Solve camera algorithm

[veclock]

Hi, where can I get more information about the algorithm Matchmover uses when solving the camera?

I'm writing a school project and I need some information about how this function works.

For example, I sometimes get this message:

"Cannot locate reference/key frames necessary for camera solving:
a minimum of 2 images with 7 points in common or 4 points with known 3D coordinates
are required to create reference/key frames"

However, I can't find any further information about what the function requires in order to calculate correctly? Also, I can overcome this minimum point limit by tracking a point twice.  

For example, after some trials and errors I've discovered that the system is linear, the footage shouldn't have lens distortion. Also, markers can't be placed only for example on the floor, at least one must be placed higher or lower in order for the algorithm to calculate correctly.

Also, the algorithm must use epipolar geometry math, right?

I need answers and sources about this information, could someone please help me?

Thanks in advance