DeepFaceLab 2.0 MVE Fork & DeepFaceLive Tutorial

Table of Contents

00:00 – Start
0:56 – DF Lab MVE Software and model files on Google drive
7:19 – Additional Software & PC Hardware recommendations
24:30 – Installing DF Lab 2.0 w/ MVE Fork
27:33 – DF Lab batch files and workspace folder contents explanation
1:02:05 – Fixing JSONSCHEMA Script error
1:02:58 – Testing demo footage of Elon/Robert – Basic XSEG masking
1:16:42 – Setting the Windows Pagefile
1:19:07 – Faceset Creation (Includes Generic XSEG, creating Faceset.pak file, etc)
1:46:45 – RTT Model & RTM Faceset files
1:49:29 – SAEHD Training Settings (Model begins training)
2:04:10 – Loss Values Explanation, continuing to train
2:17:32 – Using the “Tooth Trainer” faceset to fix teeth
2:22:02 – Teeth are fixed, back to RTM training
2:25:44 – Disabling Random Warp
2:30:41 – Enabling MS-SSIM
2:38:28 – Changing Learning Rate to 3e-05
2:40:34 – Changing Learning Rate to 1e-05
2:43:39 – Learning rate back to 5e-05, enabling LRD (Learning Rate Dropout)
2:47:18 – Enabling GAN
2:55:05 – GAN Grid patterns
2:59:02 – Setting up Deepface Live
3:01:26 – Exporting SAEHD DFM File, testing with DF Live
3:19:30 – Using Deepfacelab to make a Deepfaked Video
3:39:46 – Using the Interactive SAEHD Merging tool, Outputting the final video clip (result.mp4)
4:02:13 – Closing comments