@ reachartwork on tumblr & bskyさんのイラストまとめ

silly me i forgot to add hueshifting as an augmentation

COMING SOON IN LOOKING GLASS 1.6...

POSTPROCESSING

the lains have been BITCRUSHED

MY LAINS HAVE BEEN AUGMENTED

THE LAINS HAVE BEEN ROTATED

i fed an ai a bunch of images of lain

0.7, 0.8
0.9, 1.0

hmm

0.3, 0.4
0.5, 0.6

we are using the same 50 epoch low-similarity data-set of the lain baseline image. ultra-high, ultra-high, temp of 1.0

baseline, typical_p = 0.0, typical_p = 0.1, typical_p = 0.2 (the recommended starting point)

do you notice any real differences in the sets?

temp = 2.0 BUT varying the confidence and variability

ultra-low + ultra-high
ultra-high + ultra-low
ultra-low + ultra-low
medium + medium

SO FASCINATING

temp = 1.5, 2.0, 3.0, 10.0