Great work Andrew and your team on the new Wrap3 ![]()
Are you guys planing to have any sort of feature detection (OpticalFlow, texture analysis, template matching in UV/tangent space…) on meshes to populate the coordinate array for the non-rigid alignment?
We are trying to automate this process by detecting similar features (like corner of eyes and mouth and ears etc…) outside WrapX/3 and generate the non-rigid control points.
This is for all the meshes, meaning the generic base mesh + neutral hirez mesh + all the FACS and you can see that with around 35~40 points in WrapX on each mesh and 16 Facs it becomes pretty time intensive and an error prune task for artists.
I can see that the algorithm in Wrap3 is faster and performs better comparing to WrapX with less control points so such features could greatly benefit us and maybe others…
Again great job and let me know your thoughts,
Cyrus