Support for accessing multiple cameras through a single Harvester Core object

[genicam]

Support for accessing multiple cameras through a single Harvester Core object

[kazunarikudo]

Hi Severi,

Thank you for providing us your request!

Do you mean you would like to control multiple cameras in a single Harvester GUI?

Would this also mean that you’d like to see images from those cameras in the GUI, too?

Regards,
Kazunari

[severij]

I meant multiple camera support in case of Harvester Core, but now I actually noticed that this is a duplicate of issue #11. :) So closing this.

[kazunarikudo]

Hi Severi (@severij ),

Let me hear your opition: Which point do you see an advantage of supporting mulitple cameras within a single Harvester Core object?

Say, the current implementation maps a Harvester Core object to a single camera and it's obvisouly minimalistic and simplest (so I like this way from maintainance point of view).

Technically speaking, this way can support multiple devices just having multiple Harvester Core objects and of course, the interface is the same because you're using the same Harvester Core object.

Actually I was thinking about supporting multiple devices within a single Harvester Core object but I have changed my mind having above perspective.

Anyway, I would highly appreciate if you could tell me your point to consider if we should make it or not. In addition, it should be very helpful if you could show some lines of code that you expect to have.

Regards,
Kazunari

[severij]

I would say there is no clear advantage, except maybe being more intuitive to the user (but that might be just me). Having a harvester.device_info_list with multiple cameras makes you think that you could also connect multiple cameras with that HC object, but of course this is impossible with the current implementation.

An alternative implementation would need a separate class for Device, which could be used like

camera1 = harvester.get_device(0)  # Returns a Device object
camera2 = harvester.get_device(1)

or something like that. So the Device class would have many of the features that Harvester class currently has. Then the device objects could be used separately:

camera1.start_image_acquisition()
camera2.start_image_acquisition()

# Acquisition of images with the cameras
...

camera1.stop_image_acquisition()
camera2.stop_image_acquisition()

I think this way the process of connecting and disconnecting of the device could be made automatic, like for example connecting the device in the __init__ of the Device and disconnecting it in the __del__. If there is a separate Device class, then the role of Harvester object would be handling the cti-files and listing the available cameras.

However, I'm not sure if this is any better or more difficult from the maintenance point of view. I want to add that I'm totally happy with the current implementation of Harvester, since it works just fine. :)

Edit:
Since most of the people probably will be working with just one camera, I'm not sure if any alternative solution would be worth the effort. On the other hand the current implementation is quite simple, having just one object (Harvester).

[kazunarikudo]

Nice :-) Okay, I've got your point and would like to take a bit more time to concider a way to implement your elegant usage. My mind had been sticking on a way to use devices within a HC object anyway and it had kept preventing to reach your idea. Thank you!