kabralenguon
11-10-2010, 05:48 PM
Hi Dt's
Im not sure my 3dCamera solve is showing enough parallax.
Camera is almost on the ground plane, dollying left to right.
The 3dTracker got all the trackers from the distant green screen but none from the foreground. I fed it with the Mask to force the solution to be also in the FG area, but after the solve, al those trackers are discarded as unhealthy, and all the green ones are the far ones.
1) How can I tackle this?
to Tell Nuke please not disregard the 2d trackers that I know will help for a good distribution of features in Z.
2) How do I know this is a good solve or not? Its a bit confusing: I played around with cubes and it works in the distance (moves smoothly attched to ground plane), but just trying to match by eye the location of a cube in the FG is not going to work for judging the parallax as i'm lacking the locators that should be in that part of the picture to paste into my cube. Notwithstanding, I attempted that by duplicating my distant cube and move it in Z towards the camera to see how it goes. I found this:
A doubt:
The motion (dolly L to R) still looks kind of alright, But I wonder whether the FG cube should move faster or not ,given the distance to camera.
A certainty:
Judging other elements on the pic, I can tell that the FG cube that I translated towards camera in Z, should be at least 2 times bigger. Wrong Scale.
Does all this information means that I should start from Scratch again? Or is there a way to troubleshoot it?
Thanks!
Im not sure my 3dCamera solve is showing enough parallax.
Camera is almost on the ground plane, dollying left to right.
The 3dTracker got all the trackers from the distant green screen but none from the foreground. I fed it with the Mask to force the solution to be also in the FG area, but after the solve, al those trackers are discarded as unhealthy, and all the green ones are the far ones.
1) How can I tackle this?
to Tell Nuke please not disregard the 2d trackers that I know will help for a good distribution of features in Z.
2) How do I know this is a good solve or not? Its a bit confusing: I played around with cubes and it works in the distance (moves smoothly attched to ground plane), but just trying to match by eye the location of a cube in the FG is not going to work for judging the parallax as i'm lacking the locators that should be in that part of the picture to paste into my cube. Notwithstanding, I attempted that by duplicating my distant cube and move it in Z towards the camera to see how it goes. I found this:
A doubt:
The motion (dolly L to R) still looks kind of alright, But I wonder whether the FG cube should move faster or not ,given the distance to camera.
A certainty:
Judging other elements on the pic, I can tell that the FG cube that I translated towards camera in Z, should be at least 2 times bigger. Wrong Scale.
Does all this information means that I should start from Scratch again? Or is there a way to troubleshoot it?
Thanks!