Hello,
First congratulations for the quality of your work.
I had a few questions regarding your algorithms :
What is the most important factor to have a qualitative source for faceswapping ? Like is there a quantification of the quality of the mask or landmarks (the size, the angle...) ?
How to modify the way multiple sources and targets are handeled. If fact i'd like to have only 1 source swapping to 1 target instead of having each individual sources swapping with all the targets. Is there a way to modify the code accordingly without loosing myself in each and every file ?
Hello, First congratulations for the quality of your work.
I had a few questions regarding your algorithms :
Hope to hear from you soon !