![]() |
OpenNI 1.5.2
|
The first thing to do when wanting to create new node implmentations is to define exactly what nodes we want to implement and the input each node needs.
For example, if I have a physical device which is able to create both depth data and color image, I would want to create two node implementations: a depth generator and an image generator. It is also highly recommended to create a device implementation. The device node does not necessarily participate in the data flow, but can be useful when wanting the represent the physical device (to get serial number for example).
In the above example, we'll define our topology as follows:
Another example would be creating a module which is some sort of algorithm over another input. For example, a hand generator which identifies hands in any color image. In this case, our module will include a single node implementation, a hand generator, which will depend on any image generator.