golan.bsky.social
Artist, researcher, educator, instigator. Chaotic Good.
88 posts
3,997 followers
144 following
Regular Contributor
Active Commenter
comment in response to
post
Thank you; this is classy 🙂
comment in response to
post
marvelously obscure pixel-play, thank you for the tasty
comment in response to
post
checks out ;)
congrats on your book!!!
comment in response to
post
🎩
Thank you. I think porting/preserving someone's code is one of the highest forms of praise. JT's "Textension" piece was a bit of brilliance that needs to be rescued in full.
comment in response to
post
One of the ice cream shops in my neighborhood sells Lactaid at the counter for 50¢. I asked them how many people buy them and they said roughly 1 per hour. Alas, I've tried numerous times but can't seem to convince the other place, whose ice cream I prefer, to do the same.
comment in response to
post
CHAR! I have /always wanted/ an impossible hyperbolic paraboloid! Thank you for the birthday wishes! 🫶
comment in response to
post
Perhaps LARPA? Speak with @ntschk.bsky.social and/or @lav.io, who direct larpa.solutions
comment in response to
post
Thanks Jacob!
This summer I'll be creating a new branch of p5.plotSvg for Version 2.0 of p5.js. Apparently the beginShape()/endShape() API is undergoing significant change for the next version.
comment in response to
post
Thanks, friends, who all kindly replied with your offerings and pointers to interesting work.
In 2025, the process of researching contemporary art practices that use AI, is virtually indistinguishable from doomscrolling. You helped lift that cloud. 🌥️
comment in response to
post
That said (as others have correctly pointed out), training a LoRA for Stable Diffusion is not the same thing as training a GAN. This week I'm testing out R3GAN (github.com/brownvc/R3GAN), which is very current; I should have a notebook to share soon. But: it looks like it takes ~3 days to train.
comment in response to
post
Training a Stable Diffusion LoRA with CivitAi is a piece-of-cake drag-n-drop experience that works well in a workshop scenario with students, takes 15 minutes, uses 500 images max, and costs 50¢. You can then use the LoRA in ComfyUI with e.g. RunComfy and you're all set to make "more like this".
comment in response to
post
Congrats, Geoffrey, the article is perfect. I'm really glad you got Theo and Kyle's voices in there.
comment in response to
post
Yes, that's part of the lessons for this unit. If everything works out, they will be using their own data to train both a LoRA (for easy fun with ComfyUI) as well as a GAN (using R3GAN, which appears to be playing well with Colab).
comment in response to
post
Thanks much, Mario. Yes, taking the actual photos/scans is what I was getting at, even though of course I have lots of respect for artist-curated datasets of images-by-others. The context is an assignment for students, in which I'm trying to encourage as much ownership of the pipeline as possible.
comment in response to
post
Thanks Jack!
comment in response to
post
Thank you Hannah!! 🙏
comment in response to
post
Also see portraits made by Vik Muniz -- constructed with e.g. diamonds, sugar, trash, chocolate, etc.
comment in response to
post
Understood.
My respects.
comment in response to
post
comment in response to
post
FWIW, it's possibly worth looking at @michaelfogleman.com's implementation in github.com/fogleman/ln. There's also a Processing/Java port here: github.com/CreativeInqu....
comment in response to
post
Bravo Julien!!
comment in response to
post
ooh tx
comment in response to
post
(Whispering) Is it alchemically important to use that actual number of dead bees? I reckon that with fewer than a dozen dead bees, appropriately rotated, you could make a library of images that could be arranged by software without recognizable repetition. Cf. Chris Jordan's work, etc.
comment in response to
post
Derp! I was working quickly :)
comment in response to
post
zomg! I didn't see a class definition, nice work. I'm embarrassed 😊!
comment in response to
post
...But the code in the Node class doesn't count? :)
comment in response to
post
It's in - thanks so much for sharing.
github.com/golanlevin/p...