The lora model has been tested on the base model sd_xl_base_1.0 (works with other checkpoints) with different
CFG settings: 5–7
Steps: 25–40
Sampler: DPM++ 2M Karras / DPM++ SDE Karras.
It can also participate in nsfw in ultra realism.
The lora model has been tested on the base model sd_xl_base_1.0 (works with other checkpoints) with different
CFG settings: 5–7
Steps: 25–40
Sampler: DPM++ 2M Karras / DPM++ SDE Karras.
It can also participate in nsfw in ultra realism.