Tracking first shot, note the graph at the bottom shows good (green) features that are tracked perfectly, any red or yellow points can be filtered out as in the screenshot below so the camera solve only reads the best features to get a more accurate track..
Camera solved, 3D data shown in context on the right panel with a camera and points.
This track was very simple as there is a lot of parallax in the shot to work with and get 3D data with foreground and background points.
This shot was a bit trickier in terms of the stages needed to produce a good solve. First because there is a moving object the train must be masked out so the tracker ignores it otherwise it would try to use the points on the train as reference points when positioning the 3D camera.
Aswell as the moving train, the light moving across the floor in the foreground causes problems as the feature pixels change colour or lightness rapidly. Luckily the background stays largely untouhed by movement or light changes so a track was fairly easy but a lot of yellow tracks had to be filtered.
This was another very simple track as there are a lot of static points.
Camera solved, 3D data shown in context on the right panel with a camera and points.
This track was very simple as there is a lot of parallax in the shot to work with and get 3D data with foreground and background points.
This shot was a bit trickier in terms of the stages needed to produce a good solve. First because there is a moving object the train must be masked out so the tracker ignores it otherwise it would try to use the points on the train as reference points when positioning the 3D camera.
Aswell as the moving train, the light moving across the floor in the foreground causes problems as the feature pixels change colour or lightness rapidly. Luckily the background stays largely untouhed by movement or light changes so a track was fairly easy but a lot of yellow tracks had to be filtered.
This was another very simple track as there are a lot of static points.