Hi,

I have written some code that transforms several points. I have used the homogeneous() method in order to be able to multiply my dimension 3 vectors by the 4×4 transformation matrix:

Code: Select all

Eigen::Matrix4d Trans;
Eigen::Matrix3Xd points = Eigen::Matrix3Xf::Random(3,5);

Eigen::Matrix3Xd res = Trans * points.colwise().homogeneous()

My problem is that, while Eigen 3.2.1 returns a 3×n matrix, some previous version (3.0.5) returns a homogeneous 4×n matrix.
Currently, I'm solving it by picking only the first 3 rows,

Code: Select all

Eigen::Matrix3Xd res = (Trans * points.colwise().homogeneous()).topRows(3);

but I wonder if a better method is available, since I fear that the topRows() call might impose an additional performace load, considering that newer Eigen versions will not need it.

I don't remember we changed that behavior. Anyway, you can rather use .topRows<3>() to avoid any overhead.

Great, thanks!. I forgot the templatized version.