles. Use them to generate organic terrain features. Then run erosion on the terrain. Then run a survival algorithm on the terrain. Continue stacking simulations to add complexity.
Machine learning image generation.
Flow fields
Frame rate detection & adjustment
Printability
Custom controls
Typescript
Production components
Presets
Saving presets, database
Make an actual website around some of this
Particle systems
Whatever I make, in any given night, gets posted to wherever. So, if I make a sweet generative animation, the video of it gets posted with a link. The video creation is automated. If I make a still, that gets shared. If I generate a gallery of 100 Dalle-2 gens, that gallery gets posted (with a teaser carousel on Instagram). By starting to share the cool shit I'm doing, good things will come. I find my people, my people find me, I have an audience and a crew.
Dalle-2 inverts the creative process. Now the artist is taking ideas from the tool. But the tool is just another viewport into the world. This is a new pair of binoculars. A crazy and powerful set of fucking binoculars.
Previously, the artist took inspiration from life, people, the world. The tool takes its experience from there too. The tool is another viewport into reality, into the world that is human culture as entered into the dataset.
We have a new viewport into the world. A new search engine. We're searching cultural visual space*. *as trained into the model and as allowed by the search engine.
Make a fully-functional, aggressively minimal, raw-ass website where I'm sharing my jam and my viewpoint about whudup. Nice details you don't expect. None of the bullshit you always expect.
Exploring Spaces. This is the common theme. Let's explore generative algorithms including but not limited to generative models like Dalle-2.
Lessons learned from exploring Dalle-2 and from exploring spaces in general.
Create some really great photo to line drawing algorithms to use with the plotter. Use Dalle-2 images as input.
Circle-packing algorithm, but an initial set of circles are laid before the packing begins. Circles are imperfect.
I want to build a platform for generating and exploring digital art in various forms with the potential to charge per-render and upsell on things like prints.
I think one of the issues we have with Dalle-2 is that we don't know what's happening behind the scenes. People talk about their queries being nerfed. They say their prompts are giving shittier output today than they were 3 weeks ago. Is that true? Nobody knows. The folks at OpenAI aren't going to tell us. Maybe I just got lucky last week and my prompt gave me something amazing, but today it is giving me a long stream of garbage. Also, the short-term financial incentive for OpenAI is to get more renders, and they get more renders when the results suck. They would, in the short term, make more money if they made 50% of the resutls come from a shittier version of the model knowing that it would lead to more $$$. Is that happening? I have no idea. I don't want to have that be something I'm wondering about. I want to search a consistent vector space and not concern myself with these things.
We have the idea that image generation models like Dalle-2 may invert the relationship between buyer and seller in the art world. Currently, people with money buy art from artists whose minds they respect. If image generators completely remove the scarcity of actually good art, then anyone can acquire any amount of art. In that world, everyone experiences amazing art all the time. The thing of value is someone seeing what YOU did. People with means will to pay to have an artist experience THEIR creation. Attention is the only scarce thing left. People with money to spend will pay for attention. They pay more for either more eyeballs or better eyeballs. The most premium experience is to pay a brilliant person to give you unusual amounts of one-on-one time. They will see what you have created and give their genuine opinion. If you're an aspiring physisist, how much would you have paid to have dinner and a discussion with Albert Einstein? If you're an aspiring artist, how much would you have paid to have weekend with Picasso?
The artist still makes art. The difference is that they make their money from consulting. And shows.
And just while new fucking art. The medium maybe changes a bit. It becomes more about the person and the story. The bar is raised.
I guess the fundamental question is, what's worth spending our time on?
It's also the next /giphy. It's a free instant ephemeral visual method of communication. It's the next emoji. It's the next gif.
Eventually it is real time and able to be utilized directly from thought. At that point we will have created a new communication paradigm.

What fucking matters?

Themes

Remembering people & moments that are precious
Where are the feels?
How to help people?

Little stuff that matters

Adjusting parameters in real time, saving the parameters when I save an image, and reproducability of a sketch. Including where in the time axis the save happened.