General setup, confusion, frustration problems

[cam@anlanda.com]

Hello I am experiencing some problems with the opti track system. I am using the 8 camera set up and have :

- followed all the tutorials
- downloaded the recent arena software update
- followed the set up instructions and measured the distance apart which is 16' x 16'
- using sync cables, have double and triple checked the in out set up etc.. USB plug in connections
- meticulously re-enacted the calibration video tutorial you have online. (thanks for that)
- experimented with different exposure settings.
- meticulous set up cameras pointing at waist.
- tried this 20-25 times now in various conditions ie. at night. during the day, blocking off natural light UV's , cleaning the space of any objects, neutralizing the clutter with black plastic, tried a 6 camera and 8 camera configuration etc.

Here are pictures of the space:
http://www.anlandastudio.com/opti/116.JPG
http://www.anlandastudio.com/opti/117.JPG
http://www.anlandastudio.com/opti/118.JPG
http://www.anlandastudio.com/opti/120.JPG


PROBLEM

I have yet to get a decent calibration result (only one "excellent"ever) and very rarely above poor, fair and occasionally good. As a result am not able to record data without it getting spastic and flashing markers . (I want to use this on a larger project but I just can't go back and clean huge amounts of data. Here is a video of this in action, I recorded my screen to show you what I mean http://www.anlandastudio.com/opti/opti.html Note how it flashes and spaz's out...

It seems that there is some interference with the camera's or the exposure is wrong?

One thing I noticed is I can not get the frames to be free of dots at set up especially with the horizontal cameras in the 8 camera set up. This is what I see in those 4 cameras initially when I try to calibrate it. http://www.anlandastudio.com/opti/110.jpg I of course use block all visible dots but it makes me wonder if the cameras are not working properly etc..? Because why would I only see that in the horizontal ones.

things I am wondering...

- Would the cameras interfere with each other ? (highly doubt this but there is nothing else in the space..)
- what is with the horizontal cameras having dots in them when the vertical ones do not?
- do the cameras need to be 20' away to work properly? ( the usb cables you sell and sync cables are only 16' ? Whats up with that BTW ? )
- is the black plastic I am using to clean up the clutter in the space interfering with glares? ( I set that up because the space is filled with clutter and in your manual is says it is preferred to have a neutral environment)(if I point the cameras right at the plastic I don't see anything)
- Could there be something ie nails in the floor or electrical lines or something that I can't visibly see that is interfering?
- Poltergeists ?


My planned next step

I am going to try this set up in a completely new space .. one that is hopefully less cluttered to start with and has concrete floors.. I will again block out natural light. ( How did you ever get this working in that convention center set up as in the videos? )


ADVISE ?

If you have any thoughts on all this I am all ears.. ??

[VincentG]

OK, I have a few tips, to start off with.........

1 - The tension on the cables, will cause the cameras to be under stress, and can move very easily, which will throw off the calibration. The cable lengths, in regards to USB, are defined by USB standards. USB defines USb as maximium cable length of 5 meters, between device, and hub/repeater. As RCA cables go, you can use standard RCA cables, and use a RCA cable connector to get longer cable length.

2 - It looks like you have the upper cameras at 8 feet, and the lower cameras at about 4' high, and pointing directly across the room. The extra images that you are detecting on the lower cameras are coming from the other cameras. As they are all generating IR, they can be seen, and cause interference with each other. I would mount the lower cameras at about 5-5 1/2', pointing slightly dowwn towards the center point of the tracking area.

3 - Looking at the screen video capture, it is a little hard to tell what is happening here. Part of it canbe the calibration, as well as marker placement. If you could send me copies of your calibration file, skeleton, and sample data, we can narrow down what the Arena software sees.

[Birch]

A couple starting points :

* Strain relief : It does not look like the cables are strain reliefed. It is very important to strain relief them to prevent damage to the cameras, as well as to prevent the cameras from pulling on each other and moving. The weight of the cables hanging can pull on the tripods unless they are very solid ones, this can also cause the cameras to move. If the cameras move at all, it will degrade the quality of the tracking during and after the calibration.

* Camera aiming : How are the cameras being aimed? We recommend placing one or two markers at the center of the volume around waist or chest height, and using those to align the cameras to point toward the same area.

* Camera settings : The defaults should be sufficient for your environment (threshold = 150, exposure = 55, intensity = 15). If needed the exposure could be lowered a little, but the other settings should not need to be changed under normal use.

* Marker blocking : Is marker blocking being performed? (It should be) Can you post before and after screenshots of the marker blocking process?

* Wanding data : The 1.1 version of Arena allows the wanding data to be saved, please send a couple samples to support@naturalpoint.com

[VincentG]

When sending the sample data items, it can be easier to troubleshooting if you can start each session from a full t-pose

[leith]

the stand off distance you have there may be a bit too large. Its difficult to say without having the capture files to look at. But you may be reaching the threshold of doughnut with this one, where the cameras on the far side have issues seeing across.

the dots you are seeing are likely the cameras seeing the LED ring of the other cameras I would guess. And potentially the reflection of an LED ring in the black plastic or even the floor if its polished enough. This should not be a problem however unless you're actually flaring the lense. You should be able to get rid of the dots in calibration using the blocking features of arena.

Putting the cameras in grayscale mode can help see what is going on. I'd give it a go. But note that when the camera goes into grayscale, its exposure changes due to a framerate change. So it will be brighter than it is at full framerate. but it will show you what in the room is bright at the very least.

[Jim]

Okay, my two cents:

- For 8 cameras, they are probably placed to far apart, as Brad suggests. I would close up the square you have created by 2-4 feet. Make it around 14' between the tripods.

- As mentioned, strain releif the cables. You are using our inexpensive light stands and they will move A LOT with just the force of the cables. They will be okay on the floor you have if they are not touched at all, including cables.

- Get some USB active Extension cables. These will give you 15' more for the cable run and will help out a lot.

- You can extend the RCA Sync Cable yourself, just get some parts at RadioShack, or you can get more cables from us, and some joiners at RadioShack.

- The lower camera should be moved up to about 5'.

- The overall camera placement is good, it looks like they are pointing in the correct direction and position.

- The black plastic scares me. That stuff could reflect and shimmer, I would not use it, I would use black cloth. We have even used black bed sheets, they work well and are cheap.

I don't think you are that far away from success, just a few things here holding you up.

[cam@anlanda.com]

Thanks for the prompt response guys.. I'm a believer and can almost taste it.. just kind of driving me crazy for now.

Ok I will go try it with the cameras in closer and raise the horizontal cameras angling down slightly. I will also send any screen captures and wand data if needed.

By marker blocking I assume you mean the "block visible markers" button during calibration. And yes I have been doing that.

I had taped the tripods down and the cables as they are tight. Ill rig up something to take the strain off. I thought the idea would be to create a bigger volume but sounds like I should be going the other way...

[cam@anlanda.com]

Ok I tried it again 7 times... same results after moving the cameras in closer and raising the horizontal cameras angling down slightly.

I also watched the calibration video and repeated as close as I could. I am sending wand and calibration data and screen captures here I numbered each try sequentially. http://www.anlandastudio.com/downloadO.html

BTW next week I will be in a different space so can compare..

thanks

[Jim]

Hello:

We really appricate your detailed posts and the file links, it really helps.

Yes, I think you are close too, hope we can make this happen soon.

A caution to "wanding". Our video is just one technique, there are many and the are quite varried. The technique in the video doesn't work well for me I am usually on by back, wand in the air, and try to remain not in the cameras field of view. I also move the wand slower, but everyone develops a technique that works for them. So, if all other recommendations are followed, try a different wanding technique. Your goal is to get as many cameras to see the wand marker at the same time and cover the whole image area, without going out of the cameras overall field of view. You know, easy.

[seanboyette]

Camc,
Something else to think about - I had a 8 cam setup before and the high and low camera location on each tripod never really worked as well for me. My best results are when I had all cameras at a high vantage point, in fact I have a 12 cam setup now, and all cams are nearly 12' high and I get great results.

having them high - will allow you to have them closer together and not have the stress on the cables as well!

I would also remove the plastic - shiny = not good, if the cams are pointed to it.
pointing the cams down (using the grayscale to ensure that you get them pointed to the general capture volume, not the external stuff) will help with the clutter as well.
I can take a screenshot of my grayscale and you can see generally where I have my cams pointed.

Calibration is all a technique, the one that works best for you! I have a video where I show my technique on my site.

Sean