The shoot was very simple and short, it was just me holding my cam and filming the road outside my house, panning from left ro right and a couple of zooms but nothing drastic I believe. When I created an image sequence of this simple shoot and imported it to PFTrack, I adjusted the parameters alittle inorder to give me a good track and deinterlaced it so I would get 25p. However after the tracking, to my surprise the solving proved to be difficult once again for me despite how simple it was, it kept failing at certain frames where the camera pans from left to right resulting the most motionblur but even I wouldnt consider it that bad and there was still some tracking points on it, especially when I even manipulated the image of the footage like the contrast and sharpness to get more of a solid track for complete solving, but to no avail.
After trying this and after and many trackings later, I went and seperated the fields of the footage, thus I would get 50p. I tracked with this and to my surprise the tracking result was MUCH more solid and the solving actually solved the footage, the downside to all of this that it took waaay too long. Because of this, I dont understand why is became so successful, despite there being more frames, but the place where the track was weak and solve fails, there was a big improvement. This now questions me on how will I ever work with matchmoving footage of this kind if shot progressively (24p or 25p)?
On a final note, even with the 50p track and success solve, when playing back the footage within PFTrack, I have a feeling that there is some drift, how would I fix this if it ever was the case?
I hope you understand what my problem is and I would really appreciate it if someone helped me on this. Thanks and look ofrward hearing back guys (and gals hopefully) :thumbsup:
Post a Comment