The difference between version 1.2 and 1.1 seems subtle, with the main variations appearing in the background.
After training, version 1.0 appears to lack adequate fine-tuning, resulting in an unclear style, although it does have a unique character. It was further trained for 12 epochs and 20 epochs, leading to versions 1.1 and 1.2 respectively.
Adding other style prompts may overshadow the figure style LoRa, but it can make the image appear more three-dimensional.
