JiaxiangZheng/DeformationGraph

The constrain part in Jacobian mat may be wrong

fishfishson opened this issue · 2 comments

you may loss the derivatives of rotation mat for each vertices in constrain part.

fJac.coeffRef(row, col+9 ) = weight_regular;

I have problem about the mesh format. I use OFF file and still met following problem. Could you please provide clues to deal with it ? Such as what VV and vid should be? Thank you very much. I print the mid-process as following:
./test_FPS ./triangle.off
PrepareFastMarching...
VV before build:0
has build the vertex_triangle connectivity!
has build the triangle_triangle connectivity!
start FarthestPointSampling...
VV size: 1
vid: 18842
VV size: 1
vid: 0
max dist:=1e+15
VV size: 1
vid: 1
Segmentation fault (core dumped)

Because the author simplifies the constraint term, a point is only affected by one node. Besides, because the construction node is sampled from the original cloud, the result of that term is 0.