Trained on V-PRED
The lora is trained on the images based on the 1.5 checkpoints on here: https://civitai.com/user/Sdive/models1.5 is flawed so expect some images coming out weird.
