1399733 - Position of headset incorrect (too high) after refreshing the page and entering VR with onvrdisplayactivate handler

Status: RESOLVED INVALID

(Core :: WebVR, defect, P3)

Description

[Kevin Ngo]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/60.0.3112.113 Safari/537.36

Steps to reproduce:

1. Go to a VR-enabled page (not in an I-Frame) that handles onvrdisplayactivate.
2. Enter VR.
3. Refresh the page.


Actual results:

The page maintains VR mode. The height of the headset is too high (by about a couple of meters). The position of the controllers are correct. The headset and controllers still track.


Expected results:

The page maintains VR mode. The height of the headset and controllers match the real world position.

Comment 1

[Kevin Ngo]

On Firefox Developer Edition 56.0b11 (64-bit)

Comment 2

[:kip (Kearwood Gilbert)]

Could you try opening this page in another tab while doing your STR?

https://webvr.info/samples/00-hello-webvr.html

Before and after refreshing the page in your STR, do the values of "sittingToStandingTransform" change or pose/position show a translation that would account for the offset?

This may help us to bisect the problem between possible bugs in the browser or content.

Comment 3

[Daosheng Mu[:daoshengmu]]

I encounter this kind of issue and can be resolved by using a new profile. It seems to make browser.tabs.remote.autostart.2;true could this problem. Please help us take a try. Thanks.

Comment 4

[Kevin Ngo]

So it was not an issue with the pose or transform data, and new profile didn't do anything. It was intermittent, but the suspicion is that we enter VR early using the `vrdisplayactivate` event, we grab the VR display, and check for `vrDisplay.capabilities.hasPosition` to determine if we are on room scale, we remove a default 1.6 meter height offset which we use for 3DoF/2D mode.

So possibly, the VR display's `capabilities.hasPosition` is not set soon enough. For a workaround, I suspect adding a timeout before doing that VR display check will fix things.

Comment 5

[Kevin Ngo]

Sorry, this is an A-Frame issue where our scene entered VR before we had our event listeners attached to remove a camera offset.

Comment 6

[:kip (Kearwood Gilbert)]

(In reply to Kevin Ngo from comment #5)
> Sorry, this is an A-Frame issue where our scene entered VR before we had our
> event listeners attached to remove a camera offset.

Thanks for tracking this down, Kevin!