I have a set of 42 images made as a rotation around the object and it is failing to generate a model in Visual SFM. From what I can gather, the problem is that just doing a rotation is insufficient. There must photographs taken from above as well.
Is there any way to avoid the need for high angle shots, and just use rotational shots and still get a 3D model from Visual SFM?
This is not a problem specific to Visual SFM, but a general technical problem. The software cannot accurately determine the depth of the pieces which approach perpendicularity with the camera lens. The software uses those varying measurements and estimated angles to calculate the geometry. Check out the law of cosines.