28 Jan 2023 |
Monky Games#8592 | FYI - here is a video of the tracking working for Nolo !!! - https://youtu.be/L1-a-rHT_V4 | 05:49:13 |
haagch#5957 | presumably the rotation will drift over time as is | 07:32:33 |
haagch#5957 | the way to solve it is to write code that filters the imu and the pose together such that the movement directions detected by the accelerometer are lined up with the movement direction of the positional tracking | 07:33:48 |
thaytan#5292 | A ukf will infer the correct yaw | 08:41:09 |
| playback2396 joined the room. | 09:59:57 |
| Ornst3in joined the room. | 11:12:02 |
Monky Games#8592 | Where do you think that code should be written? OpenHMD or Monado (fusion.c)? | 20:37:14 |
thaytan#5292 | OpenHMD should do it | 20:46:35 |
haagch#5957 | openhmd doesn't give us acceleration | 20:47:06 |
thaytan#5292 | it does in my branch 😉 | 20:47:29 |
thaytan#5292 | but that's global accel, not IMU | 20:47:46 |
thaytan#5292 | It still should be OpenHMD doing the fusion | 20:51:21 |
Monky Games#8592 | Doesn't the nolo_decode_orientation in packet.c provide acceleration and gyro force? | 21:20:49 |
Monky Games#8592 | So nolo has an IMU that merges accel and gyro to provide rotation about the axises? | 21:22:16 |
haagch#5957 | imu is the part that contains the gyro and accelerometer | 21:33:37 |
haagch#5957 | openhmd reads the info but it doesn't have an api to let monado ask for the acceleration directly | 21:34:18 |
haagch#5957 | so unless we extend the openhmd api, it's only going to be possible inside openhmd | 21:34:23 |
Jakob Bornecrantz#6482 | It would be also pretty easy to port the nolo driver to Monado so you can have better integration there. | 21:34:26 |
| ᲻᲻ joined the room. | 21:44:12 |
| Supreme#9828 joined the room. | 22:13:07 |
Supreme#9828 | is openhmd still getting updates? | 22:13:08 |
Jakob Bornecrantz#6482 | ![image.png](https://matrix.org/_matrix/media/r0/thumbnail/t2bot.io/e9142947fb9dc0ac7313c24b3e14467077dac3eb?height=360&method=scale&width=360) Download image.png | 22:14:08 |
Jakob Bornecrantz#6482 | Technically yes... | 22:14:11 |
Supreme#9828 | 😛 | 22:21:14 |
pblack#8962 | for context - theres some.. whats the phrase.. 'divisive politics' around openhmd and monado but there is no question that monado is where the vast majority of the work on open source xr has been done in the last 2 years. and openhmds future as a standalone project is unclear, unless somebody steps up to really lead the project and bring some major improvements to the way it works its unclear how it can move forward, vs stagnate at best | 22:21:39 |
thaytan#5292 | If I did that, I would end up completely changing the API it exposes | 22:55:27 |
thaytan#5292 | I think it's so dumb that it plumbs everything through "an array of floats you'd better know how to interpret" | 22:55:48 |
| rpavlik#1182 joined the room. | 22:57:00 |
rpavlik#1182 | Hey don't did diss arrays of floats, vrpn has done that too for literally 25 years successfully... (But it does have tracker report types too, in addition to float arrays) | 22:57:00 |
rpavlik#1182 | 😉 | 22:57:17 |