TECNICALL SUPPORT!!!!!!!!

[Alessia8086]

We are trying to understand what actually we're doing wrong couse we hear about people recording facial markers with no problems and we cant get a decent result with 12 cameras so im pretty sure we're missing something, there is any example..

http://www.inximagine.it/errors/error/error.html

this is done with 37+4, 10 markers and last build of arena 1.7.2 as you can see markers go everywhere and even facing the cameras..

and this is done with a custom set of 53 markers positioned in specific places, and its pointless to say that its totally unusable.. dont know what to do..

http://www.inximagine.it/errors/custom_ ... error.html

[Alessia8086]

no one of the tech support cant answer?

[NaturalPoint - Mike]

Alessia -

We do want to be able to help you with this problem, however please keep in mind that you are 10 hours ahead of us. We're not trying to ignore you, however our office hours combined with the time differential make for a difficult real time support scenario.

In response to your post yesterday, I have reviewed the videos you posted as well as the files you sent. It appears that you are still having calibration issues, possibly compounded by lighting issues (not ambient, camera provided). I emailed specific suggestions to making these adjustments that I believe could help. In general, precision grayscale mode should be used, and ir/exposure settings should be experimented with during the camera preview section of the calibration process before wanding to ensure the best possible results. Also, it would do well to ensure that you provide enough samples to get a very high quality calibration, then during solving, set the quality to very high and let it calculate until everything is exceptional, and then apply. I'd encourage you to full screen the calibration wizard and make a video of your calibration start to finish like you did with the previous two examples in here.

Also, it is worth noting that camera 8 had consistantly bad quality, almost like it shifted post calibration.

[Alessia8086]

tested today with continuos instead of strobe and we couldnt even make the calibration succesfully, results where coming out allways not more then "poor" even after 40 min of calculation...

[NaturalPoint - Mike]

Alessia -

I have sent an email to you and your associates with suggestions as to alternate methods of support for this issue.

[bsteagal]

Alessia

This might be a little off the wall, but what are the specs of the system you are using to run the capture? You are making sure that you are running each usb off a different USB controller?

The reason I ask, is that I've had situations similar to this, where tracking is off and you have strange problems even when calibration is good, by using pcs that just didn't have the horsepower to keep up. You also get strange issues when using USB ports on the same controller.

Just a thought

[Alessia8086]

We're running a 12 core machines with 24 gb of ram a quadr0 4000 windows 7 pro 64 bit with 10 usb 3 and 10 usb 2 ports.
The camera are mounted on 2 different hub ( 6x2 ) on a machine dedicated just for motion capture and no other software installed except arena and motion builder.
Dont know maybe a better machine can be found at Nasa for working with mocap hehe
Jusst Kidding but well at that point we really dont know what to thing, monday we should had the final tests with natural point as soon as we got some new stuff then if it doesnt work we sell the system couse honestly we have spend 8000 dollars extra plus the price of the mocap system just to start to find some compromised but its too much work for us...
lets see...

[bsteagal]

Alessia

You are right there! now I have pc envy... lol

On the pc side though, are you making sure to connect the 2 usb hubs to different controllers on that PC? (for example, if the pc has 4 ports on the back of the pc, and 2 on the front, make sure you connect one in back, one in front)

Although I don't believe that would have much to do with it, you never know.. it is better to eliminate everything.

I run facial captures on the original v100 cameras with 8 cameras in a setup similar to yours, with the standard # of markers with no problems. There has to be something that is being overlooked. I use the 3mm markers instead of the 4mm (the bigger ones caused swapping issues)

A good test would be if someone else can use your 50+ marker setup and see what results they get...

[Alessia8086]

Hi Bryan,
yes i guess that standard setup is not giving any problems, even in our case its not perfect but works, but 37 markers are way not enough for an accurate facial animation as the one we are trying to achive and based on our experience, my first tough is that a custom configuration with 53 markers is impossible to achive at the moment with arena even with 12 cameras, and i guess this is not gonna change drastically if not adding at least other 12 cameras that i think would be really pointless, so now we are trying to test every single possible configuration to understand if this is the system we are gonna use for our next 3 feature or if its a waste of time, sorry to natural point to be direct but for 15k i expect a bit more then an amatourish configuration to work with.

[bsteagal]

Hi Alessia

You have to consider, your 53 marker configuration is not a common one. Even manufacturers like Phasespace and Vicon use a similar configuration to the one Naturalpoint is using.. 37 markers is standard and normally more than enough for facial animation.

And how much will you end up paying for Vicon or Phasespace? how about 60k or more to start.. just a thought..