I am trying to do 3D reconstruction using Photogrammetric techniques, I have two images that I want to reconstruct from

I perform feature detection and matching which gives me good matches so that isn’t an issue.

I then use cv::findEssentialMat to get the essential matrix from the matches and use cv::recoverPose to get the transformation matrix 2 relative to transformation matrix 1.

I then triangulate the points using cv::triangulatePoints(projecMatrix_1, projecMatrix_2, corr_p1, corr_p2, world_points);

Now world_points is a Nx4 matrix that describes points in homogeneous coordinates,

$$
begin{pmatrix}
x
y
z
w
end{pmatrix}
$$