Homepage › Forums › New feature › Suggestion for more accurate calibration
Tagged: calibration playspace
- This topic has 9 replies, 4 voices, and was last updated 4 years, 9 months ago by Greg Driver.
2018-08-08 at 20:15 #5252MotoFckr9kParticipant
The current calibration method for calibrating the vr playspace seems inaccurate.
If I hold my psmove controller in front of my oculus rift for calibration on different positions, the psmove controller will always be in front of the rift while driver4vr matches the position of the psmove controller and the rift together. This puts the kinect play area a couple of centimeters in front of the rift play area and makes their relative position to each other inaccurate if you rotate.
To counter this, the driver could ask to put the rift with the psmove at an angle to the kinect a couple of times. With that data it should be possible to calculate the position of the psmove controller in relation to the rift much more accurate than just mapping them together or adding an offset.
If it is unclear how i mean this i might be able to make a drawing of what i mean.2018-08-09 at 08:35 #5255
ok. you are saying that i match position 1 to 1 but they are actually in different positions. ps move is bit ahead of rift?
i think adding offset might be enough for one trivial reason – kinect is front facing device and everyone calibrates it when front facing.
it is different when you calibrate vive trackers with rift when both systems are 360 degree and there is really no ‘forward’ direction.
what do you think?2018-08-09 at 08:57 #5260MotoFckr9kParticipant
I guess you are right there. Putting a standard offset in and having the ability to change it might be enough. That would also make it less complicated for the end user as they can just leave the standard instead of having to do more calibration.
The question would of course be: How would we find out which offset we need? I guess if you can get coordinates of both the hmd and the controller, you could calculate their relative position using angles. The rift can even accurately show rotation so that could help too.2018-08-09 at 09:02 #5262
i am not assuming this is ps move. it can be any kind led ball or led bracelet.
i could make specific configuration for ps move but it would complicate things a bit for everyone.
i will experiment with some values and see.2018-08-09 at 11:24 #5263Damian CentkiewiczParticipant
PS Move with select hand tracking “PSMoveService” in Driver4VR, works really good. When I had tracking by Kinect Sceleton, was trouble with rotation. So why is it as not done? I think that many people can still using older metod.2018-08-09 at 11:29 #5264
for hand tracking PSMoveService act as Kinect tracking.
PSMoveService should take position directly from PSMoveService but it is not implemented yet.2018-08-12 at 02:13 #5267James NareyParticipant
I didn’t understand your last post.
What does it mean if I select the ‘PSMoveService – NOT DONE YET’ option for controllers? Which bit is not done yet? It seems from the forums that other people are using it.
Thanks.2018-08-14 at 07:17 #5269
yes. that is a bit of confustion that I am not sure how to get away with.
this is exactly same thing as ‘kinect: skeleton’.
and it was suppose to work in same way as PSMoveServiceSteamVRDriver but giving you support for up to date driver cause one with PSMoveService is a bit outdated.
when someone select this option instead give alert or not work at all it works as Kinect: Skeleton.
so yes. that is a story behind it and i will probably leave it as it is and make another option for steamvr driver for psmoveserivce.2018-08-14 at 12:35 #5270Damian CentkiewiczParticipant
But in this option rotation works perfect. In Kinect Skeleton I have a changing rotation but maybe I broken something in settings. Also setting rotation by Select on PSMove isn’t good, because it have wrong data of rotation.2018-08-15 at 06:44 #5271
are you saying about rotation so ps moves? then i need to investigate it further.
- You must be logged in to reply to this topic.