//=time() ?>
whenever doing anything that imitates traditional cel animation you generally have to make the FOV so tight the view is practically orthographic. the for the "matte painting walls" to look good on camera they have to look like horrible flesh slabs at regular fov 😂
training a DALL-E model to generate 90s anime aesthetic cities. trains up fast compared with a VQGAN transformer, this output is at mere 50 epochs. the big downside is you have to actually sort and classify your training data. sheesh, im making AIs so i DONT have to work hard
@Shoehead_art @dietinghippo "90s anime aesthetic screenshot of a person with purple hair"
@dietinghippo "vaporwave captain falcon by Shinichiro Wantanable|photo:-1"
@AndiJorgensen vaporwave in vivid color by studio ghibli|photo:-1