This model is a LoRA trained on about 100 pastel-toned style images, based on the SDXL-Base-1.0 model. Training was conducted on my own machine equipped with dual RTX 4090 GPUs, using the Kohya_ss backend, and I personally optimized the parameters through multiple iterations.

After generating images with various prompts, I verified that the style is well-applied. I tested the LoRA on three main checkpoints:

• SDXL 1.0

• Crystal Clear XL

• DreamShaper XL

Although I have trained many LoRAs before, this is the first one I am publicly sharing, so I welcome your feedback.