It's work thank you <3.. may i know how many steps, and how many datasets you train this Lora? Thanks in advance

This actually works perfectly and doesn't change the face like I feared from the exclusively asian samples.

Great work! These penis look great, perfect anatomy and details! Could you make a version focusing solely on realistic penis in various positions? (POV, front, side)? That would be fantastic and would be very helpful for NSFW content. Thank you!

Does not work with character LoRAs. Works great on its own though!

I cannot understand why image generation models, starting from FLUX, suffer from severe character distortion when stacking character LoRAs with other LoRAs, unlike video models such as Wan. Even after using block edit nodes, I couldn't achieve satisfactory results. While this likely depends on how the LoRA was trained, video models like Wan generally maintain character consistency quite well even when multiple LoRAs are mixed. Why is this issue so prominent specifically in image models? nice lora tho