//=time() ?>
呪文「A thick muscle bear wrestler. portrait. by Caspar David Friedrich, matte painting trending on artstation HQ」Cfg Scale: 20, Steps: 70, Sampler: plms
もうちょっと上まで描画してほしかった。
足が半分熊になってるの面白い。股間のごまかし方めっちゃ草w
When generating non-humans with #stablediffusion, turning up cfg_scale can be helpful.
This is "a minotaur wearing plate armor and holding a mace" with c=7(default}, 20, 35, and 50
@theloopcurator And it looks like this is the case, if I use the same random seeds and the same prompt but cfg_scale 5, the cool "shattering" effect on those two goes away and is just replaced with a circle.
#aiart #digitalart #machinelearning technique: #clip #diffusion
Seed evaluation with an empty prompt.
prompt:
seeds: random
resolution: 384x256
iteration: 25
model: cc12m_1_cfg
prompt weight: 11
"the vision of the ophanim by Rene Magritte, oil on canvas" cc12m_1_cfg
"angelheaded hipsters burning for the ancient heavenly connection to the starry dynamo in the machinery of night, cottagecore, tempera" cc12m_1_cfg
"a witch's garden in the enchanted mists, cottagecore, tempera" cc12m_1_cfg
"a witch's garden in winter, cottagecore, tempera" cc12m_1_cfg
"a cyberpunk cathedral by Frantisek Kupka" cc12m_1_cfg 512x512
Some cloudy mountains generated by the v-diffusion cfg model with 1024 output.
#DigitalArtist #generative #AIArtwork #illustration #cyberpunk #digitalart #ACG #nftart #NFTartist
prompt: an organic and detailed design for a cryonics preservation chamber
(CLIP Conditioned Diffusion [cc12m_1_cfg])
Cosmic Ogre, digital art, cc12m_1_cfg diffusion @RiversHaveWings
a landscape resembling The High Priestess tarot card, digital art (cc12m_1_cfg)