search

diablodale / jit.openni

Max Jitter external for OpenNI middleware, supports Kinect, Windows and Mac OSx
GNU General Public License v3.0
43 stars 16 forks source link

Body Relative #26

Closed joncorcu closed 10 years ago

joncorcu commented 10 years ago

Hi!

I´m immerse in a project where i want to use only the hands to control certain parameters. But I need this points to be body relative, not world relative.

I´m doing some tests but i can't figure out how to do this process.

If someone has already done this work or knows how to do it, would be very helpful.

Thanks for the great job Dale!

adrianfreed commented 10 years ago

Body Relative is a tricky concept. What part of the body is your origin? What axis is priveleged for rotations? Vangelis Lympouridis has a tool that works this problem quite well for the dp.kinect.

With jit.openni I have been muddling along using the "torso" as the origin. The center of mass (/user coordinate) is rather jittery.

On Oct 17, 2013, at 11:06 AM, joncorcu notifications@github.com wrote:

Hi!

I´m immerse in a project where i want to use only the hands to control certain parameters. But I need this points to be body relative, not world relative.

I´m doing some tests but i can't figure out how to do this process.

If someone has already done this work or knows how to do it, would be very helpful.

Thanks for the great job Dale!

— Reply to this email directly or view it on GitHub.

joncorcu commented 10 years ago

Thanks for the answer adrian,

The torso is my origin. I take the difference between the head and the torso and multiplied by 3, so according to Vitrubian proportions that should be users height and the maximum distance between the 2 hands.

The next step was taking the hand position and subtract the torso, and divide this value by the height. This is not working good enough but i think i am close...

I can´t run dp.kinect beacause i'm working in mac.

adrianfreed commented 10 years ago

The head value is the middle of the head not the top so that might be an issue. Similarly the hand value is not the end of the fingertips.

Building the hand one based on height builds in an assumption that the head is visible. My dancers find lots of positions where the head is not visible but the hands and torso are.

I am using Mac also.

On Oct 17, 2013, at 11:54 AM, joncorcu notifications@github.com wrote:

Thanks for the answer adrian,

The torso is my origin. I take the difference between the head and the torso and multiplied by 3, so according to Vitrubian proportions that should be users height and the maximum distance between the 2 hands.

The next step was taking the hand position and subtract the torso, and divide this value by the height. This is not working good enough but i think i am close...

I can´t run dp.kinect beacause i'm working in mac.

— Reply to this email directly or view it on GitHub.

CarloCattano commented 8 years ago

any news about this concept? im also very interested an im so new to max , any help will be apreciated