Summary of project:
Im working on VFX shot that will need a solid camera track. The footage is of a golfer walking down a fairway with the camera doing a sweeping 360 move around golfer. I’ve split the scene up to 4 sections (it will speed ramped and blurred to cover the seams) so each scene is doing 120 or so degrees (overlap needed between shots to cover seams) of the 360 camera move. Here’s the full 360 clip (this file is flat LOG). The footage I’m using to track is colored and broken down into the 4 scenes that overlap each other to make up this shot.
I get great 2D tracker data but when I 3D solve my footage, the track points on the ground are inverted in the Z space (check linked screen shot, the closest points on the footage are the farthest points from the camera). I have dumped countless hours into trying to make this work and after 30-ish attempts– it finally work. However I did not change anything major for it to work so I have no idea how to fix this issue. and I have 3 more shots to track and they were all producing the same error. But after 5 hours of trying for the second shot, it finally worked for the second time. but still not due to any particular reason. Now I have 2 more shots to track but don’t have the kind of time to sit around for 10 hours hoping it will work…
– Import .PNG sequence into Motion Tracker and set resampling to 80% and verify plenty cache space
– Auto track the scene with 1500 tracks and a minimum spacing of 35
– Clean up all bad tracks in Motion Tracker Graph View and delete the ones on the golfer and water
– Run 3D solver with focal length unknown but constant (used a prime lens)
So here are some pointed questions:
-Anyone know a solution?
-Why would C4D invert foreground points in Z space?
-At this point, I’ve done many 3D solves on the same shot with same settings, and Ive gotten various results. Some with a ton of green dots and others with very few gray/red dots. Some where the solved camera moves around the tracker points, others where the camera locks position and just pans etc.. Why would the same settings/footage produce drastically different outcomes?