[ISSUE] Brightness mismatch between video players and apps

Discussion in 'ZIDOO X9S' started by n_p, Jul 8, 2018.

  1. n_p

    n_p Active Member

    Issue is as follows - while apps put out the correct color profile/video level (gamut, brightness, ..) in general - every tested video player on the X9S has a brightness mismatch, resulting in an image thats too dark at the low end.

    Players tested:

    ZDMC (based on Kodi 16)
    - internal zidoo player
    - kodi player (dvdplayer)

    Kodi
    - kodis player

    MXPlayer
    - MXPlayer h/w accelerated

    There is a "fix" thats not really a fix -but mitigates the issue and helps to explain it.

    In the Quick settings there is a settings option on the X9S that allows you to raise the brightness of the output signal. Default is 32 - for the video players brightness to fall "where it should be" this has to be raised to 34, at which point - all "apps" get displayed with the wrong brightness (brightness of near black colors) level.

    Thats why it has to be called a mismatch.

    Also using the quick menu item to "solve" the issue for video players isn't ideal, because the video level gets raised using a quick conversion matrix (probably) that isnt as exact as a real transformation. Meaning, I'd rather get correct video levels out of this box, than to use the software equivalent of a "graphics card brightness slider".

    How this has been tested:
    TV calibrated to a known good source, "app brightness level" tested with the mobileforge app and calman. Looking at a luminance graph for 2.2 gamma: result, everything falls in line - the X9S outputs correctly.

    Then using the same calman setup to measure video levels from test videos in Kodi (kodi player ((dvdplayer)), and suddenly the luminance graph is noticably off. It only gets realighned, if you raise "quick settings menu brightness" to 34. (At which point now the mobileforege readings are off.)

    Then matched brightness test videos (black bars) accross three players (ZDMC, Kodi, MX player) - all show the same behavior.

    Conclusion:

    There is a mismatch between "normal video output" and probably "h/w accelerated video output" in brightness - which renders the X9S useless again.

    Category 1 issue, as in "Zidoo player and ZDMC are not color accurate".

    Again.


    Why it took so long to notice - I'm running on a B6 OLED where LG crushes the first near black levels on purpose - so the mismatch wasn't immediately obvious on a brightness test pattern. That said, it is, once you see the "corrected" difference.

    Also - the measurements (luminance graph in calman) were made with video levels 10% grey and above, to make sure, that LGs near black handling doesnt factor in.

    Measurements were taken in limited color modes (auto and limited settings in the x9s) on rec 709 content.

    I'll provide images of the graphs tomorrow.
     
  2. wesk05

    wesk05 Member Beta test group

    I checked Mediacenter, Hometheater 2, ZDMC (internal and external players), MX player, VLC player and Kodi. I didn't see anything wrong with the absolute YCbCr code values (64-90). The code values of the output are an exact match to the source. I only checked 4K24 10-bit YCbCr 4:4:4. The one issue with ZDMC internal player and VLC is, the output is truncated 8-bit.

    You are not using auto mode for dynamic range, are you?
     
  3. n_p

    n_p Active Member

    Thank you wesk05. Ill do a few more checks and come back with more information.

    So far I've tested on YCbCr 4:2:2, and auto dynamic range could have been enabled... The thing is, that the blacklevel difference isn't a full limited/full shift, its less than that, otherwise I'd have looked into that already.

    Also I'm on 1.4.8 - which complicates the comparison. I might have to update, but I'll do it in a last step.

    If you read this, could you specify the "truncated" part?

    I also have to try a direct connection to the TV, as my signal still goes through the eecolor signal processor, so maybe it doesnt handle the truncated 8bit correctly. (Never had similar issues with any other source device though.)

    edit:
    I suppose, thats what you mean by "truncated". That said, this shouldnt be the cause of the issue described.. :)

    I'll do a few more measurements. And will come back with the results.
     
    Last edited: Jul 16, 2018
  4. n_p

    n_p Active Member

    *hold head low*

    Cant reproduce it. (Auto settings or limited on YCbCr 4:4:4 and 4:2:2)

    The closest I come is switching from zero lut to my current lut profile (which is being retaken, while I write this), so I must have switched luts and not noticed.. :(

    Sorry for the work and quarrel caused.
     

Share This Page