[Tinwarble]

I can't find them now - 

It had been discussed in THIS thread: https://forum.kodi.tv/showthread.php?tid...pid2689145

[hdmkv]

Anyone using an Optoma 4K projector with Shield? Preferably UHD50 or the new UHD51A? I recently got the latter, and can’t get 10 or 12-bit output out of Kodi (using v18 Alpha 1) and 23.976 or 24.000. With 4K@60 HDR (like the Sony Camp demo clip), I get 10-bit 4:2:0. But, with 23.976, it’s always 8-bit 4:2:2.

With Zidoo X9S and new X20, with 23.976, I get 4:4:4 12-bit as expected. Can’t figure out why Shield won’t do same?

[wesk05]

Anyone using an Optoma 4K projector with Shield? Preferably UHD50 or the new UHD51A? I recently got the latter, and can’t get 10 or 12-bit output out of Kodi (using v18 Alpha 1) and 23.976 or 24.000. With 4K@60 HDR (like the Sony Camp demo clip), I get 10-bit 4:2:0. But, with 23.976, it’s always 8-bit 4:2:2.

With Zidoo X9S and new X20, with 23.976, I get 4:4:4 12-bit as expected. Can’t figure out why Shield won’t do same?

 Shield only supports Deep Color at 4K 50/60Hz (10-bit 4:2:0). At lower refresh rates it is 8-bit 4:4:4 for Rec. 709 or 12-bit 4:2:2 for Rec. 2020. The 8-bit that you are seeing on your projector or AVR is unlikely to be true. There are very few devices that can accurately detect the bit depth of 4:2:2 signal. Most of them display 8-bit (including some HDMI analyzers and HD Fury Vertex). 

If you do want to confirm whether your projector is treating the signal as 8-bit or 12 (10)-bit, check it out with one of the 2DQuants Quants2D test pattern that I mentioned in my previous post. If the signal is being processed as 10/12-bit 4:2:2, you should see something like in this picture- https://imgur.com/wqAlDRk . You will see 4 block wide bands, if it is processed as 8-bit as you can see in the preview below:

[wesk05]

(wish we'd call it gamut conversion to distinguish from 'same primary' colour space conversion like 601<->709)  

@wesk05 should be able to tell us - I think he's done some proper analysis of the previous 709->2020 gamut conversion that the Shield TV did. (Reporting back that it wasn't perfect but that it was better than some feared ISTR) 

To be technically correct, this is gamut mapping, but the conversion process does involve color space transformations (YCbCr to RGB to XYZ and back). I have used gamut and color space interchangeably which is really a blasphemy in the color science world ).

Correct, whatever conversion it was doing in 6.3 wasn't all that terrible. The errors were visually apparent only at the highest saturation levels and that too was dependent on the native gamut of the display. I have no idea whether this has changed at all in the 7.0 update. My Shields haven't been updated yet. As a matter of principle I do not enter into NDAs, the terms of which I don't have full knowledge of.

[Tinwarble]

As a matter of principle I do not enter into NDAs, the terms of which I don't have full knowledge of. 

There's no NDA for the Preview program, they only ask that you don't discuss thing mentioned in the Preview forum outside the Preview forum. Though, as you can see from the regular forum, many people don't even do that.  Nvidia themselves don't even always follow their own rules.

But there's nothing that you have to sign and no consequences for breaking that rule.

[wesk05]

There's no NDA for the Preview program, they only ask that you don't discuss thing mentioned in the Preview forum outside the Preview forum. Though, as you can see from the regular forum, many people don't even do that.  Nvidia themselves don't even always follow their own rules.
But there's nothing that you have to sign and no consequences for breaking that rule. 

I consider putting limitations on communication as one of the terms in a NDA. There may not be a proper NDA to sign and because of that this limitation on communication may not even have legal validity. By joining the Preview program, you agree to the SHIELD software license agreement and thereby to the Beta Program supplement in that agreement. I also consider it as a waste of time, since these beta or preview releases are really release candidates. It's for all these reasons that I decided not to participate in this program.

[Tinwarble]

I consider putting limitations on communication as one of the terms in a NDA. There may not be a proper NDA to sign and because of that this limitation on communication may not even have legal validity. By joining the Preview program, you agree to the SHIELD software license agreement and thereby to the Beta Program supplement in that agreement. I also consider it as a waste of time, since these beta or preview releases are really release candidates. It's for all these reasons that I decided not to participate in this program. 

 
Your communication is limited anyway. If you join the program you are not supposed to speak about it and if you don't then you don't know anything in which to speak about.

But as I said, Nvidia has never even attempted to enforce their rules and often break their own rules. Besides that, much of what you could speak about is what has been added by Google for Android and is already known to the general public which would make it nearly unenforceable.

I would also not say that it is a waste of time, yes the preview releases are the same as the general public release (usually), but there have been changes made before the release goes public. That is why this update was pulled before it went to the general public, because Previewers found bugs that needed to be fixed before going to the general public.

I understand your reasoning behind not wanting to be in the program, but it seems mostly arbitrary. It's your choice though, and wouldn't try and convince you otherwise. I guess it really just comes down to if you want it early or just want to wait until it goes to general release.

[wesk05]

I understand your reasoning behind not wanting to be in the program, but it seems mostly arbitrary. It's your choice though, and wouldn't try and convince you otherwise. I guess it really just comes down to if you want it early or just want to wait until it goes to general release. 

Fair enough! I'm in no hurry to get the update

[hdmkv]

Anyone using an Optoma 4K projector with Shield? Preferably UHD50 or the new UHD51A? I recently got the latter, and can’t get 10 or 12-bit output out of Kodi (using v18 Alpha 1) and 23.976 or 24.000. With 4K@60 HDR (like the Sony Camp demo clip), I get 10-bit 4:2:0. But, with 23.976, it’s always 8-bit 4:2:2.

With Zidoo X9S and new X20, with 23.976, I get 4:4:4 12-bit as expected. Can’t figure out why Shield won’t do same?

 Shield only supports Deep Color at 4K 50/60Hz (10-bit 4:2:0). At lower refresh rates it is 8-bit 4:4:4 for Rec. 709 or 12-bit 4:2:2 for Rec. 2020. The 8-bit that you are seeing on your projector or AVR is unlikely to be true. There are very few devices that can accurately detect the bit depth of 4:2:2 signal. Most of them display 8-bit (including some HDMI analyzers and HD Fury Vertex). 

If you do want to confirm whether your projector is treating the signal as 8-bit or 12 (10)-bit, check it out with one of the 2DQuants test pattern that I mentioned in my previous post. If the signal is being processed as 10/12-bit 4:2:2, you should see something like in this picture- https://imgur.com/wqAlDRk . You will see 4 block wide bands, if it is processed as 8-bit as you can see in the preview below:

 

 Thanks, I'll try the test pattern... can you point me to the post where I can download it? I did a full forum search on "2DQuants" and "wesk05" and only above post came up.

[wesk05]

 Thanks, I'll try the test pattern... can you point me to the post where I can download it? I did a full forum search on "2DQuants" and "wesk05" and only above post came up. 

This is the post: https://forum.kodi.tv/showthread.php?tid...pid2739108 The "Releases" link will take you to the GitHub page that has the download link.

Not wonder it doesn't show up, it is Quants2D

[hdmkv]

Thanks, of course now I recall that thread/post . So, just tested both the Cb & Cr light gray patterns in the HDR10 folder...


With nVidia Shield TV:


With Zidoo X9S & X20:


...and the results look very similar, if not the same. Zidoo's looks cleaner, but both look closer to the 8-bit screenshot you posted above. Not completely solid gray as it should be.

[wesk05]

Thanks, of course now I recall that thread/post . So, just tested both the Cb & Cr light gray patterns in the HDR10 folder...
...and the results look very similar, if not the same. Zidoo's looks cleaner, but both look closer to the 8-bit screenshot you posted above. Not completely solid gray as it should be.

I forgot to mention, try the non-HDR ones in the BT.2020 folder.

[Koying]

Oreo introduces "Wide Gamut" for egl: https://developer.android.com/training/w...lor-gamut/
They don't explicitely say, but looks like native bt2020, isn't it?

Anyway, as said in the other thread, this gives no control on display colorspace, nor what "surface" does.
*Maybe* having a wide gamut surface for GUI might influence the colorspace when doing mode switch, but it's only supputations (and the shield oreo I have doesn't seem to have the egl extensions mentionned in the link)

[movie78]

Thanks, of course now I recall that thread/post . So, just tested both the Cb & Cr light gray patterns in the HDR10 folder...


With nVidia Shield TV:


With Zidoo X9S & X20:


...and the results look very similar, if not the same. Zidoo's looks cleaner, but both look closer to the 8-bit screenshot you posted above. Not completely solid gray as it should be.

@wesk05  @hdmkv 

These are some great test, I always wonder if the devices are displaying the correct HDR
Going to try these files on my Egreat A5, Nvidia Sheild, Minix U9, Himedia A5 and Windows 10 PC.

Thanks, guys!

[hdmkv]

Thanks, of course now I recall that thread/post . So, just tested both the Cb & Cr light gray patterns in the HDR10 folder...
...and the results look very similar, if not the same. Zidoo's looks cleaner, but both look closer to the 8-bit screenshot you posted above. Not completely solid gray as it should be.

I forgot to mention, try the non-HDR ones in the BT.2020 folder. 

 Okay, just tested all 4 light gray samples in the BT.2020 folder...



...and, for all 4, the image is solid light gray and even, zero bars/blocks. So, this means I'm getting proper 10-bit even though PJ reports 8-bit?!