3rd Version of my AromaSensei Style LoRa.
Since the ExpMix_Line Models got claimed I had to relearn my LoRa to work with other Models. This time I trained the LoRa on Moonbeam and I think it works quite well. The Artstyle is less "cartoony" and a little more "realistic", but I still think it looks good.
Recommended Settings:
Model: Moonbeam
LoRa strength: 0.8-0.9
Cfg: 6.0-7.5
(I mostly use Lora strength 0.8/cfg 6.0 and Lora strength 0.9/cfg 6.5)
I recommend generating a batch of 4 using Euler a with 20 steps (using 0 credits), then picking your favorite and using the image enhancer. Often you'll get blurry eyes when generation, but the image enhancer usually fixes that.
