Trained on a dataset of high shutter speed captures of water in different lighting scenarios. Interesting to apply to anything. I would love to see what you create. Tip, use this embedding with the Depth Model in img2img. I will be uploading a few version that are less trained, each variant is so different.