Closed kenoharada closed 4 years ago
Hi! No fine-tuning has been done for any of the generated images. I searched for a person with similar landmarks in VoxCeleb2 and used that video for driving poses. I am not sure what kind of tip you are looking for. An important step when using faces from the wild is getting a good crop (face position and ratio).
Thank you for quick response!
I will try getting a good crop:+1:
Hi! I ended training the model.
Then I try to generate new images in one-shot setting.
Has fine-tuning been done when you create monalisa images in the demo??
And how did you select driving landmark for monalisa??
If you have tips for generate new images, could you share it??
Thanks