|
Deepfacelab is so fuckin easy. You add 2 video files - source and destination and you start 7-8 .bat files one by one. Before you start training you remove from extracted faces the blurred ones and those without faces. And you need hi end graphics card wit CUDA and at least 4GB video memory. 8-16GB is preferred (RTX cards). Even with good GPU training can take 3-4 days. You train as long as faces on preview are sharp enough.
|