You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have seen in multiple videos of your team using FLAME with existing 3D models, for example Churchill's head in VOCA. I have downloaded such models but the script fit_3D_mesh.py cannot use them as input as they have a different number of vertices and, I suspect, not exactly the same mapping between vertices and parts of the head. Thanks for the help!
The text was updated successfully, but these errors were encountered:
Yes you are right, fitting FLAME to a scan of arbitrary mesh topology requires a differentiable distance function that quantifies the difference between the scan and the FLAME surface (e.g. some scan-to-mesh or mesh-to-scan distance). You would need to replace the used distance function (i.e. the distance to the 2D landmarks, 3D landmarks, or the vertex-to-vertex distance in the provided experiments) with this new distance. Please see the FLAME paper for a more extensive and more formal explanation. Having this said, we sadly don't have such a distance publicly available due to licensing issues.
I have seen in multiple videos of your team using FLAME with existing 3D models, for example Churchill's head in VOCA. I have downloaded such models but the script fit_3D_mesh.py cannot use them as input as they have a different number of vertices and, I suspect, not exactly the same mapping between vertices and parts of the head. Thanks for the help!
The text was updated successfully, but these errors were encountered: