Trained on 500 hand picked images. This is the iteration before I try a new novel method.
443044_training_data.zip
Not-Midjourney-V3-SDXL-Beta2-000005.safetensors