All of these images were generated by using Stable Diffusion XL on local machine. Uses some of my custom-made LORA which took at around 25+ hours to train on a decent GPU card with 12GB VRAM.

Those images are really, really, really realistic. A huge upgrade from Stable Diffusion version 1.5 that reaches its peak.

Gallery Link: imgbox.com/g/uotQMQ1hwg