[Bf-committers] LibMV versus OpenCV / VXL

[Troy Sobotka]

Just read Tom M's posting on LibMV and was wondering where the
discussions have taken place for it regarding future directions.

Nuke apparently (according to the User Guide[1]) harnesses VXL for
some of its algorithms. OpenCV, for example, has a simple relevant
point optical flow tracking example (lkdemo.cpp 151 lines of code[2])
that would seem to at least be worth examining, if it hasn't already
been considered.[3]

I am wondering the upside benefit of pushing LibMV over existing options.

Sincerely,
TJS

[1] http://thefoundry.s3.amazonaws.com/products/nuke/documentation/NukeUserGuide_6.1v5.pdf
[2] https://code.ros.org/trac/opencv/browser/trunk/opencv/samples/cpp/lkdemo.cpp?rev=4240
[3] I am aware of Ton's reluctance to add another dependency, but
perhaps the complexity of computing vision here warrants it?