Diagonal scaling matrix get incorrectly initialized to a full-one matrix

[Mistariano]

Hi guys,

I wonder that ths S matrix defined in computeCov3D funcs for both fwd/bwd computations is now incorrectly initialized to full-one:

diff-gaussian-rasterization/cuda_rasterizer/forward.cu

Line 121 in 59f5f77

glm::mat3 S = glm::mat3(1.0f);

If I understand correctly, here we expect an identity matrix instead:

glm::mat3 S = glm::mat3(
    1.0f, 0.0f, 0.0f,
    0.0f, 1.0f, 0.0f,
    0.0f, 0.0f, 1.0f
);

Fortunately, such a full-one matrix with a scaled main diagonal is usually positively defined and symmetric. Thus, the reparameterized $R^TSSR$ is still a covariance matrix.

I have roughly checked that this issue is not so influential to the PSNR performance. If required and agreed, I'm more than glad to submit a hotfix PR for it.

[ys-koshelev]

@Mistariano Please beware, that according to the Sec. 5.4.2 of OpenGL API docs

If there is a single scalar parameter to a vector constructor, it is used to initialize all components of the constructed vector to that scalar’s value. If there is a single scalar parameter to a matrix constructor, it is used to initialize all the components on the matrix’s diagonal, with the remaining components initialized to 0.0.

For this reason, glm::mat3 S = glm::mat3(1.0f); provides identity matrix, not the matrix full of ones.

[Mistariano]

@ys-koshelev Thanks a lot for your kind reply. I will close this issue.