The pose matrix and its inverse

[wangarad]

First I want to say thank you for making this wonderful software!

This is not an issue, but rather a question concerning the pose matrix and its inverse;

• In the castRay() method, pt_camera_f represents pixel projected into camera space.
• As I understand, the point should be projected into world space and saved to pt_block_s.

I fail to understand why is the inverse of pose matrix (invM) used instead of the pose matrix: pt_block_s = TO_VECTOR3(invM* pt_camera_f) * oneOverVoxelSize;

AFAIK invM = inverse of pose = extrinsic matrix = world-to-camera transform. And pose matrix = camera-to-world transform.

What am I missing here?

[sgolodetz]

If I remember right, the InfiniTAM pose matrix is world-to-camera. So invM is the inverse of that, i.e. camera-to-world.

[wangarad]

Thank you for clearing this out, @sgolodetz, that makes perfect sense! I'm closing this issue.