Updates in v2:
Illustrious XL 1.0 as the base model. It seems to be better than Illustrious XL 1.1 and 2.0 as a base model to train LoRAs.
More balanced image sizes and aspect ratios in the training data. Now the LoRA works well for both 4:3 and 16:9 images.
Prompt: scenery, no humans, indoors/outdoors, day/sunset/night/rain
