Generative Art

There's a style of crochet called "wavy" or "wiggle" crochet that I've been meaning to dip my toe into, and it made me think of that.
1647984873900.png


Also, a petri dish.
 
You may think it's a bit much, but when I see stuff like this, I think back to 8088 Corruption and how the person who made that demo built a routine that tried to match each "tile" in the original image with the closest equivalent ASCII character (so a tile with a diagonal line might become a "/" etc) and I wonder how far something like that could be taken. Of course font choice would be crucial, but I wonder just how closely an image could be matched with text. And let's not even get started on overprinting (backing up the print head to print a second character over the first)...

--Patrick
 
You may think it's a bit much, but when I see stuff like this, I think back to 8088 Corruption and how the person who made that demo built a routine that tried to match each "tile" in the original image with the closest equivalent ASCII character (so a tile with a diagonal line might become a "/" etc) and I wonder how far something like that could be taken. Of course font choice would be crucial, but I wonder just how closely an image could be matched with text. And let's not even get started on overprinting (backing up the print head to print a second character over the first)...

--Patrick
That's awesome.

Ultimately, this is basically a diversion from more generative experiments that I was working on. Ultimately, it still lets me practice some of the techniques for manipulating an image and turning it into array data. The next step, which may not directly translate with this image, is vectorizing it and cutting them on the laser cutter. I am interested in using multiple cut layers in a method not too disimilar to overprinting.
 
You may think it's a bit much, but when I see stuff like this, I think back to 8088 Corruption and how the person who made that demo built a routine that tried to match each "tile" in the original image with the closest equivalent ASCII character (so a tile with a diagonal line might become a "/" etc) and I wonder how far something like that could be taken. Of course font choice would be crucial, but I wonder just how closely an image could be matched with text. And let's not even get started on overprinting (backing up the print head to print a second character over the first)...

--Patrick
 
What was the text prompt you used to generate that?
Well it's a mess but

portrait, half body, centered, a demonic woman with horns, red hair, black eyes, wearing a leather bodysuit, portrait of a person by Anne Stokes and Arina Tanemura, environment design by Artemisia Gentileschi, intricate details, action pose, moody lighting, smoke, vivid forest background

Also remastered the results one I saw a good pose
 

GasBandit

Staff member
Well it's a mess but

portrait, half body, centered, a demonic woman with horns, red hair, black eyes, wearing a leather bodysuit, portrait of a person by Anne Stokes and Arina Tanemura, environment design by Artemisia Gentileschi, intricate details, action pose, moody lighting, smoke, vivid forest background

Also remastered the results one I saw a good pose
Hah, I tried feeding that same prompt into nightcafe and got this weaksauce.
How's that "vivid forest background" workin out for ya

1663720961180.png


I guess I should have appended "with a finely detailed symmetrical face"
 
Last edited:
1663763356141.png

That's not creepy at all. :D I'm working on trying to get Cthulhu in a 3 piece suit..this was one of the intermediate results. It's not Cthulhu...but I really like how it turned out.

1663763617926.png


Getting closer. But I like the suit in the failed offshoot so much better.
 
Last edited:
Top