Tracking test

I have finally got my tracking to look right. I had a lot of problems with making boujou understand the geometry of where the floor and wall were. The calculation of the track was not in the correct place and instead produced a point cloud (the tracking points) to spread everywhere and unclear in 3D space.



Here is a side shot of my boujou track:



I did tried to set specific geometries where I knew a flat surface will be, but unfortunately I still couldn't get a good enough data of the scenes’ floor. I then decided to take the tracking into XSI to see what result I can achieve from that information. After a lot of trial and error I finally found a place where objects seemed to stick and feel as though they are in the scene.

Here is a test of some objects that have been placed around in the green screen room with some reporters:


Tracking test from Daniel Chirwa on Vimeo.

From doing this test, I understood that the reason why boujou did not understand where the geometry was set. This was mainly due to lack of information within the green screen room. What I mean by this is that the scene needs more objects to identify the edge and also the prospective when the camera moves its position. So overall, if I had to re-shoot this scene I would recommend placing square boxes around in the green screen, to not make the tracking program get confused. Hopefully now I will be able to link this tracking with my castle scene to make the shot feel seamless.