This LORA was trained on both SFW and NSFW images, tested with the dreamshaper_8 and epicrealism_naturalSinRC1VAE checkpoints.

Recommended settings for the best results:

Trigger word: painting

Sampling Method: DPM++ 2M Karras, Euler A

Sampling Steps: 24 and above

CFG Scale: 4-8

LORA Strength: 0.64-0.8

If you want more realism in semi-realistic checkpoints, use realistic or photorealistic prompts.