Im still a little confused about Masks even after reading the quide and past discussions. During the extraction stage i selected all 3 Masker (Unit, VGG-Clear, Vgg-Obs). Then, i start cleaning unwanted images and merging alignments if needed. After that i just start training. Did i miss any step in between related to the 3 mask i have selected? The video works after converting, just thought it could have been better in terms of quality.
Please advise.
Masks are not a magical quality result, they just help define what area of the image gets replaced. In fact, to get the best results you'll actually want to do the editing in a video composition software manually. Masks are provided as an option to get an easier result.
If it's the mask that you're not happy with, try looking at the masks in the manual tool. There are multiple masks available and you can change the threshold for the mask (via a separate option) to get the coverage of the mask just right. You can also edit individual masks which will give a better result than fully automatic masking.