[sahil_chaudhary]

The model was trained exclusively on stable diffusion generated images, so can be unpredictable with non-generated images, specially images with people in it