I've been playing with the VooDoo tracker and the import plugin (Thanks Anders!
it hasn't crashed my system even once!).
After searching the web for information, I was finally able to estimate the Filmback values. (camera specs - 1/2.5 inch CCD ~ 5.76 x 4.29 mm)
A few things I am trying to figure out, perhaps some one can point me in the right direction:
1 - I am assuming you use SAVE Textfile.. from the file menu in VooDoo, yes?
At least that is what I have been doing and the data is imported.
2 - The description in the VooDoo txt file labels the last column of camera data "fov : Focal Length" but the value it populates with is the Horizontal Angle of View. Focal Length (what I know) is the distance from the film(back) to the focal point of the lens, and Field of View is normally measured diagonaly is it not?
For example: txt contains fov 32.031073, Focal Length in A:M 32.03. This value does not work, the 3D FPoints slide all around. The Diagonal Opening Angle (from the camera parameters 0r calculated manually) 39.4758 appears to track.
EDIT: OK, I'm seeing things, now a setting around 70 works for the Focal Length. I think I'll take a break from tracking for a day
Can some one confirm this? I haven't had to many film clips that would track well.
3 - The distances for the 3D FPoints are not even close to reality. I'm guessing it's the "Pixel Size: [mm/pel]", but these are being calculated by the program.
example: outdoor scene the results have the camera moving about 38mm in the Z direction and I'm guessing I traveled about 4 meters (so a multiplier of 100).
Any idea what value to adjust to correct the offset? Anders, how about adding an input for a multiplier?
4 - has anyone been successful getting values entered in the Position tab of the Camera config menu (i.e. translation, Pan, Tilt, Roll) to be taken into consideration by the program?
5 - I had my best luck with 320x240 format. Tried 640x480, but haven't been successful in getting a track yet. I also tried processing the TGA's ( Blur, Sharpening, Edge enhancement, B/W....), haven't found a technique that improves the tracking.
Any recommendations on what tracks well?
All in all, it's been a very thought provoking couple of days
(and a rehash of my trig).
Here is a link to some useful data on CCD sizes and the actual receptor dimentions (not that they nessacarily use the whole receptor, but it's all I could find).CCD Sensor sizes