We also noticed that the distortion shader definitely kills the fps…
Hoping that WEBVR will be better optimized in the future!

ok, we tried it and the aspect ratio is now fixed!
but a problem is still there because our app can't use correctly the hmd parameters.
When we set in the hmd_conf window the cardboard 2015 profile [0.34, 0.55] and then we put the correct screen size, then save and reload, the view is still too much distorted.
If we set the distortion to [0,0] (using hmd_conf window) the view is really better and there is only a small distortion in the corners

You can try with this temporary link
https://hpc-forge.cineca.it/files/visit_3Dmodels_VR/public/terraemotus_1609/terremotus/terremotus_hmd.html

Any thoughts?
Luigi

wow, ok! ty! so.. is the bug already fixed in the 16.09 early developer preview?

Hi guys!
We're creating a VR version of the Terraemotus (Reggia di Caserta) project.
We're experiencing a strange behaviour in the 2015 cardboards: we have an excessive distortion of the objects in the scene (i.e. pictures on the wall have wrong aspect ratio) and the depth of the whole scene appears very flattened.
We're using the hmd_conf window in the app to set the correct parameters of carboard and phone screen, but we have not been able to fix this issue (at the moment chrome browser mobile is our testbed).
Have you got any hint for us to force the app behaviour for cardboard 2015?
TY very much!

Hi guys,
I created the italian B4W wikipedia page and I added a description of B4W in the italian Web3D page too.
I hope these pages can increase the knowledge of B4W in Italy.
Keep up the good work!

happy to ear that you are considering the issue! The secondary camera would do the trick I hope.. That's a cool workaround, It could be useful also for other things!

Thanks!

Hi!
We developed a simple scene and we ran it into hmd cardboard environment with the configuration, the new hmd_conf allows us to set webVR preset easily and the out of the box working gyro is awesome! so far so good!

We noticed that it's not possible to use the pick_object method in such configuration, probably that's disabled because of the camera barrel distortion and the consequent difference between the distorted and undistorted position (distortion managed by the browser).
In other cardboard demos we are used to a "click at the looked object" mode to interact with the scene by using the cardboard button.

By removing the the works as expected also with the gyroscope, but not in the splitted screen hmd view.

Maybe an hypothetical "pick_object_hmd()" function could pick only the central point that's not distorted by the cardboard barrel…

We saw the space disaster demo, with the math for collisions but the pick_object is so straightforward in this case, so we tried to achieve this behaviour by using the calc_ray() but that returns an exception because this function is not allowed with this camera configuration

Is there a way to interact with the scene simulating a click at the canvas center at the actual development state?
Have you got an advice for us?
TY!!

Luigi

Amazing.
Thank you for your valuable help.
We found this bug report too: https://bugzilla.mozilla.org/show_bug.cgi?id=878067
Somebody seems to say that chrome is wrong… Actually we don't think so !

The code you provided works great! Thank you very much.
In android chrome mobile the rotation is now perfect.
We still have some issue with firefox mobile… it seems that firefox has a different reference point for the w3c-defined gamma angle.

Hi!
we tried to use the gyroscope module with the eye camera. We didn't manage to use it for our purpose.
It seems that only 2 rotation angles BETA and GAMMA are supported and we have not been able to control the z angle ALPHA (as described here: http://www.w3.org/TR/orientation-event/)

any hint?
thanks!
Luigi