After countless testing runs and lesson learned, and a better dataset, I present the 1.0 version of my vore LoRA!

This is a concept LoRA that depicts the see-through of vored individuals.

Two tags were used in training: vore and mrnm. Be sure to use both :)

Recommended CFG: <3 for a softer look

Recommended Strength: 0.6~1 depending on how much "vore" you want. Please experiment.

If you want to see a more comprehensive comparison between different strengths and epochs, check out the Hugging Face page for that model which will include a BIG grid for you to compare. Some early epochs may be suitable for generating tentacle pit if you're interested.

Text encoder (CLIP) is not trained.

The versions uploaded here are just some of the runs, I upload all my training runs including a bunch of unreleased version to my Hugging Face, be sure to check them out!

License:

Basically CC BY-NC-SA.