Closed ButoneDream closed 7 months ago
Sorry for being late. This distortion happens because StyleGAN is a uni-modal model, i.e., it can only produce the results within its distribution, human face in this case. For inversion, the image input should be aligned. Also, the diffusion-based freedrag (also be provided) is recommended for real image inversion due to the higher generalizability of the diffusion model.
thanks however, i could not find how to try Diffusion-based FreeDrag demo ? could you help me ??
run the drag_ui.py in the FreeDrag-diffusion-version file
2 question 1.upload image slowly
How can this be resolved?