This model has an emphasis on bondage, latex, goo and hyper.
The dataset involved tags for features in art, use words like shading, artwork, render etc. to get the best detail. Potentially use 'cg', '3d, 'rendered' to increase accuracy to artist's style, since they use a mix of 3d assets painted over.
Recommended weighting of 0.8-1.0.
CFG Guide:
Use 5-7 for Pony, Illustrious.
Use 4-5.5 for NoobAI.
Known issues for Pony:
Latex bodysuits may get interpreted as being head to toe, turning characters into latex drones. This seems to happen frequently with equine characters.
Some categories of clothing may cause generation of additional elements, such as motorcycle clothing or bicycle shorts.
Issues with Illustrious
Faces may be low detail if further away in the image
If you have any issues with NoobAI such as prompt accuracy or anatomy issues, use the cyberfix sub model or try lowering the weight of the lora.
Description
Finally got around to making an illustrious version of the lora.
This works way better with giving latex its needed texture, however im disappointed the style doesnt carry across the pseudo 3d look that takahirosi's work has, but that's probably a limitation of illustrious being anime focused.
Faces get quite low detailed from further distances, a face detailer will be required in many cases.