This is an updated version of Racing Luka 2017 character LORA, trained with SDXL-based Illustrious model, much better than previous ponyXL-based LORA ones. Works well with most of Illustrious-based models. Should be used with various of sampler types. See images here for prompt examples.

Sampler: DPM2 or Euler A

Schedule type: Karras or equivalent (Not Automatic/Normal)

Sampling Steps: 30

*These images here are locally generated with Stable Diffusion DirectML version with Nvidia RTX 3060 12GB.