Trained with only 17 pictures of my own face.
Trained with new higher quality images at 1024x1024, more consistent higher quality generations.
_CKHv2.safetensors