[doveman]

I just noticed this in the Release Notes for v93.81 Forceware drivers (page 27, Open Issues) and thought it might be of interest to someone.


• Video color-space range for DVI-only* outputs is erroneously set to

standard mode (16-235) instead of extended mode (0-255).


* “DVI-only” means only one display is connected, and it is to the DVI output.


A new detection feature to apply Standard CSC mode to TV outputs

(including NTSC, PAL, 480i, and 576i), included DVI-only outputs by

mistake.


Note: The driver correctly applies extended mode to analog outputs, and

standard mode to TV outputs (including NTSC, PAL, 480i, and 576i).


A future driver release will correct this and apply the extended-mode color

space to DVI-only outputs.



You can work around this issue by forcing either standard or extended mode

as follows:


1. Launch regedit and determine the current primary display card by

looking in


HKey_Local_Machine\\Hardware\\DeviceMap\\Video


and note the GUID (global unique identifier assigned by Windows),

which is the long string in brackets { } at the end of the entry


ʺ\\device\\video0ʺ.


2. Look in


HKey_Local_Machine\\SYSTEM\\CurrentControlSet\\Control\\Vide o\\{GUID}\\0000


where {GUID} is the number derived from the previous step.


3 Open the ʺ0000ʺ directory and create a new DWORD called

VMRCCCSStatus and give it a value of


0x3 - to force use of the standard YUV range of 16-235


0x1 - to force use of the extended YUV range of 0-255

 

[Stereodude]

That's been in the release notes for all the 9X.XX versions of the drivers.

 

[skepticon]

I've been reading that little blurb for about a year now.

"A future driver release will correct this and apply the extended-mode color

space to DVI-only outputs."...Never gonna happen.

 

[arfster]

That's been around since 84.x - as per above, Nvidia don't fix HTPC bugs. If they did, people would have no reason to buy their new products.

 

[lstepnio]

Isn't the work-around just as good? What should we be using for a HTPC, 0x3/16-235 or 0x1/0-255?

 

[Stereodude]

FWIW: The 92.71 drivers work fine for me without any messing with the registry, and I haven't adjusted anything since I originally calibrated with the older 8x.xx series drivers that didn't have the problem.

 

[casandra9]



Quote:
Originally Posted by lstepnio /forum/post/0


Isn't the work-around just as good? What should we be using for a HTPC, 0x3/16-235 or 0x1/0-255?

Hi,

for a digital panel (incl. DLP) the 0-255 range is the choice. For an anlog TV 16-235 is probably better. But it depends on the TV.


sandra

 

[DPlettner]



Quote:
Originally Posted by lstepnio /forum/post/0


Isn't the work-around just as good? What should we be using for a HTPC, 0x3/16-235 or 0x1/0-255?

This has been discussed before, but the release notes have it backwards. It's 0x1/16-235 and 3x1/0-255.


If you are using a computer monitor, it probably expects 0-255, and an HDTV probably expects 16-235. However, your display may also have a setting in the setup menu. My HP DLP can be set for PC Levels (0-255) or default levels (16-235).


Most content (e.g., DVDs) will be encoded at 16-235. Therefore, the best picture will be obtained if you set your display for 16-235 and the nVidia drivers for 16-235. However, IMO, 0-255 looks pretty good too if that is what your display expects.


Also, it is pretty easy to try both settings and see what looks best to you.


-Dave

 

[doveman]



Quote:
Originally Posted by Stereodude /forum/post/0


That's been in the release notes for all the 9X.XX versions of the drivers.

Ah, sorry. I don't normally bother reading the notes but I was hoping they might have fixed something else. No such luck.

 

[doveman]



Quote:
Originally Posted by arfster /forum/post/0


That's been around since 84.x - as per above, Nvidia don't fix HTPC bugs. If they did, people would have no reason to buy their new products.

Hmm, aren't all their products affected by this bug?

 

[doveman]

Correction, it's actually p.19-20 if you're going by the page numbers printed on the document.


This, from p.17, makes it all the more pathetic:

--

Single display modes such as TV only, DFP/LCD only or CRT only

provide the best performance and quality from Windows Media

Center Edition.


Dual display modes such as Dualview and nView Clone and Span modes are

not recommended.

--

So single DVI doesn't work properly and Dual display is not recommended!

 

[doveman]

Now I'm wondering what colour space it's using on the DVI when it's in Clone or Dualview mode.


On my system it recognises my Sony CRT TV and there's a 'Treat this display as HDTV' switch, so maybe this puts it into Standard mode.

 

[mark007]

Wow I have been playing with my nvidia card + vmr9 recently. I do notice on my pc using dvi I have banding issues, very noticable sometimes, even with hd demos I have downloaded from the net.


The Dust To Glory demo from the microsoft site shows huge banding at the beginning of the clip (10 seconds in).


I would appreciate if anyone could check to see if these bands appear for them also. I would love a fix for this if it is indeed related to an nVidia colour bug.


http://download.microsoft.com/downlo..._Glory_720.exe

 

[Victor]



Quote:
Originally Posted by DPlettner /forum/post/0


Most content (e.g., DVDs) will be encoded at 16-235. Therefore, the best picture will be obtained if you set your display for 16-235 and the nVidia drivers for 16-235.

Multiplication by 3*3 colorspace conversion matrix is always performed during YPbPr/YUY2/YV2 to RGB conversions. 16-235 and 0-255 differ only in coefficients in that matrix. Having 256 steps vs. 220 translates to bit smaller error. If you have properly calibrated display, you will not see blacker than black and whiter then white anyway.

 

[doveman]

To complicate matters, there's a question of whether Nvidia cards send the correct signal over DVI to tell the HDTV to use the SD or HD colour space/gamut (which is unrelated to this standard/extended mode).


Certainly my TV suggests that it doesn't. It has a MTRX parameter and it changes to 1 when using my Xbox in 720p/1080i mode via component, but always stays on 0 when using my PC's DVI. Someone with a similar TV to mine (XBR960) with a Radeon X1600 Pro hooked up via component says it's working as it should.


This is from the service manual for an XBR960 TV:


"For HDMI it depends on a flag in the signal, and it can be 1 or 0. Here is where your favorite DVD player via HDMI can flip the color matrix for HD, if it's designed to do so. If the flag ("uTiny/$6D_00, byte1/bit6") is 0, MTRX is 0 (NTSC); if it's 1, MTRX is 1 (HD)."

 

[skepticon]

Doveman,


I don't expect nvidia to provide any helpful feedback on the issue, but it would probably be a good idea to report that one directly to them. They state that their email support is only for products purchased from their web site, such as the dualTV, but you can still send them feedback for other products as well. Hopefully the information will make it to someone that can do something useful with it (just don't expect a reply back from them).

 

[doveman]

Yeah I will do.


I asked them about something else the other day and actually got a reply! Mind you, it only said that their cards do what I was asking about, when they don't (at least the one I've got doesn't)

I replied within 72 hours pointing this out but haven't heard back. Can't really expect them to fix something if they think it's already working!


I need to find out a bit more about this flag. It's somewhat complicated, in that only video recorded for the HD colour space should really be displayed using it, but if the driver tells the TV to use it when it's in an HD (720p/1080i) mode, then the TV will use it all the time, even when viewing SD video. I've no idea if there's provision for this flag to be encoded in the video file, which would be better, providing that the playback software and video driver then pass it on.


Certainly with my Xbox, if it's operating in an HD resolution, the TV switches to use the HD colour space all the time. As I don't really use it to watch HD video, I just set video playback to use 480p.

 

[stanger89]



Quote:
Originally Posted by doveman /forum/post/0


I need to find out a bit more about this flag. It's somewhat complicated, in that only video recorded for the HD colour space should really be displayed using it, but if the driver tells the TV to use it when it's in an HD (720p/1080i) mode, then the TV will use it all the time, even when viewing SD video.

Whoa, back up there, I think you're confusing two issues here. Yes there are two different standards for SD and HD video, ITU-R BT.601 and ITU-R BT.709 respectively, and they do specify different matrices for converting component to RGB and vice versa.


However they both specify the same levels for black and peak white, 16 and 235 respectively, ie "video levels".


The issue with the nVidia drivers describe here is whether or not they expand from "video levels" (16-235) to "PC levels" (0-255).

 

[doveman]

I did say "there's a question of whether Nvidia cards send the correct signal over DVI to tell the HDTV to use the SD or HD colour space/gamut (which is unrelated to this standard/extended mode)." but maybe I should have left it for another topic.


It is just another example of yet another bug that Nvidia needs to fix before their cards are suitable for HTPC use (not that I'm suggesting any other make does any better).


I'm somewhat confused by your statement that the matrices are for converting component to RGB and vice versa. Is it not the case that DVI and HDMI carry digital RGB, in which case why would there be provision for a flag to tell the TV which matrix to use?