Lora based on artkoji's style. Works best at 0.8/0.9 if you're not upscaling. Use at 1.0 if upscaling.
artkoji.safetensors