Some XL-based checkpoints I've noticed are lacking in their capability to produce realistic and detailed images of vaginas. That's why I took it in my own hands to make a LoRA that makes such images close to perfect (in my opinion of course).

A DoRA for optimizing the generation of vaginas. It is trained on SDXL, and is great for both txt2img and inpainting.

My go-to Parameteres:

Sampler: DPM++ 2M SDE

CFG-scale: Around 3-5

Resolution: 1024x1024

Steps: Around 30. 20 or even 15 is often enough