This is a follow up to my WAM/Sploshing Lora. This Lora is more specifically targeted to gunge/slime only.
The Lora has been training on approx 400 gunge and Slime images that I've curated from my collection with the subjects wearing a variety of outfits, costumes in various poses and locations to give a varied data set.
The Lora works best with natural language prompts. There are are no trigger words needed for the lora. Simply specify the subject location, pose and substance is enough. Try to describe how the substance specifically interacts with the subject for the best results.
As far as settings go, The lora works well for strengths 0.5 to 1 but I prefer to weight the lora between 0.5 and 0.7.
I've been testing with a guidance of 1.9 to 3.5. Lower guidance will give more realistic images.
The usual samplers should work well Euler Simple/Beta and Deis/ddim_uniform are my go to samplers. I usually run 30 steps, but 20-40 is fine too.
It is trained with the full precision t5/clip conditioning. I have tested FP16 and FP8; both give excellent results obviously depending on the seed used.
Feedback welcome.
Description
FAQ
Comments (6)
can this lora do mud, blood, and water?
The model is not directly trained in these but I have had some success with mud etc. Flux seems to learn the general concept of Messy and how it translates to other substances. best bet is to play around. Flux is continuing to suprise me.
Outstanding! This is absolutely perfect and EXACTLY what I've been looking and waiting for! Thank you so much for this! I hope you continue working on this and update it further. :-)
doesn't work, gives a bunch of "lora key not loaded"
That I can't help with. It works well in Comfy and Swarm with FP8 and FP16 dev models as well as GGUFs
Pony version sometime please?