1. Dear Users, Happy New Year! In order to celebrate the traditional Chinese Spring Festival, Zidoo company is scheduled for a holiday from 18th Jan to 15st Feb. We will be back to work on 16st Feb. If there is urgent message, pls contact service@zidoo.tv Your understanding will be highly appreciated if our holiday brings you any inconveniences.We are looking forward to your support in the coming new year and hope you have have a happy and joyful time in 2020! Best regards, ZIDOO Team
    Dismiss Notice

Feedback [PSA] The X9S outputs wrong color levels

Discussion in 'ZIDOO X9S' started by n_p, May 31, 2017.

Thread Status:
Not open for further replies.
  1. boblo

    boblo Member

    Could you elaborate this, please? Does the HDMI in app record wrong colors too?

    And are colors displayed on games/browser way too much deviated? Is it noticeable at first sight (without special tools)? Could this issue be similar to displaying sRGB images on a Wide Color Gamut monitor? I had a WCG monitor in the past and I had to send it back because colors were horrible.

    Thanks to all
     
    Last edited: Jun 27, 2017
  2. n_p

    n_p Active Member

    The material to learn this stuff is out there. I have a chip on my shoulder with people trying to to get "simple breakdowns of everything" on the internet. If you have objectifyable data in front of you, and you ask for a subjective breakdown by a random internet user, in the end you learned nothing and still didn't understand the issue.

    First HDMI IN (recording): I didn't check for color accuracy there. But if you connect any source to it, and play back a black clipping pattern (the thing you use to set your TVs brightness level), you'll see, that the X9S expects the input signal to be full range (== pc level, == 0-255). Otherwise the black level is visibly off. As in - you don't even need the black clipping pattern to see it, if you know what you are looking for. This is an "issue" as the standard for all AV equipment is limited color range. The same way that nowadays every graphics card driver has the option to choose between video level and pc level output (limited and full), recording equipment usually has a toggle for that as well. The X9S not only is missing such a toggle, it also is defaulting to the PC standard here, which hardly makes any sense, as the box uses YCbCr (limited) by default, and tries to play in the "video realm" of things. Thats an issue. black levels don't line up with default settings even in the best case scenario, and the X9S would like to force you to switch your entire video chain to "pc level" which is not standard.

    The color errors on non video content (menus, apps, games, ...), even in the best case scenario for YCbCr are significant and clearly visible. Above 3 dE 2000 they are considered visible, above 5 they are considered really visible (without a direct comparison to another screen), and Cyan is ending up at 11, Green is ending up at 9 and Yellow at above 5, from a baseline (how the screen performs) thats well below dE 3.

    The last time I checked, also meassuring tools were not used to see things, so when I talk about visible color differences, and I did before in this thread, ... yeah...

    "But will I see them?!"
    (The question thats really hidden behind the "can someone elaborate" sugarcoating..)

    It depends, but yes.

    dE (delta Error) is a measure of noticable difference in direct viewing (a bunch of people look at color fields in a scientific experiment setting). So by that standard - yes. (See above)

    It depends also on your monitor. But chances are that it isnt wrong in the adverse direction, so - yes again.

    It depends on your color vision. But the CMF (color matching function) thats used to describe the targets and the error amount is an average from a "normal" color perceiving population already - so yes, again.

    It depends on your color vision still, as an average is just that, an average (in this case even weighted, so more a median), but probably - yes, again.

    It also depends on the display technology you are using, as more and more "modern" screens, start to kreep into the territory, where metamerism failure starts to become an issue (color primary spectrum peaks become "too" narrow, and color perception variability explodes even within the normal color percieving public), but as this issue normally only increases perceived color error, yes - again.
    -

    Ok, after all of this, here is your actual answer. We make open standards in industries, so that all kind of different manufacturers can produce stuff, that then works with other stuff. If you are not adhering to a standard, and are outputing color levels "differently", your device effectively has just become junk.

    Its this clear cut, because, when we connect a Bluray Player, to an AV receiver, to a TV, we don't wan't to add color error in ever intermediate step. The TV(/monitor) gets a little leeway because displays (reproduction) are difficult (thats why people test monitors), but all the intermediate devices do not get any.

    If your device handles color output wrong, its just trash.

    Thats why you saw wesk05 post a table with bit values before. If one of those bit levels is wrong: trash. Otherwise you just accumulate errors, the longer your (input) signal chain gets, and if you do that - none of the production/reproduction chain (camera > edit bay > colorist > codec > media format > player > TV) would work. If you remember anything out of this, remember that standards are important.

    So "will I see the error" is the self actualizing question to ask (*mumbleasamillenial*), but the actual answer is, that if the math at that point is wrong and the error is there, its already game over, and everyone agrees.

    That said, added error margin is big enough, that you will see it, even directly connected to a TV in almost all cases. (Thats what you post dE numbers for)
    --

    No, the difference will not be like "what an sRGB signal looks like on a wide gamut monitor". "Wide color gamut" means, that the color triangle you see in the Calman Screenshots gets entirely "bigger" (colors become more saturated), the X9S remains within the rec709/sRGB color space, but colors are shifted (= tint/chroma issue). So in this case, all colors in the cyan spectrum get much more blue, in the yellow spectrum they get more red, and so on and so forth. You can literally see that in the triangles in the screenshots posted. You also see the rec709/sRGB targets there.

    Also, the tables contain xy (chroma) and Y (luminance) errors as well - so you can even look at the problem more granularly.

    You see what the issue with "just give me a simple answer" is? ;) (Its people who are willing to give them... ;) )
     
    Last edited: Jun 28, 2017
    boblo likes this.
  3. boblo

    boblo Member

    Wow, what a excessively technical answer for me! :eek:

    It seems this issue should be taken seriously by Zidoo and an official response is mandatory here soon. They cannot claim it's a minor bug because X9S/X10 are media players and video is not affected. Competitors Himedia/Egreat (with HiSilicon SoC) don't have this issue and they are in the same category.
     
  4. nikos_a

    nikos_a Active Member

    It would be interesting to see measurements from another RTD1295 box...
     
  5. n_p

    n_p Active Member

    I'm being so specific, because color science currently is in a bit of a rough spot, where its not an entirely sure thing, if and when a certain person does see a color match or not (the color metamerism failure issue). But thats an issue on the reproduction side, not in signal encoding. So when your video says 15,15,15 and your video player outputs 19,25,38 (still all digitally), its not so much a color science ("how will I perceive the color") issue, its a programming error (the math used is not correct).

    So instead of saying yes, you will see the color error. I'm hedging bets. ;) According to all current standards in color science, most people will see the color difference clearly, as the margins colors are shifted by are quite big. That said, your math is supposed to be correct almost regardless, because video distribution is dependent on it.
     
  6. boblo

    boblo Member

    The most direct competitor is Zappiti 4K HDR, with same SoC and specs (including the HDMI Input). It has an image processing engine called 'Magic Pixel 2'. Some people like it, other hate it. The problem is it cannot be deactivated so it's a risky business (I've read comments about Magic Pixel that claim is not suitable for old SD videos).

    Still no official word on this? Then I'll have to change my mind and go with another media center...
     
  7. nikos_a

    nikos_a Active Member

    Most direct is probably the Beelink SEA I. As I said, I am suspicious that it is not Zidoo related but rather, Realtek-related. This is why I would like to see similar test with another 1295 box.
     
  8. mirror

    mirror Well-Known Member SUPER Administrator Zidoo TECH Supporter

    There is no so-called ”Magic Pixel“ in realtek SDK,If someone tells you It has, it must be false.
    All image quality optimizations are related to the underlying decoding.
    No one can change it, except chip factory!!
     
  9. boblo

    boblo Member

    @mirror, thanks for you reply. That's not the problem, I'm not interested in Zappiti players, but in Zidoo players.

    The issue is that @n_p has discovered a bug in color reproduction when NOT playing video. Please see THIS POST.

    I'd like tech support confirm this issue.

    Many thanks, indeed.
     
  10. mirror

    mirror Well-Known Member SUPER Administrator Zidoo TECH Supporter

    Right answer! Please remember it is a 'Media player'
     
Thread Status:
Not open for further replies.

Share This Page