Il_M1x版本v1.0 (ID: 1261614)

As usual when playing with multiple LoRA, at one point, using too many of them is slowing my potato GPU ^^;

So, for fun and tests, here is a merge of several LoRA i like to use together to give a bit of "pop" to an illustrious model (NB: even if most of the time, i roll with DPM++ 2M, this works best with Euler a).

I tend to use strength between 0.6 and 0.8, but feel free to experiment, a matrix of various strength of the same picture is in the showcase gallery :D

Recipe (made with Super merger, plus_lora, no dimension change after aligning to 32 the only LoRA not of this size):

• New Incase Style (2024) - Illustrious NoobAI: 0.3

• Battle Chasers Style - v1.0: 0.6 (after switching dimension from 8 to 32)

• Ickpot - v1.0: 0.4

• Pony People's Works - v4NoobEv1.0: 0.5

• Ri-mix - Alpha Illustrious: 0.6

Like i said, for fun and tests, nothing serious here :D Which is why, for the sake of the experiment, i also run a generation with this LoRA using a V Pred model, and the result does not looking too strange ^^;

描述:

训练词语: