For Best results: 1. Picture should be high quality. 2. Face expression should be neutral. 3. Click unlit option and change the POV during mesh to metahuman step. Note: This Open source project uses CUDA enabled graphics card to run. I used RTX 2060 in this tutorial, RTX 30xx series didn't support because of driver issue, I tried a lot with RTX 3070 but couldn't run. Links: 1. 2D image to 3D Face: 2. face-alingment package for landmarks detection: 3. Python file link for landmark extraction from the input file(place this inside the “Deep3DFaceRecon_pytorch“ folder):
Hide player controls
Hide resume playing