This will be slightly off topic, but I figure this is a smart bunch. I'm working on a perl script to help our trackers get nodal pan shots out faster. Generally I find shake to do a really good and quick 2d track and we have plenty of licenses. So, on to the issue: I've baked out the 2d track data from shake's internal tracker into their txt file output. The script is written in perl and will run externally from any programs. So far, I've got the data categorized and normalized so that the first keyframe starts at 0,0 and everything moves from there. Consider this data set:
CODE
X Y
0.000 0.000
3.334 2.365
8.256 -2.023
etc.
I want to convert these movements to the rotational values of a locked down (nodal pan) camera. Considering that in a 3d app, you could have those points in space and a camera that is constrained to aim at it, I know that the conversion is possible. Has anyone had any one encountered any formulas or ideas that may shed some thought on this topic? Essentially, once finished (tomorrow), the script will output a chan file to be read in by a 3d camera in Nuke.
Cheers,
-Scott
