[ANSWERED - TICKET] OpenNI & Orbbec crashes

Woland

You likely need to do this. Unlike the video, you will likely want to disable the wifi.

In the "Depth Camera Listener" User Actor I made for you, you might also need to go to the 'ndi source' input and re-select your NDI source.

Best wishes,

Woland

Vidasonik

@woland I don't have ethernet in my list when I open Network folder in system preferences? I've no idea what to do having spent over £100 on connectors, adaptors and the Netgear Nighthawk!

Woland

@vidasonik

I sent you a message through the ticket system. I may be able to help.

Vidasonik

Finally, I have solved the problem of depth camera not working on my new mac and having to send data to a second computer!

I have dispensed with the Netgear Nighthawk (thank god!) and not needing to connect my two macs at all! I simply connect the orbbec and webcam to my old mac and have a duplicate ableton project on the old one to trigger effects etc. The sound is only sent to Izzy.

The audio on the new mac plays out to pa with live guitar etc. I don't need to send sound to Izzy at the same time because now Izzy is controlled by the old mac. The only issue is the live guitar cannot be used to trigger effects on the old mac (which has the main izzy file which outputs to main projector. The new mac out puts a simple (non-interactive) izzy file to a separate projector.

All I have to do is press the same key on both macs at the same time to trigger the start of the show and proceed through all the scenes together. They are then both running more or less in sync!

It's taken me about 3 months to reach this position!

I have one question though, I can't understand why the orbbec orientation is the opposite of the webcam. That is, the dancers right hand is to screen right with the depth camera (correct) but to screen left with the webcam??

Best wishes, David

Woland

@vidasonik said:

I have one question though, I can't understand why the orbbec orientation is the opposite of the webcam. That is, the dancers right hand is to screen right with the depth camera (correct) but to screen left with the webcam??

Maybe you enabled the 'mirror' input on the OpenNI Tracker actor to flip the depth image?

Vidasonik

@woland Actually no I hadn't. But now I have mirror selected it matches the webcam !!

Another question, I am using an outline of the body tracked by the depth camera with DIFFERENCE actor but when I select COLOUR I only get a red outlined figure despite changing the hue offset with the HSL adjust actor?

Woland

@vidasonik said:

Another question, I am using an outline of the body tracked by the depth camera with DIFFERENCE actor but when I select COLOUR I only get a red outlined figure despite changing the hue offset with the HSL adjust actor?

Try this instead

>>>Example File<<<

Vidasonik

@woland wow - thanks so much! Looks good - Can't wait to try

Vidasonik

@woland For some reason I can't connect to the colour video node (it is faded in grey and says n/a) - only the depth video is accessible?

Woland

@vidasonik said:

I was using a Kinect. I think Orbecc would use the Video In Watcher and Live Capture for RGB video.

Also, I misunderstood. I thought you were trying to use the color video output of the OpenNI Tracker actor.

You can just disconnect the link from 'color video' and link 'depth video' to the difference actor instead.

Vidasonik

@woland - thanks again! I've used your example and changed my User Actor accordingly. Weirdly though, I get colour only when I set DIFFERENCE to GREY! When DIFFERENCE actor is set to COLOUR I just get RED again, as before. But thanks I've got there in the end. And I love the blurr effect ...

Vidasonik

@woland Hi! I'm nearly there with the depth camera now, thanks to all your help. Just fine tuning. One issue that is puzzling me is the distance between the tracked hand and the start of the coloured line. It would be great if the line emanated from the hand immediately rather than there being a delay or gap. I've checked all the settings and can't find where this distance might be adjusted? No big deal anyway. I attach the user actor patch in case you can see anywhere I may have overlooked. Thanks as ever, David

Woland

@vidasonik

It's likely the difference in aspect ratios and resolutions, at least, that's what I've had to solve before in similar situations. The depth camera is a 4:3 aspect ratio and a different resolution than the output of the Live Draw actor (which is 16:9 and uses the default video resolution set in the video tab of Isadora Preferences unless the Live Drawing actor is given a specific resolution using its 'horz/vert size' inputs.)

Using a Matte++ actor with the 'scale fg' input set to 'off', and feeding it a background the size of the stage (i.e. the output of the Live Drawing actor scaled up to the stage size) I think can reconcile the resolution difference by putting them both in the same 16:9 frame, but the openNI depth video will still be in 4:3 so it doesn't solve the aspect ratio difference.

Even after doing that you may need to scale the values sent to the Live Drawing actor by a certain amount to get the drawing point to line up with the tracked body part. Below is a simple representation of the problem. The depth camera's resolution is represented by the red box and the Live Drawing actor's output is represented by the green rectangle (which goes behind the red square). As you can see, the camera will only ever see you moving within the red box, but the Live Drawing actor can draw all the way out into the green areas.

You could solve this by zooming in on the video of the openNI depth output so that it fills the screen, but you will lose some of the depth image as a result if you do that.

You could also solve it by scaling the values so that the live drawing is limited to the area of the red box.

A third way is not to solve it and just to accept the fact that the live drawing will be further off the body part location the further from the center of the camera's vision you move.

There could be other factors and solutions that I am not remembering right now but hopefully the information I provided is useful.

Vidasonik

@woland Thanks so much! Yes this all makes total sense. I may look into these solutions or just put up with it as you say. As my PhD is around temporality it may even be good that there is a kind of delay or gap between the hand and the onset of the line