This checkpoint merge is a Rosetta Stone for drawing many of my early LoRAs. The majority of them were created with datasets of images I made using this checkpoint and then trained using base SDXL. I've begun training many of my more recent gay themed models using this checkpoint as the base. It creates much better NSFW models than base sdxl because it already knows what NSFW looks like in addition to having a danbooru trained text encoder. This model is related to the Illustrious checkpoint and also the NoobAI checkpoint, for reference. The checkpoint has been pruned to a lightweight 3.5 GBs and has been encouraged to respond to prompts for males and for gay NSFW, but only slightly, it can still reliably draw females and SFW themes. The default model style is very anime/illustrious inspired but can be prompted easily to depict more 3d-esque or pseudo-realistic visual styles. If faces render too anime-sh for your tastes or you find faces appearing animalistic(furry), the negative prompt tag no_nose is the solution, in my experience.

Positive Prompt Ideas: realistic - photorealistic - 3d_rendering - realistic_shading - realistic body hair - high resolution 8K

Negative Prompt Ideas: snout - anime_coloring - anime - cartoon - 1girl - pussy - breasts - long_hair - twins - no_nose

This is essentially an anime checkpoint with some photo-realistic weights sprinkled in. The text encoders in this model are responsive. 🤩 🤩 (they will draw what you prompt as good or better than any other checkpoint I'm aware of)

If gay NSFW LoRAs and checkpoints interest you, follow my page. I also create SFW LoRAs that are themed fantasy/magic and or psychedelic/special effects. I have about a dozen unreleased LoRAs in my catalog and I am still actively creating new ones. I also have dozens of checkpoints that are gay NSFW friendly I am working on with the end-goal being psuedo-realistic and photo-realistic checkpoints that can draw the gay.