Multi-input support #9
Description
It is currently not possible, or extremely cumbersome, to have two separate branches of a pipeline intersect as they move forward.
As an example, consider the following scenario:
- a source node that consumes a remote audio stream
- a source node that consumes a remote video stream
- a processing branch that produces text transcription from the audio
At the moment, there is no easy way to impress the produced transcribed text into the video frames.
This can be solved in a number of ways. One could be to implement multi-input buffers, capable of receiving inputs from multiple sources and syncing them together. Another one could be to delegate this process to the nodes, so that they could set their source function over different buffers.