2.5D, simultaneous anime and realism

v3.0 = wai_realcn + velvet + azoth

all examples images use perturbed attention guidance and automatic cfg nodes, with these settings:

scale 0.7

adaptive_scale 0.77

unet_block middle

unet_block_id 0

sigma_start and end -1.00

rescale 0.44

rescale_mode full

automatic cfg

hard_mode true

boost true

no embeddings or loras