Hello,
I am developing an application to measure certain parts of body movement while riding a bicycle. The movement of the knees is one of the most important motions we have to measure.
After some experiments with our 6 camera setup using Tracking Tools, I have some questions:
I had to use NatNet to get recorded data into our own software. Is or will there be a way to use a recorded timeline using the Tracking Tools SDK? Being able to use recorded data is very important while developing the actual software away from the camera setup.
Professional bikes often are very reflective, including parts that are moving (the crank for example). This results in a situation where it is impossible to completely get rid of unwanted markers being detected. If we design a system with active leds and disable the IR leds on the cameras, will that help a lot to get rid of these ghost markers?
The knee positions are very important (especially the movement towards the bike). I placed a single marker on each knee (at a certain point where the skin doesn't move relative to the bone) and recorded a timeline. When I play back the recording and select a knee point, it stays selected during the whole recording. That is why I thought I could easily trace that position using NatNet, in our own application. However, because there is not much to compare between frames (no marker size for example), there seems to be no easy way to trace such a point.
Is there a way to somehow identify a single marker? Does the tracking tools software does special processing itself to keep a persistent selection of a marker during the playback? Will the marker sizes of unknown markers be part of the natnet packets in the future?
It won't be easy to place an actual rigid body on the knees, because the attachment point is very small. But if there is no other way, we will try to come up with a design. Is there a certain "best practice" distance that markers of a rigid body have to be from each other?
A last question: I forgot to save the project after changing some rigid bodies, but I did save the timeline. Luckily, I could setup the rigid bodies afterwards, using the recorded data. I didn't seem to be able to set the ground plane using that data though. There is no error when I select the three markers of the ground plane and push the "set ground plane" button, but the axes are not changed. Is that normal behaviour? Is there a way to manually set the axes?
I'm sorry that I have so much questions, but I am thankful for each one that gets answered!
Greetings,
Jeroen
Tracking individual points
Re: Tracking individual points
Nevermind about the ground plane issue. Turns out there was a ghost marker that I accidentally selected in stead of a real ground marker. The axes are correctly setup now.
I also implemented a very rudimentary algorithm to trace the motion of a single marker by just searching for the nearest point next frame within a certain threshold. This is enough for my next tests, but I will expand the algorithm to search for intervals of frames where a marker is visible, and filter the ones that don't have a big enough interval. It would still help a lot if the marker sizes and/or roundiness would be exposed.
Another thing I noticed is that two-marker rigid bodies don't seem to support orientation or setting a pivot point. Is that intentional? Cause apart from a sign, the orientation could be calculated and if not, the pivot point could just indicate the reported position of the two-marker rigid body.
Greetings,
Jeroen
I also implemented a very rudimentary algorithm to trace the motion of a single marker by just searching for the nearest point next frame within a certain threshold. This is enough for my next tests, but I will expand the algorithm to search for intervals of frames where a marker is visible, and filter the ones that don't have a big enough interval. It would still help a lot if the marker sizes and/or roundiness would be exposed.
Another thing I noticed is that two-marker rigid bodies don't seem to support orientation or setting a pivot point. Is that intentional? Cause apart from a sign, the orientation could be calculated and if not, the pivot point could just indicate the reported position of the two-marker rigid body.
Greetings,
Jeroen
Re: Tracking individual points
3D marker size will be exposed via the API in the future.
Regarding 2-marker rigid bodies. We don't try to determine an orientation for it, but we do give access to the 3D coordinates of each marker so users can determine an orientation for their application based on the two marker locations.
Regarding 2-marker rigid bodies. We don't try to determine an orientation for it, but we do give access to the 3D coordinates of each marker so users can determine an orientation for their application based on the two marker locations.