Is it possible to redefine camera positions?

[2B||~2B]

Hi,

I'm developing an experimental set-up in which I have a 3D complex object that changes in time and I would like to get the 3D model of it every 30sec or so and I would like to try VisualSFM for this purpose, it looks really cool!

The set-up is a smaller/simpler version of the one shown here: http://www.sampleandhold.co.uk/#whatwedo

1) The object stays more or less on the same spot and the camera array is fixed. Is there a way to pre-define the position of the cameras to get more accurate results? This should reduce a lot of degrees of freedom.

2) The object is somewhat transparent. Are there any recommendations for the type of marks to be used (shape? color? size?) to maximize feature detection?

Cheers :-)