Hi,
I have imported my own face mesh into the Audio2Face app from 3ds Max. I followed the online tutorials for character transfer but when I press mesh fitting, it throws me a “Factor is exactly singular” error. Interesting after post wrapping and while testing the audio2face I see a slight movement, not on the lips but on top of the head. I am attaching the screenshots of the matched nodes and also console error and also the final recording for your reference
a2f_output.mkv (566.8 KB)
.
Are there any prerequisite characteristics a mesh needs to have inorder to generate lip sync? I am not sure if I am doing anything incorrect. Let me know if I can get this to work.
Thanks!
Hi @ravi.konkimalla! Appreciate you reaching out to us. I’ll get a developer to take a look at your post!
1 Like
Hi @ravi.konkimalla ,
Are you able to share the head mesh and your setup scene so I can take a closer look?
Thank you.
Would it be okay, if I share the face mesh and my setup via email?
yes you can.
My email address is esusantolim @ nvidia.com
Hi @ravi.konkimalla ,
I loaded your file in Maya to see if there’s anything wrong with the mesh, and noticed there’re some single floating polygons on the mesh.
In the image I switch to face selection mode, then double click on the mesh to select the entire polygon island and move them. there are a few left behind as circled in the picture.
The problematic face id FaceShape.f[4026:4027] FaceShape.f[5985:5986] ;
Here is an example when character transfer is done. It looks like I need more work on tagging the lips :)
Thanks a lot Edy!! I will make sure to remove those floating polygons and try it again.
Thank you!
