Are you sure you are getting the transformation matrix? If I read right your right eye values and align them into the matrix (assuming you dumped them line by line) I get:

```
In [9]: np.matrix('0.64759 0.00000 0.12824 0.00000; 0.00000 0.78750 0.00000 0.00000; 0.00000 0.00000 -1.00020
...: -0.20002; 0.00000 0.00000 -1.00000 0.00000')
Out[9]:
matrix([[ 0.64759, 0. , 0.12824, 0. ],
[ 0. , 0.7875 , 0. , 0. ],
[ 0. , 0. , -1.0002 , -0.20002],
[ 0. , 0. , -1. , 0. ]])
```

Which does not seem to be a transformation matrix (at least in 3D space).

EDIT: I now realized I just took the numbers from the message preview in this thread and did not check the original thread you linked. Now I see what are the matrices, but still, they do not look like transformation matrices to me

I was looking at Unity doc Unity - Scripting API: Camera.GetStereoProjectionMatrix, but it does not really help much either.

Do you know from which space/coordinates to which space/coordinates is this supposed to work? For me the strange is non-zero fourth line and the obvious “non-symmetric” property. Usually the transformation matrices look symmetric (in 3x3 submatrix and apart for the signs) and have the last row zeroes. For example the matrix OpenVR uses does not even define the 4th row (as it is zeroes anyway) and looks like this for the right eye:

```
[[ 0.9848078 , -0. , -0.17364816, 0.03486158],
[ 0. , 1. , -0. , 0. ],
[ 0.17364816, 0. , 0.9848078 , 0. ]]
```

This is the matrix which transforms the coordinates from the eye view to the head view. This is a bit counter intuitive, because for the rendering one needs its inversion (i.e. from head view to the eye view), but I guess they had some reasons to define it this way.

EDIT2: Now I realized that you are talking about a **projection** matrix and not a **transformation** matrix. Well then in that case, I guess it does not say much about the underlying views.