Multicam in C# -- processing the frameset

[SChop3]

Hello, I'm looking at processing the data from multiple cameras in C# and I have a few questions with regards to the appropriate way of processing the dataset.

With a single camera

When there's only a single camera connected, I use the Pipeline class to get a frameset

pipeline = new Pipeline();

cfg = new Config();

PipelineProfile profile = pipeline.Start(cfg);

frames = pipeline.WaitForFrames();

depthframe = frames.DepthFrame;

colourframe = frames.ColorFrame;

To align the colour frame to the depth camera I use the Align class on the resulting frameset

FrameSet colourindepth = align.Process(frames);

With multiple cameras

Using code snippets from the examples, for multiple cameras I'm using a different approach. (Code simplified for brevity)

using (var ctx = new Context())

{ DeviceList cameras = ctx.QueryDevices();

// For now, going to loop through the cameras

foreach (Device camera in cameras) {

FrameQueue depthqueue = new FrameQueue();

FrameQueue colourqueue = new FrameQueue();

SensorList sensors = camera.QuerySensors();

Sensor depthsensor = sensors[0];

var depth_profile = depthsensor.VideoStreamProfiles.First(); Sensor coloursensor = sensors[1]; var colour_profile = coloursensor.VideoStreamProfiles.First(); //Start sensors depthsensor.Open(depth_profile); depthsensor.Start(depthqueue);

coloursensor.Open(colour_profile);

coloursensor.Start(colourqueue); // Receive frames Frame colourframe = colourqueue.WaitForFrame(); Frame depthframe = depthqueue.WaitForFrame();

}

}

Questions

However, in this case I only fetch a single 'Frame' object at a time. How do I align the colourframe to depth?

I've tried to use the .waitforframeset method but it doesn't seem to be supported for this method. Also, the Align class requires a Frameset object?!

How do I get a frameset in this instance? Should I be using a pipeline to get the frames? Do I pass the handle to the camera when constructing a new pipeline?

Any help would be greatly appreciated

[SChop3 · Accepted solution]

Hello all,

I've worked on this looking for a workaround and have found that the key is to stick with the pipeline method. The configuration class allows you to create a custom pipeline based on a camera's serial number

I'm still starting with the camera context as before but this time I'm using a pipeline with a specified camera

1. using (var ctx = new Context())
2. {
3. DeviceList cameras = ctx.QueryDevices();
4. // For now, going to loop through the cameras
5. foreach (Device camera in cameras)
6. {
7. cfg = new Config();
8. cfg.Enabledevice(camera.Info[CameraInfo.SerialNumber])
9. PipelineProfile profile = pipeline.Start(cfg);
10. }

Then, when you use that profile object to wait for frames, it will wait for frames from the specified device and give you a frameset you can use in an Align object.

If anyone else is using C# for multicam I'd reccomend this approach.

Apologies if the code is slightly wrong, it's from memory.

[MartyG]

Thanks so much for sharing your results with the community, Wiredchop! C# resources are always especially welcome.