What is better solution for face jitter? The landmark smoothing can help to some extend but it is not perfect. Can augmenting the cropped faces with random shift in position help?
In the paper Adaptive Blending is mentioned but only static blending is found in the repo. Can you share adaptive blending implementation.
How to train ghost in higher resolution like 512px?
What is better solution for face jitter? The landmark smoothing can help to some extend but it is not perfect. Can augmenting the cropped faces with random shift in position help?
In the paper Adaptive Blending is mentioned but only static blending is found in the repo. Can you share adaptive blending implementation.
How to train ghost in higher resolution like 512px?
Thanks