Loading...

This topic explains the section "Camera XYZ Calibration" of the ID Tag node, for general information about IR ID tracking and links to other chapters please see the topic ID Tag Tracker.

This section includes all settings for the setup of the fundamental XYZ World description needed when working with multiple cameras in 3D and 2.5D mode. If you work with one camera only or only in 2D mode, you have the choice to skip this step and work with a Range filter node instead.

In general, if one determines a XYZ coordinate, first, an origin of the 3D space is needed, meaning the coordinate system's origin, the orientation of the X-, Y- and Z-axis as well as their scaling.
Only when the cameras share the same 3D space information, they can agree on correct XYZ position values for a tag. In other words, only if the origin for each camera can be determined accurately, their reported tag position coincides. If the origin from one camera does not correspondent with the one from another camera, they position the tag differently. As long as both cameras see the tag, its final position is averaged but as soon as it is covered for one camera, it can not be averaged anymore and could jump significantly.
The origin calibration also means that the cameras themself are positioned in the 3D space, this is why this section is called "Camera XYZ Calibration". This can be observed in the 3D mode in the Preview section.

For calibrating the origin for one camera or all of them at the same time, you need to measure and mark a rectangle somewhere on your stage (i.e. four co-planar points). For most applications a 1-2m² spot that is seen by all cameras is ok, but this depends on the distance and orientation of the camera(s). For example, if a camera sees a very large area, a 1m² spot could be too small to achieve accurate results. Enlarge the rectangle until it covers 1/4 - 1/3 of the camera image seen in the Preview. Ideally it should be positioned in the center of the camera image. For complex applications, you might need to constitute several rectangles for different cameras.
For the sake of simplicity, you could mark four points on the floor. To achieve a better accuracy, the rectangle should include the origin itself, the coordinate 0,0,0. Now, you may decide how the orientation of your 3D space should be. For most scenarios, it makes sense to place the XZ plane onto the ground. A positive X axis means "move right" in Pandoras Box; a positive Z axis "move backwards"; a positive Y axis "move to ceiling". Having these axises in mind measure the XYZ position for each point of the rectangle (in meters!) and copy the data to Widget Designer into the "Origin Settings" dialog as described below. It's a good idea to number the corners 1-4 to remember the order. If you need to work with more than one rectangle, measured these points starting from the 0,0,0 point!

Note: Widget Designer now also offers a "5 Point Calibration Mode". Instead of calibrating the cameras with four points forming a rectangle, you can now also calibrate with five points. This method is of interest when you need to achieve a very precise tracking or if you are not satisfied with your results using a rectangle. In order to make text more readable, the manual will talk about four points or a rectangle, but all information applies comparably to the "5 Point Calibration Mode".

In the next step, the four points belonging to one rectangle will be found in the image of each camera. There are two modes available, an automatic and a manual one. Depending on the camera's position, the rectangle's shape looks different and this is how the camera's position is calculated.

As said above, there are two modes available to find the four points of the drawn rectangle in each image of a camera: the "Mouse Mode" and the "ID Tag Mode" which is easier and faster but requires four different tags.

If some cameras do not see the rectangle, a second one needs to be drawn, measured and equipped with ID tags. Redo all above steps, only for step 2a, do not select "All Cameras" but a single one.