Camera placement

[Amos Meyers]

My laboratory is involved in a study of kinematics/kinetics of Olympic weightlifting and we're having a heck of a time positioning our cameras (8-camera BTS Smart-D) to pick up all markers throughout the full range of motion during snatches and c+j's. Our two biggest issues are getting at least two cameras to pick up the ASIS markers at the bottom of the squat and the shoulder markers (prominent point of shoulder on top of acromion, since the barbell would hit a marker at the AC joint, our usual landmark) at the top of the press.

I think our best option is a "double diamond" configuration. If you picture it like a compass, with the subject facing north, we would have cameras placed high (~8ft) at N, S, E, W, and cameras low (~4ft) at NW, NE, SW, SE. Before we take our cameras off their permanent mounts, does anyone have experience with MoCap and a motion like squats (where the ASIS, if you mark it, is likely to be obscured). How did you deal with it?

TL;DR - How to place cameras to best pick up awkwardly placed markers?

Cheers

~ACM

[Alec Black]

Re: Camera placement

I would think about marker placement rather than camera placement. For our lab we don't ever change the camera positions, but we do think carefully about marker placements. We would likely place markers on the ASIS's, PSIS's and maybe the iliac crests as well. We would then create virtual marker positions of the ASIS markers from the other markers (PSIS and iliac crests) and use those when we lost the ASIS markers. Same would hold true for other "difficult" markers.

Good Luck
Alec

[Dustin Bruening]

Re: Camera placement

I am not familiar with the BTS cameras, but Alec is right in that you can certainly play around with your camera configuration, but you may never see an ASIS marker that is occluded during a deep squat, or an AC marker with a bar on top of it (not to mention that it is likely going to get knocked off). You are better off tracking those points using additional markers. For a deep squat, a marker triad on the sacrum may be your best bet for pelvis tracking (or PSIS and Iliac crests as mentioned). AC markers or shoulder joint centers can be tracked in a lot of ways depending on model detail - if you need good detail, you could try 2 markers on the clavicle and one on the posterior acromion, which should give you sufficient spacing and stability.

-Dustin

[yajur]

Re: Camera placement

Thanks for your nice information.

[James]

Re: Camera placement

I am not familiar with the BTS cameras, but Alec is right in that you can certainly play around with your camera configuration, but you may never see an ASIS marker that is occluded during a deep squat, or an AC marker with a bar on top of it (not to mention that it is likely going to get knocked off). You are better off tracking those points using additional markers. For a deep squat, a marker triad on the sacrum may be your best bet for pelvis tracking (or PSIS and Iliac crests as mentioned). AC markers or shoulder joint centers can be tracked in a lot of ways depending on model detail - if you need good detail, you could try 2 markers on the clavicle and one on the posterior acromion, which should give you sufficient spacing and stability.

-Dustin

Thank you for sharing these wonderful tips.

[Edmund Cramp]

Re: Camera placement

How to place cameras to best pick up awkwardly placed markers?

Assuming that they are available, the first step is to look at the raw 2D data files from each camera to verify that the problem is caused by the camera failing to see the marker - not a problem with the markers being seen but not being reconstructed into 3D points. This also checks that the cameras are actually working - and will tell you which cameras see the markers, and which cameras do not.

Once you've verified that the problem is that the cameras are working but are not seeing the markers, walk around to each of the cameras, place your head near to the lens and simply watch the subject move through the trial. When you find that you can't see the marker(s) during the subject motion, have the subject repeat the motion and see if you can move yourself to a position where the markers can be seen.

If you find positions in the lab where you can see the markers then consider moving the camera to that position and viewing angle - remember that a minimum of two cameras must be able to see a marker at the same to to reconstruct a 3D point.

Of course, when you move a camera to a new location, you'll lose the 2D data that the camera was contributing from the old location ... and this may make the overall data quality worse. In this case you can either change the marker model or purchase additional cameras specifically to capture the problem points - but remember, if you can't see the markers when the subject squats; then no amount of additional cameras is going to help.

[Paulo Lucareli]

Re: Camera placement

Dear ACM

I'm using a cluster with 3 markers on PSIS and during static trial a pointer to create virtual LASI and RASI. I have this processor in case you are using Smart Analyser.
If the changrs in cameras position doesn't work you should try the same strategy to acromio markers.

Best

Paulo

[Amos Meyers]

Re: Camera placement

I'm using a cluster with 3 markers on PSIS and during static trial a pointer to create virtual LASI and RASI. I have this processor in case you are using Smart Analyser.

Prof. Lucarelli -

We are using a smart analyzer. I sent you an email requesting your protocol, but never received a response. Could you get in touch with me?

Thanks,

~ACM

[Paulo Lucareli]

Re: Camera placement

I think your mail going to my trash past.

Would you mind send me another email to paulolucareli@uninove.br or paulolucareli@me.com?

My pleasure to help you.

Paulo

Movement Analysis Laboratory
Hospital Israelita Albert Einstein
Universidade Nove de Julho
Sao Paulo - Brazil