could you provide some test data?

[ForrestPi]

I have tested some data collected by myself.
the results are terrible.
could you provide some test data showed in your readme?

[Qiulin-W]

Hi, I added some videos to ./assets, have fun!

[ForrestPi]

@Qiulin-W thanks providing test videos.
Animation demo seems good. but the face swap demo I tesed by myself still seems not so good.

[ForrestPi]

resultSAFA.mp4

@Qiulin-W

[Qiulin-W]

resultSAFA.mp4
@Qiulin-W

For face swapping, it is something like "copy and paste" in our framework, which relies on a face segmentor and we deform the source face so that it can fit to the target face. The GAN loss during training mainly optimizes the deformation field instead of making realistic face swapping (e.g. face boundary).