Forum Discussion
genetransfer
12 years agoExplorer
My Hydra Ik Integration Progress...
Just thought I'd show my current progress with the hydra in my engine. The Ik is performed with a ghost Ik chain for positional limits for the PhysX actors and the constraints are done with simple joints. I use dynamic bodies for the upper arm and forarm and a kinematic body for the hands. the joints go (parent child) torso-> upper arm, Upper Arm -> forearm, hand -> forearm. pretty happy with the results. will use this as a reference to control the player character bodies so as to provide world more stable world collisions. the bending over is just interpolation based on the hydra going lower than arms at rest length and twisting as arms are closer to some child null points attached to the torso.
anyway here is a short video (sorry can't embed vimeo here)...
https://vimeo.com/78870388
anyway here is a short video (sorry can't embed vimeo here)...
https://vimeo.com/78870388
15 Replies
- Felix12gHonored GuestLooks great. I'd love to see a video of the arms knocking around other physics objects when you get a chance.
- FritoExplorerWow that looks really really close to motion capture
- cyberealityGrand ChampionVery nice.
- genetransferExplorerThanks guys. I didn't expect the results to be so good just from 2 controllers, I was really surprised.
"Felix12g" wrote:
Looks great. I'd love to see a video of the arms knocking around other physics objects when you get a chance.
started working on that yesterday, so will show a video as soon as it's finished off. - AnonymousI hope that IK tech is as good as your video makes it look!
I was very concerned about the PrioVR campaign to discredit the STEM IK demo: http://www.kickstarter.com/projects/yeitechnology/priovr-get-your-ideas-moving/posts/621751 - genetransferExplorer
"Zoide" wrote:
I hope that IK tech is as good as your video makes it look!
I was very concerned about the PrioVR campaign to discredit the STEM IK demo: http://www.kickstarter.com/projects/yeitechnology/priovr-get-your-ideas-moving/posts/621751
In my integration I only use an Ik ghost chain for the position of the final effector. the actual bending is constrained by the PhysX joints. and being the hand is a kinematic actor which is governed by the hydra it gets a 1:1 look somehow to match my arms in all the posses you see in the video.
I have to try pretty hard to get a a twisted joint (usually if twisting the hydras in an unnatural way and putting them behind me) but it's easily solved by just stretching out your arms and the joint pops back into place.
So if I can do this with only 2 trackers having 5 is going to be awesome :), especially with improved latency!The main issue I'm trying to get around at the moment (being I'm using PhysX 2.8.4) is the slerp drive is a little crazy and not sure why. I want to use the bodies in the video (being they give stable response to movement) to govern a 2nd set that will be for interacting with the world. But I'm sure I'll find a way. but if not I'll just use the above rig directly as I can live with hand going through a table if I have to.Solved...will show video as soon as I add object interaction :D . - MattWalesHonored GuestThis is very cool. I'm thinking about getting the Hydra today.
- KuboldExplorerI just got it and I am sooo disappointed by Hydra. I had Playstation Move for a while on my PS3; Hydra is not even near the motion capturing quality of Move. Dealy, drift, rotation and translation error, horrible calibration...
I remember playing table tennis on Move - lighning fast full arm swings, running across the whole room, spinning the ball with a slight twist of the wrist, with a move I trained in real life - flawless motion capture, to the milimeter. Unthinkable when using Hydra.
Somebody on forum said that magnetic field based controllers are best for VR. Now that I tested Move and Hydra, I strongly disagree. Visual systems rule. I also would risk a theory, that Oculus Rift's tracking would be flawless if it used opticals rather than gyroscopes etc. And it would have positional tracking automatically. - raidho36ExplorerYou must be having something wrong - the Hydra has only 6 ms lag (delay), it doesn't drift whatsoever and calibraion can be as simple as pushing a button while holding your hands in certain position, but depending on the game, may be not required whatsoever - sensors always get absolute readings in same units, you can completely rely on them, calibration is only there to account for individual body metrics. Error is really there, and it's pretty big if you move away, but the error is mostly just jittering so you can smooth it out a lot while sacrificing a bit of latency - but the high jittering is due to extremely low EMF, they should've made it 1000x stronger (would still be less field strength than a cell phone), they're resolving it with STEM system. Optical tracking has major flaws, biggest one of them is, surprisingly, low precision and lag - the best timing you get is about 10 milliseconds with really low resolution and therefore precision, in high resolution you easily get 25 ms lag while precision still isn't the greatest, I estimate if you're 10' away from the PS Eye, you get half the precision of Hydra 3' away from the base, and this is honestly VERY generous estimation for the Move.
But you don't see none of that Hydra shit it in the Move games because it's ironed out carefully and people took serious effort to make it run smooth, because the whole game is built around it, as opposed to lunch-break Hydra integration. - genetransferExplorer
"Kubold" wrote:
I just got it and I am sooo disappointed by Hydra. I had Playstation Move for a while on my PS3; Hydra is not even near the motion capturing quality of Move. Dealy, drift, rotation and translation error, horrible calibration...
I remember playing table tennis on Move - lighning fast full arm swings, running across the whole room, spinning the ball with a slight twist of the wrist, with a move I trained in real life - flawless motion capture, to the milimeter. Unthinkable when using Hydra.
Somebody on forum said that magnetic field based controllers are best for VR. Now that I tested Move and Hydra, I strongly disagree. Visual systems rule. I also would risk a theory, that Oculus Rift's tracking would be flawless if it used opticals rather than gyroscopes etc. And it would have positional tracking automatically.
I can't speak for anything you've played as I myself haven't played any games or demos with the hydra yet (no time) other than my own and the hydra tuscuny demo. For my implementation to calibrate all you have to do is put your arms by your side and pull the left trigger (after activating calibration mode - left bumper + right bumper). and the tracking really accurate and just shy of 1:1 movement. Rather than posting another video I want to put a demo of it up so I can get feedback on what people think of it,so would be interested in see what you think when I've finished getting the base framework down. (hopefully have a simple test ready in couple of weeks, it's dx11 only so you might not be able to test it but would be interested in honest responses as to how it feel like to use, especially those who have had bad hydra experiences).
From some of the video I've seen of optic tracking, it has some big misses as well and the developers don't hide this, they are more than willing to point this out but it is a pretty good solution while we are waiting for stem and the like. stem looks like a great improvement on the hydras, and I've really enjoyed working with the hydras so stem will probably blow me away.
but I really can't comment on your experience as, like I've said, I havn't used them in anyones else game. but from my own play with them, they have been great with the rift. without the rift I wouldn't bother with them though myself.
Quick Links
- Horizon Developer Support
- Quest User Forums
- Troubleshooting Forum for problems with a game or app
- Quest Support for problems with your device
Other Meta Support
Related Content
- 6 months ago
- 10 months ago
- 6 years ago