LoRa trained on 40 images of mostly Margot Robbie's face at 768x768 resolution. Recommended with LoRa weights between 0.8 and 1.4 and with semi-realism models (dreamshaper, reVanimated etc.)
v1.1 : Retrained on full images at 1024x1024 pixels. Can directly generate 1024x1024 pixel images without upscaling.