krpano 1.19 - Pre Release 3

  • the form and amount of distortion depends on the individual headset and can't be auto-detected

    Howdy Klaus,

    Admittedly I'm new to VR mode but I have to wonder. In the case of cardboard the viewer parameters should be the same across all headsets. I would guess that when being displayed on a phone in a browser it would be in a cardboard or similar type viewer so KRP could do the correction automatically. What headset / phone / browser combinations require a distorted display?

    BTW, showed off the VR Mode tour at a trade show over the weekend. Lots of positive reaction. YOU make us look good.


  • The lens distortion of the Oculus Rift itself isn't perfect - it depends on a lot of factors like the exact position of your eyes inside the headset. The only way to have an bit a control of that distortion is via the Oculus Rift configuration tool.

    But still somehow krpano has this distortion/bending while other panoramas don't. For example, https://mozvr.github.io/panorama-viewer/ behaves better even when you download it and replace the landscape image (which has no vertical lines) with an image with a vertical line. Both tested with Firefox Nightly and Oculus DK2 on the same computer.

  • Tell me if I'm wrong,
    it seems that toggleVR doesn't call webvr_onentervr ?
    I added webvr_act_as_gyro_in_portrait_mode(); in webvr_onentervr, works perfect when webvr.enterVR() is called, but when webvr.toggleVR(); is fired then webvr_act_as_gyro_in_portrait_mode(); isn't executed ???

  • Quoted from "jeromebg"
    because one of my clients had loading problems with a tour I did.
    I have indeed a problem on iphone, sometimes some cube face just don't load and I have no errors in the log...

    No error would be strange...

    When the server is slow or busy and reports a loading error or times-out krpano still tries the loading again retrycount times before reporting an error in the log, no loading and no error is not normal.


    Found the problem after hours of reviewing my code, very simple in fact...
    It's only the mobilevr_wakelock that is not very friendly with safari on IOS (at least IOS 9.2) , settings it to false solved all my strange cube faces loading problems !
    YEEEEEEPY !!!!!!!!!!!!!

    will 1.19.4 have a safrai friendly hack for wakelock ?

    THX !

  • Hi,

    In the case of cardboard the viewer parameters should be the same across all headsets. I would guess that when being displayed on a phone in a browser it would be in a cardboard or similar type viewer so KRP could do the correction automatically.

    All 'cardboard' headsets are requiring lens pre-distortion!
    And the lens are different between many headsets, even between so called 'Google Cardboard' headsets!

    And it's not possible to auto-detect the lens type and position as there is no connection between the Cardboard and the phone (it's basically just a cardboard or plastic box with lens).


    What headset / phone / browser combinations require a distorted display?

    All!

    But still somehow krpano has this distortion/bending while other panoramas don't.

    Please don't mix up the display on the computer with the Oculus Rift with smartphone - in case of the Oculus Rift, it can display the undistorted view on the PC and inside the headset you will get the distorted one. When using the WebVR API the browser is the only one that is responsible for the lens distortion!

    When testing make sure on desktop make sure to test the real WebVR API and not the mobile fake mode!
    I would recommend testing using the Chromium WebVR build for Windows.

    What is your testing environment? (ALL details!)

    Tell me if I'm wrong,
    it seems that toggleVR doesn't call webvr_onentervr ?

    It does...

    will 1.19.4 have a safrai friendly hack for wakelock ?

    Sorry, but I've no problems with the wakelock...
    The iOS wakelock is just a page-reload-try-and-abort every 15 seconds.
    There is no other wakelock hack so far I known. Other scripts are basically doing the same.

    Best regards,
    Klaus

  • Please don't mix up the display on the computer with the Oculus Rift with smartphone - in case of the Oculus Rift, it can display the undistorted view on the PC and inside the headset you will get the distorted one. When using the WebVR API the browser is the only one that is responsible for the lens distortion!

    When testing make sure on desktop make sure to test the real WebVR API and not the mobile fake mode!
    I would recommend testing using the Chromium WebVR build for Windows.

    What is your testing environment? (ALL details!)

    I'm using Windows 10, Firefox Nightly, Oculus DK2, NVidia GTX 960 with recent drivers. What other details do you need to know? When I mentioned the distortion I was always talking about Oculus DK2, I'm aware that I cannot use the WebVR fake view to judge distortion. Can you reproduce the problem with the attachment I posted earlier today? Or can anyone else here with an Oculus confirm the problems? Here's the link again: https://www.dropbox.com/s/c9jhd6z6iw97…ortion.zip?dl=0

    I cannot use the Chrome WebVR build as it crashes for me.

  • Hi,

    I'm getting strange rendering errors with the current Firefox Nightly inside the Oculus Rift, but Chrome WebVR is working fine for me (I'm on Windows 7).

    Please keep in mind that all about WebVR is still experimental!
    The API itself, the browers and the Oculus runtime and drivers aren't very stable too today.

    But from krpano side and how krpano uses the WebVR API everything should be right - krpano uses the WebVR according to the specs and there is not any rendering distortion added by krpano of course. When using the WebVR API, krpano renders the pano directly without any distortion direct by using the WebVR projection matrix and the WebVR screen settings.

    Maybe try somehow (don't ask me how) to get Chrome working - for me it's WebVR support was always working way better than the Firefox Nightly one...

    Best regards,
    Klaus

  • Maybe try somehow (don't ask me how) to get Chrome working - for me it's WebVR support was always working way better than the Firefox Nightly one...

    I will try once more with Chrome tomorrow. If someone knows more recent Chrome WebVR builds than the ones on https://drive.google.com/folderview?id=…bW9WTHMtOWMzNjQ, please let me know.

    For Firefox Nightly, I recommend turning *off* "Enable multi-process Nightly" as it might not work well together with WebVR. Plus, of course, making sure that Nightly is actually up-to-date. In case you have the head tracker attached, try removing it, it helps me get less judder.

  • Sorry, but I've no problems with the wakelock...
    The iOS wakelock is just a page-reload-try-and-abort every 15 seconds.
    There is no other wakelock hack so far I known. Other scripts are basically doing the same.


    Hi Klaus, I do believe you have no problems with wakelock but don't understand why I have some, only on iphone & webvr mode ?
    Here is exactly the same tour, one with wakelock false, the second one with wakelock true:

    http://www.covecreekproductions.com/tours/kenmont/vr.html

    http://www.covecreekproductions.com/tours/kenmont/vrlock.html

    One the one with wakelock set to false I have no problems on iphone in webvr.
    One the second one with wakelock set to true I have no problems on iphone in normal mode, but in webvr mode some cube faces won't load, or I have some loading errors on preview files or hotspots.

    Thoses problems only occurs on my iphone, not on my LG G3.
    Iphone 6+ IOS 9.2

    Any ideas ?

  • Hi Jerome, Klaus,

    i confirm this on my iPhone 6 iOS 9.2

    first url is ok
    second keeps loading and only shows loading and the preview in vr mode.

    tonight i 'll update to 9.2.1 .. we'll see if that makes a difference.

    Tuur *thumbsup*

  • Hi,

    okay, I see now and I think I know what the problem is:

    The iOS wakelock hack Javascript code looks like that:

    Code
    setInterval( function() 
    {
      window.location = window.location; // request a page reload by redirecting to itself
      setTimeout(window.stop, 0); // stop the loading one tick later
    }, 15000);
    • Here a function will be called every 15 seconds,
    • and inside that function the 'window.location' setting will be set to itself to request a reload of the current page.
    • This page reload-request resets the internal iOS Safari turn-off timer - this is the 'wakelock'.
    • Then directly after that reload-request in the next Javascript 'tick' (via setTimeout+0), the window.stop function will be called to stop that loading request.


    And the problem here - that window.stop call also stops the loading of all other elements. So when there was something loading exactly during this step, it will fail (sometimes silently, sometimes with error as it seems). This is also the reason why it was sometimes working and sometimes not.

    I will check this in detail and will try improving that for the next release - probably by internally telling the loaders that there is currently an iOS wakelock in progress and to automatically retry/redo the loading in this case.

    Best regards,
    Klaus

  • I create a hotspot with url as textfield.swf, and html text, but in VR mode, this hotspot disappeared,
    if the hotspot is an image, it still appear in VR mode,
    why?

  • I will check this in detail and will try improving that for the next release - probably by internally telling the loaders that there is currently an iOS wakelock in progress and to automatically retry/redo the loading in this case.


    Great, thanx Klaus for looking !

    if the hotspot is an image, it still appear in VR mode,
    why?


    If you read the docs you'll see that only image hotspots are supported for now in VR mode.

  • hello, thanks for the new krpano version. I'm new to this vr viewer. Is there a user's instruction or something like that?
    can I use it on windows PC? I opend the demo link in Chrome, it seems that I can't control it with the mouse.
    I clicked the "Look here to start",but nothing happens, and I clicked the "Enter VR", the screen splited and I still can't control it with mouse.

    something I do wrong? or I need some device or equipment?

    sorry for the silly questions.

    by the way, my system is windowsXP

    Edited once, last by relationcg (January 27, 2016 at 3:52 AM).

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!