Applying post-processing ...

[PTozo]

The RealSense Viewer shows these post-processing options, which improve the quality of the depth buffer considerably:

What's the best way to apply these in my own app? I notice that the SDK has functions like rs2_create_decimation_filter_block() and rs2_create_temporal_filter_block(); presumably there's a certain optimal order in which to apply these?

Also, gaven't dug into them much yet but I'm assuming they can be applied repeatedly for further smoothing -- I'm strongly tempted to try this in post-processing (i.e. offline) unless someone thinks this won't work / is a bad idea for some reason ...?

[MartyG · Accepted solution]

RealSense SDK 2.0 has a post-processing sample program to refer to.

https://github.com/IntelRealSense/librealsense/tree/master/examples/post-processing librealsense/examples/post-processing at master · IntelRealSense/librealsense · GitHub

[PTozo]

OK -- I'm getting this working but I'm finding it super annoying that it requires Imgui (which isn't included in the RealSense SDK) and isn't in the default RealSense samples along with 'capture,' 'multicam,' etc.

[PTozo]

Also, sorry to be critical, but when I see lines of code like this --

filtered = filter.filter.process(filtered);

-- I can't help thinking that better variable naming conventions might be in order.