Some racket games waiting for the STEM (Hydra too limited)

[drifter]

Hey rifters !
Just to confirm a few things about the Hydra...

As a bit expected, the Hydra shows its limits with this kind of game (racket games):
- The in-game calibration can't be robust enough because of the non-linear tracking
Depending on where I do my calibration, I can have an error up to 15-20 cm ! (Where I need a margin of a few cm max)
- For the same reason, the tinkered head positional tracking using the spare Hydra is not really satisfactory
It needs to be very precise and must feel natural for this type of game.
Also I'm at the limit of the effective range of the Hydra, if I move back a step, my head start to jitter (not pleasant !)
- There is some interference between the two controllers
At a certain distance (up to half of the range), my right Hydra passing between the base and my head will interfere with my head tracking (small lateral movements)

As a bonus I would like to get rid of these 2 annoying wires (the Rift wire is less a trouble).
Also the lack of haptics is a problem if later I want to aim for a true sim (it's even more important for a tennis sim).

So... vivement 2014 !

Here's a video showing that when the calibration is not too badly done, it's possible with the Hydra to have quite a good control on the tiny ball (power, direction & effects).(Effects are disabled in the video, still working on it).

(don't watch it in your Rift (or only on pause), head positional tracking + fast head movements + low framerate = vomitland)

[youtube-hd]http://www.youtube.com/edit?video_id=_4 ... watch&ns=1[/youtube-hd]

I still have a tennis game project going in parallel (must make a video one of these days).
The lack of accuracy is a bit less a problem (the racket head being much larger), but it's not so joyful either...

So since a few months my projects with the hydra are going a bit slowly (I have another project - not involving the Hydra - to keep me busy), but I will continue to work on it from time to time, I know it's not good to have several projects at the same time, but I like to vary the pleasures and it's nice to work sometimes on something completely nausea-free.

++

[cybereality]

Cool.

[Fredz]

Nice that you did continue working on that, the result looks pretty good, can't wait to see what you've done with your tennis game.

Also nice that you could confirm (although quite disappointing and somehwat expected) that the Razer Hydra was not accurate enough for such tasks, that's what held me off backing the STEM project, since they didn't really talk or give hard numbers about accuracy.

Did you try using PS Move controllers combined with a PS Eye for head and hand tracking in your games ? The vibration-based haptic technology could also give some sort of feedback.

[drifter]

Cool.

Heya, monsieur cyber !

Nice that you did continue working on that, the result looks pretty good, can't wait to see what you've done with your tennis game.

Merci m'sieur Fredz

Also nice that you could confirm (although quite disappointing and somehwat expected) that the Razer Hydra was not accurate enough for such tasks

yeah I knew from the start that these projects were very hazardous : some games working only with an HMD which still is not a reality (even if we are all strong believers here), and playable only with a more or less accurate motion controller (the successor was even not announced yet).
Let's say they are long-term projects, one day the hardware will be good enough...
Now to cover my butt I'm on a more short-term project since I got my Rift in june, it's the complete opposite, it's even playable without Rift (but of course it will be much better in VR )

that's what held me off backing the STEM project, since they didn't really talk or give hard numbers about accuracy.

You are right (dunno if the robustness of the tracking is quantifiable ?).
I myself backed it because I'm an optimistic dude (also the guys from Tactical Haptics found the STEM more robust, well we'll see...)

Did you try using PS Move controllers combined with a PS Eye for head and hand tracking in your games ? The vibration-based haptic technology could also give some sort of feedback.

Didn't try, the lack of absolute positional tracking wouldn't be a problem ?

Edit: wow I just saw you already updated your wiki, amazing

[Fredz]

You we right (dunno if the robustness of the tracking is quantifiable ?).

Actually it's often advertized for high-end magnetic tracking devices, that's why I'm not convinced that the consumer-market ones are as rigorously tested.

Intersense IS-900 :
- precision : 0.75mm 0.05°
- accuracy : 3mm 0.5° RMS
- range : 9m²
- frequency : 180 Hz
- latency : 4 ms

Polhemus Patriot :
- precision : 0.00046" 0.00381° at 12" range
- accuracy : 0.06" 0.40° RMS (measured at 0.1" 0.75° RMS by an independent study)
- range : 1.52 m
- frequency : 60 Hz
- latency : 18.5 ms

Didn't try, the lack of absolute positional tracking wouldn't be a problem ?

It DOES support absolute positional tracking thanks to the glowing orb tracking with the PS Eye, along with the 3dof rotational tracking from the MARG in the PS Move.

You can also optically track up to 5 controllers with the PS Move API which should be more than enough for head, torso and hand tracking. And in this kind of game I think line of sight shouldn't be a problem.

[cybereality]

The Hydra has distortion issues, but they are repeatable. In theory you could calibrate for the distortion and correct for it.

[zalo]

Allegedly all the old CAVEs used electromagnetic tracking in the beginning, but later switched to optical because they could skip the calibration step (and building the frames out of wood).

[geekmaster]

All you need to do to calibrate is dedicate a hydra button to "trim" where when pressed your VR arms do not move until you release the button (presumable with your real arms coinciding with your VR arms). The software would keep track of all the calibration points where you released the "trim" button, and interpolate between them. More intuitive in-game than a separate calibrate mode...

Likewise, you could have "trim" controls (much like on R/C transmitters) to compensate for gyro drift. You could tweak it a little to miminize drift when needed. The automatic drift compensation in drivers does not always work in all cases (as you can read in various forum posts), so manual adjustments would help in those cases.

Also, even when magnetometer values are not calibrated, they should be stable enough to be used as arbitrary "landmark" reference points to prevent drift.

Anyway, compensating for calibration problems should be a simple and intuitive manual procedure to be done naturally at the touch of a button (like ratcheting in some games).

[drifter]

Sorry if I'm a bit late.

@Fredz
Thanks for the info, I will take a closer look at the Playstation stuff (true I was more focused on PC devices).
Line of sight should be ok, yep, at least it should see the collision (and the frame before ^^).

@cybereality & geekmaster
Oh, yes now I remember some discussions about a calibration map
http://www.mtbs3d.com/phpBB/viewtopic.p ... on#p135172
Ok I will measure many more points, specially the points far from the (small) sweet spot. Something like this :
- Calibration sequence : head, torso and then (stretched arm) forward, left, upward and right (the forward point staying the best input to get the length for the real arm).
- optionally, if a more complex calibration is needed (or wanted), saving calibration data to disk (several slots if you need to move and/or rotate the base station)
- then performing IRT linear interpolation between the two 3D matrices to get the correction vector
I am right ? Hrm... Brain already emits smoke.

[Fredz]

If you implement positional tracking with the PS Eye/PS Move you could even calibrate the Razer Hydra using this.