This model was trained on high quality erotic/amateur/hard/softcore images and used Realistic Vision v2 and epiCRealism v1 as the base.
Trained with a noise offset of 0.15.
All example images were generated using only this model with no embeddings, lora's, noise, extensions, upscaling, restore faces or CLIP skips (I hate when you can't reproduce sample images too).
First 4 images are 512x800, rest are 576x768.
See training data for a list of all available captions/concepts.
VAE not included but I recommend the default "vae-ft-mse-840000-ema-pruned".
Thank you to Realistic Vision and epiCRealism for the base of this model and all the model creators merged with these.
