madVR - high quality video renderer (GPU assisted) [Archive] - Page 996

mclingo

23rd March 2018, 11:54

OLED graduation has been discussed to death on AV forums, its not as simple as the panel itself. Early panels had issues where poorly compressed material would create quantisation errors and blocky noise in dark areas but a movie with no compression artifacts looks perfect in dark areas. Later panels dealt with this with what we assumed to be better dithering algorithms but most issues on OLED are caused by poor source material, push great in, get great out, push crap in ..... I dont notice any banding at all on my OLED, even at near black, unless its on the source and mine is gen one. There were some differences in the panels, the later 2015 models tended to be better and a few lucky people like me got a really good example, although I had to send two back with terrible near black vertical banding which is the real Achilles heal of OLED. It can still be a bit of a panel lottery but its a lot better than it was, i'd still buy my current OLED over any LCD, FALD or otherwise as i'm a dark room viewer and OLED cant be beaten in this field, once you go OLED there is no going back unfortunately :)

ryrynz

23rd March 2018, 12:07

Yeah there ain't no beating that perception of depth and clarity that OLED brings. Every day I see the C7, A1 and EZ950 and Sammy's Q8 while good, don't compare. I'll never buy an LCD TV :P

bran

23rd March 2018, 12:18

It would be useful to have a user with an AMD card and the same display. Then you could determine if its the extra 2-bits that is causing the problem.

THIS.

Exactly what I've done - exchanged nvidia (12b) for AMD (10b) on my 2015 flagship Samsung JS9505 (FALD). Banding much reduced.

huhn

23rd March 2018, 12:36

was there still banding?

bran

23rd March 2018, 13:44

was there still banding?

Some, but not nearly to the same extent. And it's hard to rule out the material as well. Tested using Planet Earth II, which some reviewers have noticed banding in the skies as well.

j82k

23rd March 2018, 14:06

I'd suggest using test patterns and not movies for testing stuff like banding. With movies you never know if the source is already bad.
This site has a ton of HDR10 test patterns and they're free.

https://yadi.sk/d/RPrX2C7l3HEjPq

Warner306

23rd March 2018, 15:25

Some, but not nearly to the same extent. And it's hard to rule out the material as well. Tested using Planet Earth II, which some reviewers have noticed banding in the skies as well.

So what happens if you choose 8-bits? Does the banding get even better? Some banding sounds strange. The processing still isn't doing its job.

Warner306

23rd March 2018, 15:26

Which HDR mode will I want .

NV HDR or OS HDR ? does it matter? NV HDR happens when it auto detect and enable TV HDR automatically when I run the movie with Potplayer + MadVR . OS HDR happens when I open the movie with HDR Enabled in the windows display settings.

I would go with the Nvidia HDR. Otherwise, you will be switching the OS switch every time you watch an HDR video.

GCRaistlin

23rd March 2018, 16:13

These screenshots are taken in MPC-HC by Alt-I: with EVR (http://tinypic.com/m/k4fsz8/4), with madVR (http://tinypic.com/m/k4ft03/4). No noticable differences.
These screenshots are taken by PrtScr: with EVR (http://tinypic.com/m/k4ft3p/4), with madVR (http://tinypic.com/m/k4ft5c/4). It may be noticed that EVR's one has more grain (see the sleeve). Why?

(Tinypic changes picture size, nothing can be done about that. Here (http://mir.cr/Q08T7IYT) are the original images).

oldpainlesskodi

23rd March 2018, 16:14

I would go with the Nvidia HDR. Otherwise, you will be switching the OS switch every time you watch an HDR video.

That's what I always thought....

The Martian (2015) 4k HDR. The scene from 00:10.10 looks extremely different when played with Madvr passthrough with metadata, vs madvr passthrough without metadata and nvcp set to use default color settings and the OS HDR toggle on.

Does any else have this film and test as above?

K

Win10 (latest fast ring) and GTX 1080ti (latest driver)

mclingo

23rd March 2018, 16:49

when you say different what are we talking about here exactly?, darker, lighter, washed out, over saturated, grainy, banding?

oldpainlesskodi

23rd March 2018, 16:58

when you say different what are we talking about here exactly?, darker, lighter, washed out, over saturated, grainy, banding?

Colour reproduction (more vibrant with correct colour tones) sharpness, clarity/image depth and the absence of aliasing = using nvcp default colour with OS HDR toggle as per my previous post.

In fact, I've just tried about 30 or so HDR movies and the differences I'm seeing have me scratching my head

K

mclingo

23rd March 2018, 17:15

I'm running on an AMD 460 so cant test i'm afraid but when I play HDR movies using the windows FILMS a& TV app with the windows HDR toggle on they look very similar to the output of MADVR suggesting an NVIDA HDR issue to me.

Not sure why you 'd want to run HDR that way though, but I know there have been some issues getting HDR to work properly with some NVIDIA driver versions, AMD has been working fine with HDR for a long while now but did have some issues with earlier RX 4XXX drivers if memory serves.

AMD is pretty much play and go at the minute for HDR, no issues at all. I'd like to move to the more powerful NVIDIA cards myself but there are just too many issues, just dont think NVIDIA put the time in to HTPC playback AMD does.

Asmodian

23rd March 2018, 17:34

Warner306

23rd March 2018, 18:03

That's what I always thought....

The Martian (2015) 4k HDR. The scene from 00:10.10 looks extremely different when played with Madvr passthrough with metadata, vs madvr passthrough without metadata and nvcp set to use default color settings and the OS HDR toggle on.

Does any else have this film and test as above?

K

Win10 (latest fast ring) and GTX 1080ti (latest driver)

It is not clear what you are comparing here...

With OS HDR on and madVR set to passthrough, the Windows method should be used. If select "send metadata," is the Nvidia API activated? Or is this another Windows method: full passthrough vs. partial passthrough? Doesn't the madVR OSD indicate this?

It would have to be determined what each setting is doing. I think the OS HDR switch has to be off for the Nvidia API to work. Most people are using the Nvidia API and checking send metadata, so this hasn't been brought up as an issue.

It has been stated that the Windows OS HDR is capable of producing a different image, but not how it does this.

Manni

23rd March 2018, 18:08

HDR has been working well for me, with Nvidia, since madVR implemented the Nvidia API. I created a 3DLUT with DisplayCal and calibrated and profiled with Calman, it was a bit shocking how bad the gamut coverage is in HDR mode. :(

Gamut coverage is far less important for color reproduction that saturation tracking and colour volume, so you want to check these too (Calman has two good layouts to measure colour volume and a few good layouts to measure linearity).

A display with "only" 90% gamut coverage can produce a more accurate picture than a display with 100% gamut coverage if the former has a better linearity and a higher color volume.

j82k

23rd March 2018, 19:21

That's what I always thought....

The Martian (2015) 4k HDR. The scene from 00:10.10 looks extremely different when played with Madvr passthrough with metadata, vs madvr passthrough without metadata and nvcp set to use default color settings and the OS HDR toggle on.

Does any else have this film and test as above?

K

Win10 (latest fast ring) and GTX 1080ti (latest driver)

I just tried this and I can't really see a difference between NV HDR and OS HDR, so it can't be extremely different.
I also tried some HDR clipping test pattern and they also looked identical between NV and OS HDR.

One thing I really like about OS HDR though is that the windows interface isn't blindingly bright.

Also on latest win10, latest nvidia drivers.

oldpainlesskodi

23rd March 2018, 20:43

I just tried this and I can't really see a difference between NV HDR and OS HDR, so it can't be extremely different.
I also tried some HDR clipping test pattern and they also looked identical between NV and OS HDR.

One thing I really like about OS HDR though is that the windows interface isn't blindingly bright.

Also on latest win10, latest nvidia drivers.

Odd, must be my set up then if you tried the Martian movie - With NV i see oversatuarted colours, whilst not showing all the gradients, and some ringing artifacts on the small stones on the martian land as it zooms in, but with the OS HDR I get done of those issues, and the image is far clearer/vibrant.

K

Asmodian

23rd March 2018, 21:46

Gamut coverage is far less important for color reproduction that saturation tracking and colour volume, so you want to check these too (Calman has two good layouts to measure colour volume and a few good layouts to measure linearity).

A display with "only" 90% gamut coverage can produce a more accurate picture than a display with 100% gamut coverage if the former has a better linearity and a higher color volume.

By gamut I meant color volume. The 2D gamut, or the primaries measured at 100% brightness, is not really the gamut as they are defined in the specs, even though we often look at them like that. Linearity is less important when using a 3DLUT.

It was the Calman reports that were depressing, the primaries are quite good so at a quick glance the coverage looks great but as soon as you measure the full 3D space it doesn't look so good. It is fine for viewing, HDR content looks great, but it is a strong reminder that the HDR specs are well outside what current consumer displays can do.

Warner306

23rd March 2018, 22:03

Odd, must be my set up then if you tried the Martian movie - With NV i see oversatuarted colours, whilst not showing all the gradients, and some ringing artifacts on the small stones on the martian land as it zooms in, but with the OS HDR I get done of those issues, and the image is far clearer/vibrant.

K

Did you turn off the OS HDR switch when using the Nvidia API?

Manni

23rd March 2018, 23:54

By gamut I meant color volume. The 2D gamut, or the primaries measured at 100% brightness, is not really the gamut as they are defined in the specs, even though we often look at them like that. Linearity is less important when using a 3DLUT.

It was the Calman reports that were depressing, the primaries are quite good so at a quick glance the coverage looks great but as soon as you measure the full 3D space it doesn't look so good. It is fine for viewing, HDR content looks great, but it is a strong reminder that the HDR specs are well outside what current consumer displays can do.

This is because you measured BT2020, which isn't relevant for 99.99% of current content. Most recent consumer displays cover at least 90% of DCI-P3, which is enough to reproduce well most current content.

Linearity is always important. It's easier to correct linearity with a 3D LUT, but you can still have linearity issues with a 3D LUT depending on the number of points used in the cube etc.

Regarding color volume, you want to measure DCI-P3 within BT-2020. BT2020 is only a container. Unless your display is really poor (or is a projector), it shouldn't be so bad if you measure the correct color volume (i.e. the effective one, not the theoretical one).

BatKnight

24th March 2018, 02:48

How come the one user has no problems with Windows OS HDR at 12-bits? It must be the tone mapping done by the display.
I've been doing further testing and managed to capture better screenshots (download and present them at 1:1 resolution) to exemplify my findings:

NV HDR triggered by madVR (output 10 bit) BANDING (http://www.helloutro.com/banding.png)
NV HDR triggered by madVR (output 8 bit) NO BANDING (http://www.helloutro.com/nobanding8bitNVHDR.png)
OS HDR enabled at Windows 10 (output 10 bit) NO BANDING (no screenshot available, too bright)

I've also noticed that with OS HDR disabled, and sending the metadata at madVR (NV HDR), the OSD only shows D3D11 fullscreen windowed (10 bit) when MPC-HC's seek bar isn't showing, otherwise it shows (8 bit). Keeps alternating from (10 bit) to (8 bit) when the mouse cursor moves down to the seek bar, for example.

But when with the HDR toggle enabled at the Win 10 display settings and madVR no sending metadata to the display (OS HDR), the OSD always shows D3D11 fullscreen windowed (10 bit), even when you move the cursor to the seek bar, or exits fullscreen. It ALWAYS shows (10 bit) at the OSD.

And, at last, D3D11's fullscreen exclusive always shows D3D11 exclusive (10 bit) and has NO BANDING.

So, to conclude, D3D11 exclusive NV HDR = Win10 OS HDR (both banding free)

Hope this helps further understanding why Windows' HDR toggle results in perfect picture without banding, and madVR sending the metadata to the display (NV HDR) on windowed mode doesn't.

foxyshadis

24th March 2018, 07:53

Jeesh, what is it with this site and image uploads? It's not a video, it's a simple tiny image. Anxious for you guys to look it over. I see other folks post and their image shows immediately? Please, someone explain what I've done wrong.

In manage attachments I uploaded a 900k .jpg direct from my PC to here and posted. It said pending approval.
I removed attachment and checked post to see that it was gone.
Uploaded .jpg to imgur.com, then copied link to here. It said pending again. Noticed it said 200k max.
I removed pending image again.
I cropped image down to nothing (162KB), uploaded from my PC to here and posted. Pending this one too.
Removed and checked post to see it was gone.
Uploaded 162KB to imgur.com, copied url, and posted. Says pending again.
Help?

Report your own post, say "approve plz" and it'll get approved. I don't think the 200k thing is a real limit.

Unfortunately vBulletin in all of its wisdom notifies mods of reports but does NOT notify us of pending approvals, so the only option is to wade into threads on the off chance someone might have attached something. There's the mod panel, but it's... lousy, at best. I check on this thread twice a week, but it's so active I miss a lot. Now that we all work way more than full time, the rules page really needs a Rule #0: Report your own post for prompt service.

oldpainlesskodi

24th March 2018, 08:36

Did you turn off the OS HDR switch when using the Nvidia API?

Yes

K

j82k

24th March 2018, 10:57

When using OS HDR (MadVR send metadata disabled), how does that even work? Does windows send the metadata to the display?

tyrindor

24th March 2018, 11:06

Warner306

24th March 2018, 11:13

When using OS HDR (MadVR send metadata disabled), how does that even work? Does windows send the metadata to the display?

I don't know. Windows is able to trigger HDR mode, but may do the tone or gamut mapping rather than the display.

Warner306

24th March 2018, 11:16

I've been doing further testing and managed to capture better screenshots (download and present them at 1:1 resolution) to exemplify my findings:

NV HDR triggered by madVR (output 10 bit) BANDING (http://www.helloutro.com/banding.png)
NV HDR triggered by madVR (output 8 bit) NO BANDING (http://www.helloutro.com/nobanding8bitNVHDR.png)
OS HDR enabled at Windows 10 (output 10 bit) NO BANDING (no screenshot available, too bright)

I've also noticed that with OS HDR disabled, and sending the metadata at madVR (NV HDR), the OSD only shows D3D11 fullscreen windowed (10 bit) when MPC-HC's seek bar isn't showing, otherwise it shows (8 bit). Keeps alternating from (10 bit) to (8 bit) when the mouse cursor moves down to the seek bar, for example.

But when with the HDR toggle enabled at the Win 10 display settings and madVR no sending metadata to the display (OS HDR), the OSD always shows D3D11 fullscreen windowed (10 bit), even when you move the cursor to the seek bar, or exits fullscreen. It ALWAYS shows (10 bit) at the OSD.

And, at last, D3D11's fullscreen exclusive always shows D3D11 exclusive (10 bit) and has NO BANDING.

So, to conclude, D3D11 exclusive NV HDR = Win10 OS HDR (both banding free)

Hope this helps further understanding why Windows' HDR toggle results in perfect picture without banding, and madVR sending the metadata to the display (NV HDR) on windowed mode doesn't.

On my computer display, I can't see a difference, but I believe you. Sounds like a strange technical bug. Good to know. FSE it is.

It's probably not the tone mapping that is the problem.

Manni

24th March 2018, 11:38

Of course. Its not like NVIDIA doesn't support 10-bit, they just only offer 8-bit and the highest format supported in the settings. My PC screen here which only supports 10-bit also makes NVIDIA offer that in the control panel.
So you could take the EDID from your screen as-is, and just remove the 12-bit modes, and it should instead offer you 10-bit options.

Good suggestion, but not as straightforward as this it seems.

I've asked HD Fury to create a couple of test EDIDs with 12bits disabled, so that the Vertex/Integral/Linker reports exactly the same EDID with 12bits disabled. I asked for the one I usually use (full capability as well as a few others). None of these resulted in 10bits appearing in the NV CP. I'm talking RGB 4:4:4 of course, we already have 8/10/12bits options in YCC 4:2:2.

Apparently its because there is no standard for VIC with RGB 4:4:4 10bits, so there is an element of luck regarding what your display is going to report. In fact I think 10bits wasn't supported at all in HDMI 1.4 for UHD resolutions, which might be one of the reasons why nVidia has limited 10bits options.

So it looks like we would need nVidia not to expect VICs to be able to force 10bits for all displays using the methodology you suggested. Or possibly find another way to alter the EDID reported by the display, but the Vertex/Integral is quite good at doing this so I don't think that's an option.

Don't ask me what VICs are, I have no idea, but you probably do. :)

oldpainlesskodi

24th March 2018, 11:52

I've been doing further testing and managed to capture better screenshots (download and present them at 1:1 resolution) to exemplify my findings:

NV HDR triggered by madVR (output 10 bit) BANDING (http://www.helloutro.com/banding.png)
NV HDR triggered by madVR (output 8 bit) NO BANDING (http://www.helloutro.com/nobanding8bitNVHDR.png)
OS HDR enabled at Windows 10 (output 10 bit) NO BANDING (no screenshot available, too bright)

I've also noticed that with OS HDR disabled, and sending the metadata at madVR (NV HDR), the OSD only shows D3D11 fullscreen windowed (10 bit) when MPC-HC's seek bar isn't showing, otherwise it shows (8 bit). Keeps alternating from (10 bit) to (8 bit) when the mouse cursor moves down to the seek bar, for example.

But when with the HDR toggle enabled at the Win 10 display settings and madVR no sending metadata to the display (OS HDR), the OSD always shows D3D11 fullscreen windowed (10 bit), even when you move the cursor to the seek bar, or exits fullscreen. It ALWAYS shows (10 bit) at the OSD.

And, at last, D3D11's fullscreen exclusive always shows D3D11 exclusive (10 bit) and has NO BANDING.

So, to conclude, D3D11 exclusive NV HDR = Win10 OS HDR (both banding free)

Hope this helps further understanding why Windows' HDR toggle results in perfect picture without banding, and madVR sending the metadata to the display (NV HDR) on windowed mode doesn't.

Odd again my end. I always use D3D11 FSE, and I see differences in tone mapping and clarity etc with the OS HDR trigger.

K

Update - I have narrowed the settings down that show the differences using the OS HDR toggle. Using Dsplayer Kodi (last build), if I set to use Full Screen in system settings, and set MadVR to either windowed or FSE, it looks the same as using the OS HDR toggle off and use MadVr to send metadata. However, if I set use FullScreen window in Dsplayer, and set Madvr to FSE, using the OS HDR toggle, Madvr OSD reports Windowed mode, and I get vastly improved clarity and tone mapping.

Perhaps if anyone else if using Kodi, they could validate my observations?

huhn

24th March 2018, 11:54

an EDID has nothing todo with luck...
https://docs.microsoft.com/en-us/windows-hardware/drivers/display/overriding-monitor-edids

Manni

24th March 2018, 11:59

an EDID has nothing todo with luck...
https://docs.microsoft.com/en-us/windows-hardware/drivers/display/overriding-monitor-edids

I know what an EDID is. I'm only saying this because Nev suggested that simply not reporting 12bits ability in the EDID would make the 10bits option appear in the NV CP. I assume he has tested this with other displays and that it worked.

I tested this with mine and it doesn't.

So instead of saying that Nev's suggestion doesn't work, I'm saying that there must be an element of luck that makes it work with some displays and not others.

ryrynz

24th March 2018, 12:18

http://i.imgur.com/VLzJqQfm.jpg (https://imgur.com/VLzJqQf)

huhn

24th March 2018, 13:52

he never said he uses an edid override. he just uses a screen with no 12 bit support. and why would you trust an external device to do such a thing.

a blind guess a displayport device. that can do 60Hz UHD 10 bit since ages.

the part of the edid with a bit deep information should be block 167.
my windows 10 doesn't care about the extend edid block so i couldn't bring it to override this block yet and it'S a freaking pain to do it "by hand" in windows 10...

167 is this block:

00 FF FF FF FF FF FF 00 34 A9 96 A2 01 01 01 01
00 19 01 03 80 80 48 78 0A DA FF A3 58 4A A2 29
17 49 4B 20 08 00 31 40 61 40 01 01 01 01 01 01
01 01 01 01 01 01 08 E8 00 30 F2 70 5A 80 B0 58
8A 00 BA 88 21 00 00 1E 02 3A 80 18 71 38 2D 40
58 2C 45 00 BA 88 21 00 00 1E 00 00 00 FC 00 50
61 6E 61 73 6F 6E 69 63 2D 54 56 0A 00 00 00 FD
00 17 3D 0F 88 3C 00 0A 20 20 20 20 20 20 01 21

02 03 4D F0 57 10 1F 05 14 20 21 22 04 13 03 12
07 16 5D 5E 5F 62 63 64 61 60 66 65 23 09 07 01
78 03 0C 00 10 00 38 3C 2F C8 8A 01 02 03 04 81
41 01 9C 06 16 08 00 18 00 67 D8 5D C4 01 78 80
03 E2 00 4B E3 05 1F 01 E4 0F 00 00 78 56 5E 00
A0 A0 A0 29 50 30 20 35 00 BA 88 21 00 00 1A 66
21 56 AA 51 00 1E 30 46 8F 33 00 BA 88 21 00 00
1E 00 00 00 00 00 00 00 00 00 00 00 00 00 00 85

nevcairiel

24th March 2018, 14:03

my windows 10 doesn't care about the extend edid block so i couldn't bring it to override this block yet and it'S a freaking pain to do it "by hand" in windows 10...

Just use CRU, disabling 12-bit is just a checkbox there, and it updates the EDID in the Windows registry for you.
If this works to make NVIDIA offer a 10-bit mode depends on a few unknown factors apparently. But then I don't really have a use for it because my Sony TV supports 12-bit input properly.

huhn

24th March 2018, 15:00

after giving up on it and reset everything.

i can now select 10 bit for 422.
420 is still limited to 12 bit and RGB too.

makes totally sense to me obviously. i broke something so whatever.

BatKnight

24th March 2018, 15:05

On my computer display, I can't see a difference, but I believe you. Sounds like a strange technical bug. Good to know. FSE it is.

It's probably not the tone mapping that is the problem.
I understand it's difficult to see the banding because the screenshots have the colors washed out. But I can assure you it's quite noticeable when displayed with the correct gamut.

D3D11 FSE is a no go for me, because everytime i right click or need to exit fullscreen the display flickers and goes black, and even sometimes goes out of range and I have to unplug and plug back the HDMI cable.
Even though I need to toggle the OS HDR at the display settings before watching an HDR movie, I still find it less anoying. (my opinion)

I'll wait for madshi's input on this. Maybe the NV HDR D3D11 windowed 10 bit's banding it's something that he could work on his side.

huhn

24th March 2018, 15:10

the is no banding on the image because the image are both dithered to 8 bit that simply how the screenshoot function works.

Warner306

24th March 2018, 16:47

Warner306

24th March 2018, 16:48

I understand it's difficult to see the banding because the screenshots have the colors washed out. But I can assure you it's quite noticeable when displayed with the correct gamut.

D3D11 FSE is a no go for me, because everytime i right click or need to exit fullscreen the display flickers and goes black, and even sometimes goes out of range and I have to unplug and plug back the HDMI cable.
Even though I need to toggle the OS HDR at the display settings before watching an HDR movie, I still find it less anoying. (my opinion)

I'll wait for madshi's input on this. Maybe the NV HDR D3D11 windowed 10 bit's banding it's something that he could work on his side.

You might be better of using 8-bit dithered until this issue is fixed. Your eyes probably won't notice the difference.

Warner306

24th March 2018, 16:49

Odd again my end. I always use D3D11 FSE, and I see differences in tone mapping and clarity etc with the OS HDR trigger.

K

Update - I have narrowed the settings down that show the differences using the OS HDR toggle. Using Dsplayer Kodi (last build), if I set to use Full Screen in system settings, and set MadVR to either windowed or FSE, it looks the same as using the OS HDR toggle off and use MadVr to send metadata. However, if I set use FullScreen window in Dsplayer, and set Madvr to FSE, using the OS HDR toggle, Madvr OSD reports Windowed mode, and I get vastly improved clarity and tone mapping.

Perhaps if anyone else if using Kodi, they could validate my observations?

I don't think I understand you. Setting the Kodi GUI to fullscreen windowed vs. true fullscreen changes the image? How? It makes the Nvidia API work or not work? What would the GUI have to do with the video player?

oldpainlesskodi

24th March 2018, 17:17

I don't think I understand you. Setting the Kodi GUI to fullscreen windowed vs. true fullscreen changes the image? How? It makes the Nvidia API work or not work? What would the GUI have to do with the video player?

Ok, thought I was pretty clear. Yes, using fullscreen windowed, using the OS HDR toggle with Madvr set to HDR passthrough but not send metadata (and thus bypass the Nvidia API HDR switching) produces a completely different image (for the better).

As to why? good question, hence my post.

I think you are on the Kodi forum, so perhaps you could try, as above, and see if you notice any difference on your setup?

K

brazen1

24th March 2018, 17:33

Can someone explain to me how to setup a profile to automatically enable 3D when a 3D movie is launched via kodi with MPC-BE as an external player? The MadVR's setting "automatically enable 3D" doesn't work, I must check the box manually prior to watching.

I've seen some people mention that this is possible but could never figure it out. For reference, I use 3D frame packed MKVs.

I posted the instructions a few pages back. The profile code is there. 3D has to be enabled globally, in display settings of the O/S. This is what madVR enables and disables as needed. It also has to be enabled in the NVidia control panel as needed. When you play an HDR title, stereoscopic is automatically disabled in the NVidia control panel. Next time you go to play 3D, it won't because it's disabled. So, it either has to be manually turned back on in the NVidia control panel or automated. My guide explains how to do this automation by using .bat files. It applies to all players. If you have resolution profiles with this madVR 3D setting, make certain the setting is on and adjusted for all the resolutions in the profile, not just the 3D one. ALL of them.

To add to the 8bit/10bit discussion, don't confuse FSE with windowed fullscreen. Double clicking the player window takes it out of FSE but is not the same thing as windowed fullscreen. This is always going to render 8bit. HDR is fine windowed or FSE. Makes no difference. 4:2:2 has an 10bit selection yes and would be wonderful except it's limited only, not full.

Nothing is going to trigger Windows HDR and advanced color switch on and off as needed. This is why we use madVR with private API's. They do.

In the future, Windows update is promising an HDR calibration tool. I imagine this means the HDR switch in Windows will be able to remain on all the time and allow us to calibrate the desktop to a reasonable usable view while providing calibration for video. I suppose drivers will need to update so that 8bit limit in Windows desktop with HDR will still allow native 10bit of HDR titles.

How any of this will affect using madVR private API's vs Windows on all the time HDR will probably be slight since our recent tests show there is little or no difference between O/S HDR or GPU HDR except currently one is manually turned on and the other is automated. In the future, the automation should be moot if the calibration ability is worth a damn. How madshi will react, if at all, is a mystery but the tests we are doing now should be fuel for his interest as it is ours.

I want to add, going by memory, there is some difference between O/S HDR and GPU HDR. One you might notice: Look in your panels brightness and contrast setting when selecting madVR to use the private API's and pass it through. Then look at the display settings when you use the O/S HDR. This may be why some claim big differences. GPU HDR passing through evokes the display to calibrate. O/S HDR isn't evoking the display to calibrate, rather producing a set look hence why windows desktop looks like junk. Metadata isn't in use. I could be wrong about some or all of this.

MADSHI! WHERE YOU IS?

foozoor

24th March 2018, 18:16

Madshi is dead... :devil:

EDIT:
He is on another forum... and posted a new test release.
http://madshi.net/madVRhdrTest3.rar

brazen1

24th March 2018, 18:18

Nah. I'm fairly certain, well no, positive he's lurking. And his main perception is: Until moral improves, the beatings will continue.

el Filou

24th March 2018, 18:22

Stop pestering madshi expecting ETAs for new releases or immediate reactions.

brazen1

24th March 2018, 18:33

Lighten up. No one in recent history has made any demands of the sort.

Asmodian

24th March 2018, 19:56

This is because you measured BT2020, which isn't relevant for 99.99% of current content. Most recent consumer displays cover at least 90% of DCI-P3, which is enough to reproduce well most current content.

No, I measured DCI-P3, of course my TV doesn't do BT.2020. The primaries are not anywhere close to BT.2020's either. Peak Luminescence Target: Relative, ST 2084 HDR (PQ), DCI P3, and D65 in Calman. I have also attached my measured color volumes, viewable with Calman's Color Volume Visualizer (https://calman.spectracal.com/CalMAN-Color-Volumizer.html).
HDR: Color Volume (http://www.mediafire.com/file/l6hf4o022ah5shn/HDR%20DCI-P3%200.05%201000.csv)
http://i1222.photobucket.com/albums/dd496/asmodian3/1a198cd1-8fb1-40ff-9e7e-523477fe9349_zps5dldj6pc.png

SDR (BT.1886): Color Volume (http://www.mediafire.com/file/sbhau6ybd5bfgm4/SDR%20DCI-P3.csv)
http://i1222.photobucket.com/albums/dd496/asmodian3/9c6fd4f2-30d4-472d-8b22-52d5b5920a50_zpsfvhwyin9.png

Linearity is always important. It's easier to correct linearity with a 3D LUT, but you can still have linearity issues with a 3D LUT depending on the number of points used in the cube etc.

Nothing a 3DLUT can do is going to help a 50% coverage volume but madVR's full resolution 3DLUT really can help non-linear displays look a lot better. Nothing you can do short of getting a new display actually fixes a badly non-linear display but a 65x65x65 3DLUT can be a big improvement. Argyllcms can even do a full 256x256x256 3DLUT, but I have never had that improve anything. I actually haven't used Calman with any significantly non-linear displays but I am sure it would help too.

I said linearity was less important, not that linearity was unimportant. ;)