Want to know about the Faceswap's Face Extraction process? Got tips, ideas or just want to learn about how it all works? Then this is the place for you
Please explain, I do according to the instructions, but since English is not a happy language, some actions are not clear on this.
in the extract tab, got a set of pictures and json
sort
I delete the bad and unnecessary in the folder. After in the alignment tab I indicate these settings.
Further it is not clear. I make extract from the original video and json. It turns out, a copy of the finished data. Was this step skipped? It serves as a kind of backup.
I do the same with the second video
doing the merge
what to do next. Since training takes 2 folder and 2 json. I tried it, out of 700 pictures I worked, two hours and did not finish, I had to stop. I made smaller requests for 100 images, I did not finish the hour of work either. Created a folder with files. Where it should then be applied. How to use an envelope, not enough files.
Not sure what you're looking for here. Most of those options are optional and you can just ignore them. You can hover over any setting and it'll give you help text explaining what it's doing.