Update Kinect 2 Input to Azure input and Add features

[strangerattractor]

Update Kinect 2 Plugin to Azure and take https://github.com/keijiro/Akvfx as a guide.

- [ ] depth texture should be available to use (either in VFX or Shadergraph, or else...) like in akvfx. It becomes a usable texture in the /buffer folder, if I'm not incorrect. This is very convenient to use. -> #129

- [ ] depth threshhold should be adjustable in the settings like in akvfx (this is very handy for a lot of reasons) -> #129

- [ ] make different depth camera modes switchable if possible: ->#138

- [ ] if possible, check if a "cleaned" depth image is necessary (to fill in for undetected depth pixels) -> requires elaboration

- [ ] color output texture should be available to use like in akvfx. -> #129

- [ ] Update Joint detection to Azure Kinect like before. -> #137

- [ ] Update motion speed detection like before. -> #136

- [ ] Best case scenario is, that depth texture is ready to use as "position" in shader graph and "position" in vfx graph. -> #135

[chikashimiyama]

hmm... the requirement is more like an adaptation of Akvfx.

[chikashimiyama]

The task is too big to tackle. I will separate tasks into several chunks.

[strangerattractor]

ok, I also wrote the requirements after looking into the akvfx functionalities and things that I found handy to have and to use. so we can still cut stuff. as always

[chikashimiyama]

If you want to make a big change in software. It's better to granulate tasks and define many small tasks so that we can give up a feature or prioritize a certain feature easily.

[chikashimiyama]

made #129 as the first contact with Kinect Azure.

[chikashimiyama]

• Keijiro uses k4a.dll which is a C# wrapper of KinectAzure SDK.
• /Internal/ classes use a lot of unmanaged memory tricks which might be very experimental.

[strangerattractor]

I also saw a one point in the code he actually says: interface is experimental and will break in the future

[chikashimiyama]

https://github.com/microsoft/Azure-Kinect-Sensor-SDK is C and C++ project. https://microsoft.github.io/Azure-Kinect-Sensor-SDK/master/index.html

C# wrapper is available but there is no documentation of the API even in v1.4.0 alpha. The azure project is still very immature.

[chikashimiyama]

moved three ACs to #129

[chikashimiyama]

if possible, check if a "cleaned" depth image is necessary (to fill in for undetected dpeth pixels)

@strangerattractor

Can you elaborate on this AC?

[chikashimiyama]

make different depth camera modes switchable if possible:

Is this also a feature of Keijiro's demo?

[chikashimiyama]

Update motion speed detection like before.

I'm not sure what you meant by this. Did I implement this? if so, where can I find that? I interpret this as the "difference between previous and current frame" #136

[strangerattractor]

if possible, check if a "cleaned" depth image is necessary (to fill in for undetected dpeth pixels)

@strangerattractor

Can you elaborate on this AC?

I checked with the new kinect. And I think I can work without additional cleaning. I had been looking at stuff like this: https://blogs.wcode.org/2017/01/howto-use-foreground-removal-with-the-microsoft-kinect/

I think I'm fine now with this topic

[strangerattractor]

Update motion speed detection like before.

I'm not sure what you meant by this. Did I implement this? if so, where can I find that? I interpret this as the "difference between previous and current frame" #136

exactly. sorry for the not being clearer here