Video wird geladen...
Video konnte nicht geladen werden
When you make images with code you have more control over the granular details.
24,692 Aufrufe • vor 14 Tagen •via X (Twitter)
22 Kommentare

this is so cool and beautiful

Thanks Guillermo :)

this is giving me Bret Victor - Inventing on Principle demo vibes. very nice!

it would be so crazy to do animations with manim, but with gated APIs (like you did with the p5 paint) animations with code, but in a very unique style CC: @SergioPaniego

@SergioPaniego would be super cool!

Pretty cool

Beautiful!

I like your horizon slider that’s a good idea

love itttt

Thank you :))

This is far more interesting than a single generated pass I could see art in this way be more “editable” and therefore more human imo

pioneer fr

Nice

how do you usually decide which details to prioritize in your code?

this is sick, can i play with the site ?

hopefully sometime next week! In the meantime you can add ur email here to get notified

Do you find the biggest gain comes from deterministic geometry, or from being able to edit individual parameters after generation?

with code, you can tweak specifics like dimensions and spacing instantly

wow

depends what the code talks to. svg and shaders hand you exact coordinates, so the control is real. a wrapper around a diffusion model is still prompt roulette with extra syntax.

This is all javascript, there is no diffusion. Those intermediate shapes you see as you adjust are for showing a fast preview since the full painting takes about ~1-2 seconds to render. It's very much deterministic

wow, it’s quite beautiful



