Add possibility to return the linear transform in align_with_landmarks

[baba-yaga]

It would be nice to have a possibility to return the linear transform by align_with_landmarks function in pointcloud.py. It may then be re-used with .apply_transform on other objects too. For instance, I fit a mesh to an image img of its rotated view, then apply the inverse transform to the camera showing Image(img) to find the position of the camera in the original mesh space which gives the same view as on img. I am going to describe this in Discussion soon.

Technically, I suggest adding argument return_transform=False in the function call and return lmt at coming to the last 3 lines of the code if it is True.

I wonder if the non-linear transform can be returned from warp function too. It can then be saved and applied later.

[marcomusy]

Hey Sergei, this is already possible for both linear and non-linear operations!
Check out example: examples/basic/align5.py. Try adding

print(s2.transform)

will print:

vedo.transformations.LinearTransform at (0x7f4790b453d0)           
name          : LinearTransform
concatenations: 1
inverse flag  : False
matrix 4x4    :
[[ 1.781236, -0.576785,  0.051535,  0.344465],
 [ 0.577714,  1.758522, -0.286335, -0.019739],
 [ 0.039791,  0.288202,  1.850269, -0.026763],
 [ 0.      ,  0.      ,  0.      ,  1.      ]]

this is a vedo.LinearTransform which you can later use with some_other_mesh.apply_transform(T)

[baba-yaga]

Indeed! Sorry for missing it - thank you!