Aiming to be a more generalized base model, staying closer to the original v1.5 style, except with higher native resolution support (1024x1024), better coherency, better anatomy, better faces, easier prompting, better embedding and LoRA support.
Can be a good base for further training, too.
I only used MBW merges with Base Alpha at 0, so the Text Encoder should be intact afaik.
~Recipe:
v1.5 + mbwFlat50*SuccubusV3 + mbwFlat50*ChimeraV4 + mbwFlat50*HydraV1 + mbwFlat50*WyvernV9
+ mbwFlat25*PirsusEpicRealism1_3 + mbwFlat25*AmIRealV3
Download link:
(~5.2GB)
Attachments
-
grid-11380.png -
vie-by-christopher-nolan-a-detective-standing-in-t.jpg -
z-grid-0725-2597049991-woman-full-shot-dramatic-8k.jpg -
chen-tight-latex-minidress-leaning-forward-cleavag.jpg -
io-anime-character-version-of-name-highly-detailed.jpg -
house-on-a-hillside-abstract-landscape-highly-deta.jpg -
d-anthropomorphic-animal-by-artist-and-artist-high.jpg -
hoto-of-a-funko-pop-puppy-figurine-highly-detailed.jpg -
6-sims-4-character-version-of-name-highly-detailed.jpg -
0-simpsons-cartoon-drawing-of-name-highly-detailed.jpg -
grid-11388.png