madVR - high quality video renderer (GPU assisted)

[ryrynz]

First off determine whether madVR is even related to what you see by changing the renderer.

[Manni]

I only have one custom mode in the Nvidia Control panel set for 23Hz and I tried this driver hoping that I may not even need to set that..
I get a frame repeat every twelve minutes as opposed to a frame drop every three minutes with 385.28 when not using the custom mode.
However, once again like every other driver after this it won't add the custom mode if I change the drop down box to select manual timings, so back to 385.28 I go. Pathetic.

Ayy, I got it to stick by modifying 24Hz instead and added that to the list of resolutions madVR can switch to. Hurrah, now no frame drops/repeats with 398.11.

Are you using HDMI or DP? I thought the new f/w only improved DP, which I don't care about.

Also please could you detail what you are doing so that 24p applies to 23p?

When editing 23p with MadVR custom res, it creates a 24p custom mode in the nVidia CP that is applied with 23p content. That's the way it works here, with all drivers (at least with 4K23, but I think you're using 1080p).

I don't understand what you are doing/modifying, so please could you detail step by step?

Thanks!

[ryrynz]

I'm using HDMI. This is just the results of driver update not the DP firmware, I haven't applied that update.
What I did was via the Nvidia control panel, I haven't saved any custom modes with madVR because Nvidia's custom mode works so well.. when Nvidia's drivers save the config that is.

[nevcairiel]

Not all cards even require the fw update, my 1080Ti didn't, and to be clear, it only updates the UEFI part of the firmware, not the whole thing.

[Manni]

I'm using HDMI. This is just the results of driver update not the DP firmware, I haven't applied that update.
What I did was via the Nvidia control panel, I haven't saved any custom modes with madVR because Nvidia's custom mode works so well.. when Nvidia's drivers save the config that is.

OK thanks.

Not all cards even require the fw update, my 1080Ti didn't, and to be clear, it only updates the UEFI part of the firmware, not the whole thing.

Yes I guess it depends on the brand/model, my 1080Ti (EVGA) did need the fw update. I applied it even though I'm not using DP right now.

I don't think it should make any difference with HDMI, I think the reference to 10bits being handled by the driver is about DP, not HDMI. I haven't tested the 398.11 driver yet though.

I expect the video levels will still be borked, but I'm curious to see if it improves the 3D performance even further than 397.93, and also maybe the 8bits HDR bug will be fixed. These two bugs seem to be JVC only, as few seem to be affected.

[madjock]

No joy from me with latest Drivers 398.11.

Strangely I did a clean install, went in and checked and has been stated it sits at Full as defaults for colour, but I also noticed it had saved my custom 23.976 profile from the previous drivers.

Started a Video and it calmed down to 1 frame drop every 12 Minutes which is no good, but I noticed even though the composition said 23.976, the display was now higher than previous driver at that custom res and was now 23.9767 and hence the quicker frame repeats.

Deleted the custom res, and started the same video agaiin, and yes the composition is sat at 23.976, so I assume this is Nvidias fix and it looks like it ignores custom res well at least for me.

So created a custom res and like some other drivers it would not save either.

Re-Installed 390.77 and set it up as before and back to 1 frame repeat in hours. So no joy for me with them. :scared:

[jkauff]

No joy from me with latest Drivers 398.11.
This was the first driver install that didn't wipe out my madVR custom resolutions (not a clean install, which probably would have).

I've had the best success creating my custom resolutions in CRU, and then optimizing them in madVR. No "GPU rejected this mode" messages that way.

[Mano]

First off determine whether madVR is even related to what you see by changing the renderer.
I switched around the directshow setting and

madvr: tearing
Haali video render: tearing
Video mixing render 9 (windowed): tearing

Video mixing render 9 (renderless): no tearing
Enhanced video renderer: no tearing
System default: no tearing

VLC also shows tearing similar to madvr.

So where do you think the problem is?

[huhn]

what system are you using, please make a screen of the OSD and did you changed any settings in your GPU control panel.

my blind guess is an optimus system

[Manni]

Report on 398.11 (see sig for config)

The Good:
- Asio4all compatibility isn't broken anymore, so you can use it with REW etc :)
- 3D (1080p23FP) timings improved (same as 397.93, around 13-14min for a frame repeat vs every 3min with any of the former drivers)

The Bad:
- They have broken 12bits support entirely in custom res. Even if you create the custom res with say 4K30 12bit selected, the custom res will always be 8bits and won't let you select 12bits (as before). Former drivers didn't let you select 12bits, but the bit depth way greyed out and 12bits was in fact applied. There is no way around that. So not an issue if you want 8bits, but if you want 12bits that's a no go.
- JVC specific: the video levels are still borked, but we're back to 390.65 level, where you can actually get the correct level by setting GPU to Full, MadVR to 0-255 and the JVC to standard. No other combination works, including the correct one which should be to set MadVR to 16-235 for a display set to Standard video levels. I couldn't get good results with Enhanced (0-255) on the JVC either. But at least, you can get the correct levels, unlike 390.x, 391.x and 397.93 which are not usable with the JVCs.

The Ugly:
- JVC Specific: HDR in 8bits is still borked and produces the Magenta bug at all frame rates unless you disable HDR metadata in the HDR tab in MadVR. Pre 391.35 only produce the Magenta bug in HDR at 60fps.

Conclusion:

I didn't test the banding in 10bits in full screen windowed because 10bits is disabled with the current MadVR build.

I tried to create a custom RES using nVidia CP (no back porch accessible, so not good enough) and CRU (wouldn't accept my front porch of 1276). So although 3D is improves, Asio4All incompatibility is fixed, video levels are half borked but usable and the magenta bug is fixable by disabling HDR metadata, the lack of 12bits custom res is a no go. It also looks like the driver doesn't always save the custom modes, or reverts to 60p in the GUI when the actual frame rate is lower.

Big mess of a driver for JVC users, possible a good choice for those who don't need custom res or can get one with CRU.

As for me, back to 385.28, as the only downside there is a frame repeat every 3mn vs 13-14min in 3D...

[x7007]

I switched around the directshow setting and

madvr: tearing
Haali video render: tearing
Video mixing render 9 (windowed): tearing

Video mixing render 9 (renderless): no tearing
Enhanced video renderer: no tearing
System default: no tearing

VLC also shows tearing similar to madvr.

So where do you think the problem is?
Do you have Vsync Off in the NVCP ? what GPU do you have ?

[giulianoprs]

Hello everyone, how do you think my OSD? for now only fullHD, when I get the new 4k tv I'll try the HDR .thanks to everyone.

[NoTechi]

Report on 398.11 (see sig for config)

- JVC specific: the video levels are still borked, but we're back to 390.65 level, where you can actually get the correct level by setting GPU to Full, MadVR to 0-255 and the JVC to standard. No other combination works, including the correct one which should be to set MadVR to 16-235 for a display set to Standard video levels. I couldn't get good results with Enhanced (0-255) on the JVC either. But at least, you can get the correct levels, unlike 390.x, 391.x and 397.93 which are not usable with the JVCs.


Which chain are you using with 385.28. GPU full, MadVR 0-255 and JVC enhanced?

NoTechi

[Manni]

Which chain are you using with 385.28. GPU full, MadVR 0-255 and JVC enhanced?

NoTechi

Nope, GPU Full, MadVR 16-235 and JVC Standard.

[NoTechi]

Thx Manni :)

[Mano]

Do you have Vsync Off in the NVCP ? what GPU do you have ?

I am using sapphire R9 290.

Driver version 13.12 without the AMD Control centre.

p/s: also exclusive fullscreen mode seems to get rid of the tearing.

[huhn]

Hello everyone, how do you think my OSD? for now only fullHD, when I get the new 4k tv I'll try the HDR .thanks to everyone.

don't attach screen like this it need admin approval first.

@Mano

ok i try one more time what system of windows is used and can you make a screen of the OSD?

[Mano]

@Mano

ok i try one more time what system of windows is used and can you make a screen of the OSD?

Windows 7.

By OSD you mean this? https://i.imgur.com/ztaVlC5.png

[ryrynz]

Why is your screen using 1280x720? Any reason for using d3d9 without present several frames in advance?
Tried d3d11? Any apps running that could be possibly interfering with the display? Shut all non essential apps down.

[huhn]

did you disable aero?

[Mano]

Why is your screen using 1280x720? Any reason for using d3d9 without present several frames in advance?
Tried d3d11? Any apps running that could be possibly interfering with the display? Shut all non essential apps down.

my screen is 1440p, 1280x720 is probably the size of the anime file.

d3d9 is probably the default setting, how do i change that?.

did you disable aero?

Yes, i play games in borderless window mode (so i can run smaller resolution than 1440p, less GPU taxing) and don't want to introduce input lag due to vysnc.

[ryrynz]

You're not upscaling to 1440P, otherwise draw wouldn't say 1280x720 and you'd be luma upscaling on the OSD showing as Image > "Something"

rendering > general settings > use Direct3D 11 for presentation
You've been here a while, how do you not know the basics?

ENABLE VSYNC.

[huhn]

by disabling aero you disabled vsync so you will get tearing end of story

[giulianoprs]

sorry! a link to see my OSD to good the same? can you give me an opinion on my OSD status? now only for fullHD, thanks to all, hello.

https://imageshack.com/i/pnsK1Yj2j

[Mano]

You're not upscaling to 1440P, otherwise draw wouldn't say 1280x720 and you'd be luma upscaling on the OSD showing as Image > "Something"

rendering > general settings > use Direct3D 11 for presentation
You've been here a while, how do you not know the basics?

ENABLE VSYNC.

Strange, "use Direct3D 11 for presentation" is already select inside madvr setting but it is still showing D3D9 in the OSD.

Look like enabing aero/vsync removed the tearing.

p/s: Also after enabling aero, OSD changed from D3D9 to D3D11 in the OSD.

by disabling aero you disabled vsync so you will get tearing end of story

Thanks. Is there any way to make run vsync for MPC-HC alone?

Because i notice for output like Video Mixing Renderer 9 (renderless), even though aero is disabled, it still doesn't trigger tearing.

[huhn]

yes vsync is there to stop tearing that what it does.

d3d11 needs aero.

if you don't want to use aero you have to use FSE or overlay what amd doesn't support or you can get tearing the rest is luck.

[Neo-XP]

Hi there,

I encounter a color problem when converting HDR to SDR.

SDR BR :

https://s15.postimg.cc/g4p294nkn/image.png (https://postimg.cc/image/g4p294nkn/)

HDR UHD BR (downscaled, 200 nits) with tone mapping curve set to "BT.2390" :

https://s15.postimg.cc/a3rdc2o3r/UHD_-_BT._2390.png (https://postimg.cc/image/a3rdc2o3r/)

HDR UHD BR (downscaled, 200 nits) with tone mapping curve set to "clipping" :

https://s15.postimg.cc/hjqmxvjiv/UHD_-_clipping.png (https://postimg.cc/image/hjqmxvjiv/)

When the tone mapping curve is set to "BT.2390", red color becomes violet/magenta, but details in highlights are preserved.
When the tone mapping curve is set to "clipping", colors are closer to HDR (the red jacket is in a beautiful red), but details in highlights are lost.

Both options are not right. Why can't we have the best of both worlds technically?
Is there any way to improve the HDR to SDR conversion with the "arve custom curve" option? How to use it?

:thanks:

EDIT : It is a bug I guess, see here : https://forum.doom9.org/showthread.php?p=1843997#post1843997

[Betroz]

ENABLE VSYNC.

Enable VSYNC in nVidia controlpanel when using MadVr? If so, that was new to me too :D

[jkauff]

Enable VSYNC in nVidia controlpanel when using MadVr? If so, that was new to me too :D
Necessary if you have "present a frame for every VSync" selected under "use Direct3D11 for presentation" under Rendering/General Settings.

[huhn]

leave the nvidia settings as they are.

[Warner306]

Both options are not right. Why can't we have the best of both worlds technically?
Is there any way to improve the HDR to SDR conversion with the "arve custom curve" option? How to use it?

:thanks:

It's pretty hard to compare the original 1080p Blu-ray to the UHD version. In that shot, it looks like most of the information is brighter than 100 nits, so the entire color balance would be different. It is hard to determine which shade of red is the correct one between clipping and BT.2390, but it does look like a slight hue shift with BT.2390. It depends on the shade of red used in the original grade, and it's impossible to know exactly what shade was used. The clipping shot has so many pixels that are hue shifted or producing noise that's hard to see the overall color balance. Red must also be clipping, so it could also have a hue shift in the clipping shot, but it doesn't look that noisy. It depends on whether that shirt is brighter than 200 nits. If it isn't, then the tone mapping is wrong.

Overall, BT.2390 looks the best, but maybe the shirt is a little off. It is exactly the type of thing madshi is working on at the moment.

Are you using dumb or complex scientific gamut mapping, and does changing these options make any difference with the shirt? It would be helpful to know if this is influenced by BT.2390 alone or gamut mapping done after BT.2390.

[Neo-XP]

It's pretty hard to compare the original 1080p Blu-ray to the UHD version.

That is true, but I also compare the UHD version on my UHD TV to the SDR conversion, and with BT.2390 tone mapping curve, the colors are off (most noticeable on reds).
The 1080p BR colors are closer to HDR than the SDR conversion done with BT.2390 in this example.

Are you using dumb or complex scientific gamut mapping, and does changing these options make any difference with the shirt? It would be helpful to know if this is influenced by BT.2390 alone or gamut mapping done after BT.2390.

After testing all modes, I am using "dump mode - convert gamut late", because complex scientific modes produce weird colors (orange faces for instance).
Changing the gamut mapping algorithm does not change the shirt color in this case.

dumb mode - convert gamut late (BT.2390):

https://s15.postimg.cc/b2zap8hkn/dumb_mode_-_convert_gamut_late.png (https://postimg.cc/image/b2zap8hkn/)

complex scientific mode - RGB clip (BT.2390):

https://s15.postimg.cc/qbp82zvtj/complex_scientific_mode_-_RGB_clip.png (https://postimg.cc/image/qbp82zvtj/)

[madjock]

It's pretty hard to compare the original 1080p Blu-ray to the UHD version. In that shot, it looks like most of the information is brighter than 100 nits, so the entire color balance would be different. It is hard to determine which shade of red is the correct one between clipping and BT.2390, but it does look like a slight hue shift with BT.2390. It depends on the shade of red used in the original grade, and it's impossible to know exactly what shade was used. The clipping shot has so many pixels that are hue shifted or producing noise that's hard to see the overall color balance. Red must also be clipping, so it could also have a hue shift in the clipping shot, but it doesn't look that noisy. It depends on whether that shirt is brighter than 200 nits. If it isn't, then the tone mapping is wrong.

Overall, BT.2390 looks the best, but maybe the shirt is a little off. It is exactly the type of thing madshi is working on at the moment.

Are you using dumb or complex scientific gamut mapping, and does changing these options make any difference with the shirt? It would be helpful to know if this is influenced by BT.2390 alone or gamut mapping done after BT.2390.

Is this not why you can adjust saturation and lumination reduction in the other extra drop down boxes, I went for complex scientific mode - RGB Clip and 30% and 70 %, only as I saw it in another forum.

Still not sure if comparisons should be made against a Bluray version, all the reviews I have read are that UHDs compared to BRs, the main thing they usually mention is the colours being different, now I know we dont have an HDR display, but if you do get the UHD to look like the BR, whats the point ?

[Neo-XP]

Ok, forget it, there are bugs in madVR I guess, look at this.

In BT.2390 - dump mode - convert gamut late:

https://s15.postimg.cc/cahtv2uh3/convert_gamut_late.png (https://postimg.cc/image/cahtv2uh3/)

In BT.2390 - dump mode - convert gamut late + "restore details in compressed highlights":

https://s15.postimg.cc/k38hn1dav/convert_gamut_late_restore_highlights.png (https://postimg.cc/image/k38hn1dav/)

Is it normal that the color of the shirt changes like that?!

[madjock]

Ok, forget it, there are bugs in madVR I guess, look at this.

In BT.2390 - dump mode - convert gamut late:

https://s15.postimg.cc/cahtv2uh3/convert_gamut_late.png (https://postimg.cc/image/cahtv2uh3/)

In BT.2390 - dump mode - convert gamut late + "restore details in compressed highlights":

https://s15.postimg.cc/k38hn1dav/convert_gamut_late_restore_highlights.png (https://postimg.cc/image/k38hn1dav/)

Is it normal that the color of the shirt changes like that?!


To be honest the more options that are in HDR->SDR the more confused I get. For example here is a review of La La Land which I guessed the stills were from.

https://ultrahd.highdefdigest.com/43176/lalalandultrahdbluray.html

https://www.avforums.com/review/la-la-land-4k-ultra-hd-blu-ray-review.13608

http://www.blu-ray.com/movies/La-La-Land-4K-Blu-ray/170219/

So for me, how does it deal with all this extra colour and changes from Bluray to UHD, does it flatten them all and have a more simple setup if possible to take it back to in comparison a blandness if the difference is so great.

Unsure, I think when there was less options I was strangely happier, or more ignorant, now I don't know what to pick, and I think it changes movie to movie.

[Neo-XP]

Apparently, madshi is still testing these conversions to find the best result. For instance, I guess the "clipping" option is there for testing purposes only. If there are some useless options left then, I am sure he will remove them.

@madshi Do you confirm there is a bug when using any dumb mode and enabling "restore details in compressed highlights"?
The strange thing is that I get almost the same wrong colors with this option than with the "complex scientific mode - RGB clip" algorithm.

Dumb mode - convert gamut late:

https://s15.postimg.cc/5gc4j86o7/dumb_mode_-_convert_gamut_late.png (https://postimg.cc/image/5gc4j86o7/)

Dumb mode - convert gamut late + restore details in compressed highlights:

https://s15.postimg.cc/y5z0fvidz/dumb_mode_-_convert_gamut_late_restore_details_in_compressed_h.png (https://postimg.cc/image/y5z0fvidz/)

Complex scientific mode - RGB clip:

https://s15.postimg.cc/7kwhkasvb/complex_scientific_mode_-_RGB_clip.png (https://postimg.cc/image/7kwhkasvb/)

Is it possible to have a quick fix for this if the next version is not ready yet?

[madjock]

Apparently, madshi is still testing these conversions to find the best result. For instance, I guess the "clipping" option is there for testing purposes only. If there are some useless options left then, I am sure he will remove them.

@madshi Do you confirm there is a bug when using any dumb mode and enabling "restore details in compressed highlights"?
The strange thing is that I get almost the same wrong colors with this option than with the "complex scientific mode - RGB clip" algorithm.

Dumb mode - convert gamut late:

https://s15.postimg.cc/5gc4j86o7/dumb_mode_-_convert_gamut_late.png (https://postimg.cc/image/5gc4j86o7/)

Dumb mode - convert gamut late + restore details in compressed highlights:

https://s15.postimg.cc/y5z0fvidz/dumb_mode_-_convert_gamut_late_restore_details_in_compressed_h.png (https://postimg.cc/image/y5z0fvidz/)

Complex scientific mode - RGB clip:

https://s15.postimg.cc/7kwhkasvb/complex_scientific_mode_-_RGB_clip.png (https://postimg.cc/image/7kwhkasvb/)

Is it possible to have a quick fix for this if the next version is not ready yet?

Strangely more seems to be talked about for projectors than here, but you cant comment on those threads as you get told to post in this one where no one really talks about the options :) I know we do not have projectors but an option is an option surely ? :D

http://www.avsforum.com/forum/24-digital-hi-end-projectors-3-000-usd-msrp/2954506-improving-madvr-hdr-sdr-mapping-projector-26.html

[Asmodian]

I think it changes movie to movie.

This is true. :(

When there is no proper standard commercial releases do it in different ways and HDR hasn't been properly standardized yet. There are too many "HDR" displays that are not really capable of HDR but still want to claim to display it. HDR mastering optimized for them can be pretty different from HDR mastering actually optimized for a display with a 4K peak white.

Also, I think projectors tend to see less difference between SDR and HDR because even HDR usually doesn't use brightness levels higher than those used during SDR. If there is no change in peak white HDR and SDR are effectively the same thing, as far as the display is concerned.

[Manni]

Also, I think projectors tend to see less difference between SDR and HDR because even HDR usually doesn't use brightness levels higher than those used during SDR. If there is no change in peak white HDR and SDR are effectively the same thing, as far as the display is concerned.

This isn't always true, it depends on the screen size/gain. A projector is calibrated to 50nits for reference white in a dedicated room for SDR. With HDR, I get up to 210nits in high lamp with HDR, with diffuse white in the content at 100nits and on screen at 25nits, that's 4-8 times more brightness for the highlights. It's a very visible difference if the projector is properly set-up and calibrated.

However, my screen is only 88" diag 16/9, so people with a larger screen and the same projector will get less dynamic range unless they go for a high brightness laser or similar.

Overall, if you have the adequate projector for your screen size, HDR does provide a very significantly increased dynamic range. It's just that most people don't take HDR (or 3D) into account when they chose the PJ/screen/projector, and they end up with a too dim picture. In that case they often have 50nits (or less!) for both SDR and HDR, and in that case HDR makes no sense, except to get the immersive audio.

Don't forget that bluray is mastered to 100nits but watched at 50nits in a dedicated room.

HDR is mastered to 100-4000nits (depending on the title), but with diffuse white around 100nits in the content displayed at 25-50nits on screen, it can be watched with great effect if you have at least 100-120nits peak white. You certainly don't need more than 300-400nits in a dedicated room to have more brightness than you can cope with in very bright scenes. An OLED with 600nits peak white in HDR is borderline too bright in a dedicated room with no ambient light.

[Asmodian]

Thanks for the info. I am surprised people calibrate to 50 nits for SDR peak white if the projector can do 100 nits, even in a completely dark room. Maybe 80 nits but 50 seems obviously too dim to me. Diffuse white at 25 nits on screen would be very dim with a lot of the HDR content I have, at least as viewed on my OLED. I use 100 nits peak white for SDR, and diffuse white in HDR, in a completely dark room. I picked it over 80 and 120 nits after testing for a few days at each.

Was 50 nits chosen as a minimum reasonable brightness or is that actually thought ideal even if a higher brightness is possible?

[Warner306]

Dumb mode - convert gamut late:

https://s15.postimg.cc/5gc4j86o7/dumb_mode_-_convert_gamut_late.png (https://postimg.cc/image/5gc4j86o7/)

Dumb mode - convert gamut late + restore details in compressed highlights:

https://s15.postimg.cc/y5z0fvidz/dumb_mode_-_convert_gamut_late_restore_details_in_compressed_h.png (https://postimg.cc/image/y5z0fvidz/)

Complex scientific mode - RGB clip:

https://s15.postimg.cc/7kwhkasvb/complex_scientific_mode_-_RGB_clip.png (https://postimg.cc/image/7kwhkasvb/)

Is it possible to have a quick fix for this if the next version is not ready yet?

I am assuming you are outputting both at BT.2020? If the HDR -> SDR is BT.709, that could throw things off.

That second last post definitely looks like a bug.

I'm still at a loss to determine if dumb or complex scientific is more accurate. The complex scientific shot doesn't look that bad, either. It is promising when your HDR display is close to the 1080p Blu-ray, but there is a fair to good chance it is also using dumb gamut mapping. dumb will destaturte more than scientific, but it consistently changes the hue in almost random ways, so it is not very precise or predictable in its correction. Scientific has maintained greater saturation and the warmer colors might have been intended, or it might not. Or, the person grading the 4K master intended dumb gamut mapping to correct when necessary. Doesn't seem likely. It's hard to tell.

That is a good example. I would compare dumb mode in madVR to your HDR display and see if you can spot any differences.

[COOLak]

Hi all. Basically I've been using MadVR for years, with very modest settings that have always suited my needs and never caused issues. Now I encountered a video that's 1080p@30fps, and for some reason it's like a slideshow. Almost unmoving in fullscreen and still choppy and laggy in the windowed mode. My TV is also 1080p and supports 30fps, to which it changes automatically via MPC-HC (not sure why, but this option never worked with MadVR for me, so I'm using MPC-HC capabilities for that). My video card is Zotac GeForce GTX 1050ti, the OS is Windows 7 x64. I'm using the conjunction of MPC-HC, LAV Filters, and MadVR. The CUDA mechanism is used. If I select the default MPC-HC renderer, everything is fine, but with MadVR it's just how it is, completely unwatchable. I don't want to disable MadVR solely to be able to watch some select videos, so I'm asking for your help in configuring it properly. Considering that the video resolution matches my screen, it's weird, because MadVR shouldn't neither upscale nor downscale anything in this case, and I don't know why it's causing this video to lag. I'm attaching my settings.bin and the data from MediaInfo about this video. Please take a look. Thanks.

[XTrojan]

Small question, my TV is calibrated to Rec709 2.2 which I have set in MadVR.

Sometimes I happen to watch 4k, which is rec2020. Will MadVR adapt to 4k Rec2020 or am I forced to change the tv settings back and forth everytime?

[Asmodian]

If you have your display set to "this display is already calibrated" at BT.709 + pure power 2.2 then madVR will covert BT.2020 content to be correct on your TV, without changing any settings in your TV.

Note: any colors outside the BT.709 gamut will be clipped, you cannot display all of BT.2020 on a BT.709 display.

[XTrojan]

If you have your display set to "this display is already calibrated" at BT.709 + pure power 2.2 then madVR will covert BT.2020 content to be correct on your TV, without changing any settings in your TV.

Note: any colors outside the BT.709 gamut will be clipped, you cannot display all of BT.2020 on a BT.709 display.

It's a JS9000 I believe it can display around 65% of BT2020 colors.

But since it's configuered for BT709 you're saying the colors outside that will be clipped?

Is it better to just change the color space to "native" on the TV? Is there any point "calibrating" for Rec2020, since you can't go outside the color space of BT2020, or well no TV can so far, so it's better to just put it to "native"?

All in all, getting correct color for HDR and BT2020 seems very confusing...

[ryrynz]

I don't know why it's causing this video to lag. I'm attaching my settings.bin and the data from MediaInfo about this video. Please take a look. Thanks.

Screenshot your OSD plz.

[jmone]

I can create the following Custom Resolution which works better than the default for dropping / repeating frames. The issue is that it seems to only allow RGB 8-Bit colour space and this causes issues on my JVC PJ (Purple push). I need to set YCbCr 4:2:2 @ 10-Bit for correct colour and while I can do this using the STD resolutions, the NVidia control panel only allows RGB 8-Bit when using the Custom Res.

Does anyone know if it is possible to have a Cusom Res for timing with YCbCr 4:2:2 @ 10-Bit ?

Thanks
Nathan

https://jmone.org/nextcloud/index.php/s/trzDacfPC8Bm7di/preview

[Manni]

Thanks for the info. I am surprised people calibrate to 50 nits for SDR peak white if the projector can do 100 nits, even in a completely dark room. Maybe 80 nits but 50 seems obviously too dim to me. Diffuse white at 25 nits on screen would be very dim with a lot of the HDR content I have, at least as viewed on my OLED. I use 100 nits peak white for SDR, and diffuse white in HDR, in a completely dark room. I picked it over 80 and 120 nits after testing for a few days at each.

Was 50 nits chosen as a minimum reasonable brightness or is that actually thought ideal even if a higher brightness is possible?

50nits for reference white for SDR isn't preference, it's reference. That's the cinema reference (47nits), which is translated to home cinema, even thought the content itself is mastered to 100nits. Some prefer a brighter picture, and there's nothing wrong with that, but reference in a home cinema with a projector in a dedicated room with no ambient light (you were talking about projectors) is 50nits. If you have any ambient light, then you'll most likely need more than 50nits. I can tolerate up to 60-65nits for bluray, anything above that is not comfortable (to me). I know some people who use 80nits or more with a projector, but usually they don't have a bat cave and tend to have some lights on. Also with a projector, opening the iris to raise peak brightness also raises black levels and lowers on/off contrast, unless you use a dynamic iris. You don't have this issue with OLED, so it might be less of a compromise to target a higher reference white with a flat panel.

I have no idea about your OLED. It probably has at least 500 to 600nits, so chosing 100nits for reference white makes sense if it's comfortable to your eyes as you still have some headroom for the highlights. It doesn't make sense with a projector that has 100-120nit peak white as you have no headroom for highlights if you set diffuse white to 100nits (apart from the fact that it would be blinding, at least in my room). The most I would want for dark scenes is 100nits in content = 50nits on my screen, or a brightness factor of 2. But I wouldn't be able to use this for the whole film unless the brightness factor is adjusted dynamically. Apart from the excessive brightness, there wouldn't be enough headroom for highlights in most cases (with titles with content up to and above 4000nits) and you'd lose a lot of detail.

Yes 25nits reference white will be too dim for quite a few titles, yet anything above that will be too bright for others. That's why I said 25-50nits (on screen) for diffuse white in HDR. This is also why I use three different custom gamma curves in my JVC and MadVR in passthrough mode with the Vertex selecting the most appropriate calibration according to the HDR metadata, until MadVR can either adjust its conversion to each title or provide a variable to select a different profile according to the HDR metadata (not just Max mastering brightness). We'll get there in the end, but for now Madshi is working on the tone-mapping itself. We've agreed to put dynamic conversion on the back burner for now until the algo is set (and the options simplified). So it will take a little while.

With a peak white of 120nits in HDR (normal lamp with 1200 hours on the bulb, I get around 175nits in high lamp), I use a target brightness of 480nits in MadVR, which means a diffuse white of 25. That's way too dim for some titles/scenes (that's one of the issues with MadVR's current implementation with projectors), but it's the best "one-size-fits-all" compromise for the average title, and I'm not blinded with bright titles. :)

[XTrojan]

Is it normal for composition rate to be 60Hz at Exclusive FullScreen mode when the display is at 23.976?

Whenever I switch back to windowed or normal fullscreen composition is 23.976 (normal)

Edit: It doesn't matter if it's D3D 9 or 11, whenever FSE is enabled it says composition rate is 60.000 Hz even if the display is at 23.976 and the content as well, should one be worried?

[huhn]

what windows version is used and is it a dual screen setup?