This model was trained on SD1.5.
CFG 6-8 and anything above 20 sampling steps.
Model used: realisticVisionV20_v20
trishahershberger.safetensors