this is a simple ik solver and server that fits a kinematic model to ik targets coming from the network.
it is meant to be a filtering bridge between kinectv2 and occulus/unity demos since the kinectv2 sdk orientations are really noisy
- Generate a skeleton definition:
$ python model.py > skeleton.jsonyou may tweak either model.py or the resulting file for your needs.
- calibrate the scene
$ python calibrate.py <ip>where <ip> is the target source server, with default port 9001.
the mapping between source targets and model anchors is performed in
target.py, so tweak it as needed.
the default skeleton definition/target mapping works for JSON targets such as:
[{"y": -0.4968303, "position": "SpineBase", "z": 1.70303965, "x": -0.0443293042}, {"y": -0.178456277, "position": "SpineMid", "z": 1.6178565, "x": -0.0205915235}, {"y": 0.130871385, "position": "Neck", "z": 1.51529324, "x": 0.00289628538}, {"y": 0.262803853, "position": "Head", "z": 1.45563316, "x": 0.0410893}, {"y": 0.0461130179, "position": "ShoulderLeft", "z": 1.51446247, "x": -0.164839}, {"y": -0.230310664, "position": "ElbowLeft", "z": 1.54731178, "x": -0.255745471}, {"y": -0.4476466, "position": "WristLeft", "z": 1.494228, "x": -0.277606815}, {"y": -0.499036223, "position": "HandLeft", "z": 1.4880296, "x": -0.2682711}, {"y": 0.0364198051, "position": "ShoulderRight", "z": 1.56479108, "x": 0.166294485}, {"y": -0.2141666, "position": "ElbowRight", "z": 1.63231707, "x": 0.193412572}, {"y": -0.448038846, "position": "WristRight", "z": 1.50331831, "x": 0.16579701}, {"y": -0.490654945, "position": "HandRight", "z": 1.50715077, "x": 0.135519311}, {"y": -0.480613261, "position": "HipLeft", "z": 1.65557861, "x": -0.127024859}, {"y": -0.67532146, "position": "KneeLeft", "z": 1.30374789, "x": -0.141852066}, {"y": -1.00822043, "position": "AnkleLeft", "z": 1.18105888, "x": -0.141313359}, {"y": -1.03534722, "position": "FootLeft", "z": 1.04560268, "x": -0.123352721}, {"y": -0.490618467, "position": "HipRight", "z": 1.67325032, "x": 0.0403243974}, {"y": -0.7525866, "position": "KneeRight", "z": 1.29755676, "x": 0.124624327}, {"y": -0.8104322, "position": "AnkleRight", "z": 1.67638087, "x": 0.009866156}, {"y": -0.789838, "position": "FootRight", "z": 1.50924647, "x": 0.0458623022}, {"y": 0.0550712757, "position": "SpineShoulder", "z": 1.54378343, "x": -0.00290112034}, {"y": -0.566544235, "position": "HandTipLeft", "z": 1.47139823, "x": -0.270116746}, {"y": -0.4969024, "position": "ThumbLeft", "z": 1.4606638, "x": -0.23709558}, {"y": -0.545468, "position": "HandTipRight", "z": 1.49029338, "x": 0.111190423}, {"y": -0.490386277, "position": "ThumbRight", "z": 1.54815781, "x": 0.1521317}](see my other repository kinect-server for the corresponding kinectv2 target source server)
- start the server
(this part is highly specific to my needs. if you need something generic please wait until I make this part customizable)
$ python server.py <ip>this will connect to a target server <ip on port 9001, and serve
joint angles for the occulus demo on port 9000 as well.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent