The LoRA was created using a dataset primarily composed of highly rated images collected from WAI, HassakaXL, and SmoothMix.
License: MIT
Checkpoints:
Wai-illustrious
HassakuXL
Recommended Sampler / Scheduler:
dpmpp_2m / sgm_uniform
dpmpp_2m_sde / karras
CFG / Steps:
CFG: 3.6 / 27
CFG: 4.6 / 50