AR tracker only supports Kinect or RGB cams. #3 uses RGB currently, so it's not using the ZED's depth data.
The Kinect version subscribes to a colored point cloud, rather than a separate image & point cloud stream, and just relies on RGB data in the point cloud to reconstruct the image for OpenCV. The ZED wrapper can also do a colored point cloud, so it may be as simple as that. If not, it should be a trivial amount of work to pipe in a separate image and point cloud topic.
AR tracker only supports Kinect or RGB cams. #3 uses RGB currently, so it's not using the ZED's depth data.
The Kinect version subscribes to a colored point cloud, rather than a separate image & point cloud stream, and just relies on RGB data in the point cloud to reconstruct the image for OpenCV. The ZED wrapper can also do a colored point cloud, so it may be as simple as that. If not, it should be a trivial amount of work to pipe in a separate image and point cloud topic.