madVR - high quality video renderer (GPU assisted)

[tp4tissue]

I know how to clamp a gamut, and I know how to pull in the corners manually.

It doesn't look good in HDR, and it looks even worse in SDR.

I like exaggerated outer edge colors. The only time stretching the gamut becomes a problem is if there is non-linear push towards the edge, which luckily is rarely the case, and most TV has at least 1 mode that spreads the coordinates evenly outward.

Most real scene colors and memory colors only occupy a tiny sliver of the rec709 space, and they're mostly so desaturated that even if pushed alot, they don't look so out of the ordinary.

[huhn]

This isn't true, you just need to learn how to use it.

You can get excellent results with FMM, using gamut Auto in SDR (with color 25) and custom BT2020 (with color 20), at least on the S90C. The CMS works fairly well as long as you find the right balance between different saturation levels and gives near reference results. I feel no need to use a 3D LUT in SDR.

then why is the fmm settings that are shared with custom gamut?
yes you can calibrate fmm just like any other mode but that doesn't make it any better.

the CMS clamps the gamut but works in native mode. you can literally change colors by changing the CR setting...

even low end rtings has totally trash reds in SDRpre and post calibration and they barely run any patches and calibrate to gamma 2.2...
The hdr or sdr gamut?

I don't think I fully understand the issue you're describing. I'm still 1310. :scared: I still have my own gamut bugs.

FMM is always garbage, always, I never use it. It's an incorrect pretense to clamp gamut to exact edges of rec709 or dci p3.
sdr i don't care about HDR. and my bug is not been reported by anyone else yet and just hope i'm the only person on earth having that. i can literally go and use native and i will get the same error as seen with bt 709 compared to game mode.

ohh you have 1310 you can do something very very funny with that firmware. take this image place it in the center unscaled go to FMM at 60 or lower refreshrate and pump the sharpness. plz tell me that doesn't need 1308 or older.

that's the greatest bug i have ever seen.

the gamma response out of the box is fine the gamut is a joke in SDR and i'm ignoring my gamut bug i have here.

there is also a sadly german test that shows that this TV my exhibit psychovisual error even if a 2nm spectro says it is white it will not look white to a human eye. the way this panel output's blue is not how reality is using blue the band is way to narrow.

[Manni]

then why is the fmm settings that are shared with custom gamut?
yes you can calibrate fmm just like any other mode but that doesn't make it any better.


It makes it better (along with movie mode, which behaves the same) because you can calibrate that mode closer to reference than any of the other modes, in both SDR 709 and HDR BT2020. the other modes track a lot worse and are basically not usable for a reference calibation. You only need to disable PC mode and game mode to access FMM with all the same settings and calibration tables as when using a non-PC source.

Plius if you use Calman (I do for manual calibration, I only use Colorspace when I need a 3D LUT) you can get excellent results with the 1D LUT and you can also use the DDC controls, which is infinitely better, faster (even with IP Control in the latest CM version) and more accurate than using the menus.

Now if you don't care about reference, or if you have a bug that no one else has, that's something else.

Anyway, all this is off topic here, better discussed in one of the Samsung thread.

[huhn]

it subsamples in gamma light. reference is out the window.

this literally will produce wrong colors just read a colored checkerboard pattern.
if it would subsample in linear light the colorimeter would be none the wiser.

[tp4tissue]

ohh you have 1310 you can do something very very funny with that firmware. take this image place it in the center unscaled go to FMM at 60 or lower refreshrate and pump the sharpness. plz tell me that doesn't need 1308 or older.

that's the greatest bug i have ever seen.

which image huhn?

[iakoboss7]

Hello, it seems that today the "madVRhdrMeasure205.zip" expired and as such i only see a black image.

sadly 206 has issues for me (flickering colors), is there a version without an expiration date?

[Siso]

Is there a possible way to save madvr's "increase or decrease source black level"? Save option only works for gamma. I'm on a pc monitor and 2.4 gamma looks dark, so I'm stuck with 2.2 gamma.

[huhn]

which image huhn?

https://madshi.net/madVR/ChromaRes.png

Hello, it seems that today the "madVRhdrMeasure205.zip" expired and as such i only see a black image.

sadly 206 has issues for me (flickering colors), is there a version without an expiration date?

none without major bugs.
112 doesn't time out and smooth motion works. i don't have it at hand. 113 will crash with smooth motion.

these version have the win 11 device page bug. if you only have one device you are usually not affected.

Is there a possible way to save madvr's "increase or decrease source black level"? Save option only works for gamma. I'm on a pc monitor and 2.4 gamma looks dark, so I'm stuck with 2.2 gamma.

file name should work:
blacklevel: blacklevel=-50 to +50
source: https://forum.doom9.org/showpost.php?p=1709814&postcount=3

these work on folders a folder is actually part of the file name.

[Siso]

file name should work:
blacklevel: blacklevel=-50 to +50
source: https://forum.doom9.org/showpost.php?p=1709814&postcount=3

these work on folders a folder is actually part of the file name.

Can you elaborate more please? I'm using decrease source black level -3.

[huhn]

i never use this.

try to add blacklevel=-3 to the file name with spaces around.

[Siso]

i never use this.

try to add blacklevel=-3 to the file name with spaces around.

A folder with that name blacklevel=-3?

[Manni]

it subsamples in gamma light. reference is out the window.

this literally will produce wrong colors just read a colored checkerboard pattern.
if it would subsample in linear light the colorimeter would be none the wiser.

I have no idea what you are talking about. Have you disabled game mode and PC mode?

[tp4tissue]

https://madshi.net/madVR/ChromaRes.png


What am I looking for huhn :rolleyes:, I already know it drops to 4:2:2 lower than 30hz. 30 60 120 work fine though.

I'm toggling the sharpness up and down in movie mode at 24hz ?

[huhn]

A folder with that name blacklevel=-3?
a folder the file is place in or the file in question
What am I looking for huhn :rolleyes:, I already know it drops to 4:2:2 lower than 30hz. 30 60 120 work fine though.

I'm toggling the sharpness up and down in movie mode at 24hz ?

try 60 if it is not just one side of the screen that version doesn't have the issue. i only saw that on the release version of my TV and only in fmm movie was fine.

I have no idea what you are talking about. Have you disabled game mode and PC mode?
you can't use FMM with it so when i tested it yes.

it is well known that scaling and subsampling will alter color incorrectly. it is also very easy to proof.

this is only a problem with a PC and video games because of RGB. if you subsample RGB you have to do it in linear light or colors will shift.
120 HZ

FMM can't do 4:4:4.
reset preset only change warm2: https://i.postimg.cc/BQV5jFBV/image.png
it does this wrong because of subsampling.
graphics:
reset preset warm2 colorspace auto: https://i.postimg.cc/1XRhT2g4/image.png
it does it correctly if you ignore that my screen is utterly broken. there shouldn't be any green but well...

everytime 2 different colors are next to each other the resulted subsampled pixel is objectively wrong linear light can mitigate this.

[Siso]

a folder the file is place in or the file in question

I want to use it globally, not to rename each file I play...

[huhn]

your files are in a folder so you know add that to the folder name.
if you stream that's it.

you can also go to device -> "your device name" -> properties -> custom levels... and rise black this should have a similar effect.

[Siso]

Rising the black level in custom levels, makes the blacks gray, I want to make the blacks more black.

[huhn]

yeah right there was a -.
that's the job of gamma but maybe your level are just wrong.
the black you see with 0 is as dark as it get's more is clipping.

[iakoboss7]

none without major bugs.
112 doesn't time out and smooth motion works. i don't have it at hand. 113 will crash with smooth motion.

these version have the win 11 device page bug. if you only have one device you are usually not affected.

thank you!

I tried 113 (and all the versions in https://www.videohelp.com/software/madVR/old-versions ) but none works (113 hangs mpc-hc, the rest black image).

what other options do i have to continue using madvr till 207 comes out (hopefully without the color flickering of 206).

edit: 206 works fine with Smooth Motion Disabled, but smooth motion is super important :/

[Sunspark]

You don't need smooth motion if you match the display refresh rate to the video refresh rate.

It doesn't have to be the 1x multiplier either. E.g. if you have a 120 Hz display, you can make a custom mode of 119.88 Hz and that's 5x 23.976 Hz.

I spent way too long yesterday playing with pixel clocks on my deck.

[iakoboss7]

You don't need smooth motion if you match the display refresh rate to the video refresh rate.

It doesn't have to be the 1x multiplier either. E.g. if you have a 120 Hz display, you can make a custom mode of 119.88 Hz and that's 5x 23.976 Hz.

I spent way too long yesterday playing with pixel clocks on my deck.

i would have to change refresh rates back and forth all the time then in my use case, as my monitor has a freesync range of 38-76.1hz, i could use 71.928 but that would create issues with my games (as it would be just outside of LFC thus killing any "freesync" bellow 38hz).

[Manni]

you can't use FMM with it so when i tested it yes.

it is well known that scaling and subsampling will alter color incorrectly. it is also very easy to proof.

this is only a problem with a PC and video games because of RGB. if you subsample RGB you have to do it in linear light or colors will shift.
120 HZ

FMM can't do 4:4:4.
reset preset only change warm2: https://i.postimg.cc/BQV5jFBV/image.png
it does this wrong because of subsampling.
graphics:
reset preset warm2 colorspace auto: https://i.postimg.cc/1XRhT2g4/image.png
it does it correctly if you ignore that my screen is utterly broken. there shouldn't be any green but well...

everytime 2 different colors are next to each other the resulted subsampled pixel is objectively wrong linear light can mitigate this.

You need to disable both game mode AND PC Mode.

Yes, it's a well known issue that if you use SDR there is a subsampling issue, I reported that a long time ago.

There is no chroma subsampling issue if you enable NV HDR, we've had that discussion already.

I use 4K100, 4K119 and 4K120 as a workaround for the motion bug in fw 14xx. With OS HDR enabled, there is no chroma subsampling issue with FMM.

I use JRVR to tonemap everything to HDR10 / BT2020, and the picture is very accurate with all content, so enabling OS HDR isn't a problem. I play mostly HDR content, so having the OS default to HDR 4K119 means that there is no HDMI resync 99% of the time.

I have no idea why you are getting such poor results with FMM. With the recommended settings, I get reference results for both SDR and HDR.

[huhn]

change the observer to ICtCp.
de2000 close to ignores to bright colors.

OS HDR is not doable with madVR windows will assume sRGB so i would need a 3D LUT just to remove that major error. a shader is most likely better suite but sRGB is a total mess.
HDR has the issue of to bright primaries for SDR blue barely changed but is now massively brighter. volume is usually ignored for reason beyond me.

[Manni]

change the observer to ICtCp.
de2000 close to ignores to bright colors.

OS HDR is not doable with madVR windows will assume sRGB so i would need a 3D LUT just to remove that major error. a shader is most likely better suite but sRGB is a total mess.
HDR has the issue of to bright primaries for SDR blue barely changed but is now massively brighter. volume is usually ignored for reason beyond me.

I always use dEITP (or de2000 JNDab) especially when working on gamma as it takes luminance more into account.

I don’t know about madVR, I stopped using it except as a TPG with build 113. Too many uncorrected bugs. I exclusively use JRVR now for video playback and there are no such issues when mapping SDR to HDR if using OS HDR.

[huhn]

but the screens issue isn't gamma it is primaries brightness CIE2000 doesn't care when a color is like 15 % to bright it will go completely nuts when your gamma is.

btw. i wanted to test fmm HDR: https://i.postimg.cc/k5DsxZ43/image.png oh boy.
maybe i shouldn't touch a TV i need a new mainboard i guess.

[Manni]

but the screens issue isn't gamma it is primaries brightness CIE2000 doesn't care when a color is like 15 % to bright it will go completely nuts when your gamma is.

btw. i wanted to test fmm HDR: https://i.postimg.cc/k5DsxZ43/image.png oh boy.
maybe i shouldn't touch a TV i need a new mainboard i guess.

As I said, I use de ITP, especially for gamma work. I also use it to check everything else. I have nothing like your errors after calibration (and I check at five levels of saturation and five levels of luminance), but I use FMM Auto Rec-709 (color 25) for SDR and FMM Custom BT2020 (color 20 with custom CMS) for HDR.

Yeah, get that set fixed, it doesn’t behave as it should.

[huhn]

just to be absolutely clear i'm currently in this state obviously that screen would be gone without results as this:
https://i.postimg.cc/QCtNFFwg/image.png
because game mode still works 100% as it should.

i will proof that later on AVS that the screen even through it looks ok using de2000 is objectively trash.

the screen is currently 3D LUT calibrated and there is no way in hell a 3D LUT could fix that DE2000 shows a de of 1.5 for red and ictcp shows 3 and it looks like a 3. the to bright primary red creates terrible results.

[flossy_cake]

Rising the black level in custom levels, makes the blacks gray, I want to make the blacks more black.

Madvr looks at the file path of the currently playing file and if it contains a string such as [blacklevel=-3] then it will automatically apply -3 clicks of blacklevel.

So all you need to do is add [blacklevel=-3] to the name of your top level folder in which all your media files are contained, cause then the path of every file now contains [blacklevel=-3].

Alternatively you can adjust the near-black gamma at the GPU gamma table which takes effect globally so it will apply to everything not just video files played through MadVR. See post on previous page, but you'll need AMD or Nvidia GPU dithering to keep it free of banding (the NVidia driver dithering is a little buggy sometimes doesn't apply on startup and needs a reboot, but I find increasing startup delay to 10 seconds in the config.ini seems to have resolved that...for now).

See the tips.txt file included with MadVR which lists all the available tags. I find them very useful for example correcting files which were encoded with the wrong YUV matrix I'll put eg. [matrix=601] for some streaming service versions of older shows cause the dullard dummies at Netflix or Peacock or whatever have no clue about colour encoding. And I've used the [blacklevel] tag for some older shows that have a baked in black pedestal. And of course you can use them to apply profiles too. This tagging system is one of the main reasons I require MadVR over other renderers, it just gives me so much flexibility, like if a season of DVD show is too soft I can just tag that season like [profile='adaptivesharpen low'].

[Manni]

just to be absolutely clear i'm currently in this state obviously that screen would be gone without results as this:
https://i.postimg.cc/QCtNFFwg/image.png
because game mode still works 100% as it should.

i will proof that later on AVS that the screen even through it looks ok using de2000 is objectively trash.

the screen is currently 3D LUT calibrated and there is no way in hell a 3D LUT could fix that DE2000 shows a de of 1.5 for red and ictcp shows 3 and it looks like a 3. the to bright primary red creates terrible results.

Yes, as I said AVS is a better place to discuss this, but there is something badly wrong in your system (settings, madTPG, meter, GPU, driver, whatever).

You posted a magenta saturation chart showing dE errors up to 27 in FMM:


FMM can't do 4:4:4.
reset preset only change warm2: https://i.postimg.cc/BQV5jFBV/image.png
it does this wrong because of subsampling.
graphics:
reset preset warm2 colorspace auto: https://i.postimg.cc/1XRhT2g4/image.png
it does it correctly if you ignore that my screen is utterly broken. there shouldn't be any green but well...


I have no idea why or how you're getting these results, but it has nothing to do with chroma subsampling with FMM.

I took linearity measurements with FMM in SDR (OS HDR disabled, so 4:2:2 chroma). Initial settings are those I recommend in my thread for SDR rec-709 (colorspace Auto/rec-709, color 25) after setting the white balance at 100% white to D65 and correcting gamma (BT1886=-1 baseline + gamma autocal with CM). See https://www.avsforum.com/threads/2023-samsung-77%E2%80%9D-s90c-4k-qd-oled-review-dolby-vision-support-recommended-settings-calibration-results-tips-and-comparison-with-jvc-nz8-projector.3302053/?post_id=63344903#post-63344903 for more details. Here are the results (worst case scenario, I didn't warm up the meter or panel, just switched on and measured):

dE2000: https://mega.nz/file/syFgVCgD#KIuwsdMcp5DPilgkY8yfo26Z1dAim-9RdwUr2lBDBVM

dEITP: https://mega.nz/file/djlVWYSb#drLwG32StzYRjFuA8rlLwCW6AXyW-lA0wFZhI2F1JdA

I could use the CMS to get red at 100% sat much closer, but this would wreck the other levels and make them undersaturated, so I don't really care about red at 100% sat, there is very little content there.

Nothing looks wrong.

I could add a 3D LUT (JRVR can use madVR 3D LUTs) but I don't feel the need to. I use other sources (non-HTPC) and they all look great with this SDR calibration. I don't use this calibration with the HTPC as I use JRVR with OS HDR to tonemap SDR to HDR BT2000 and get 4:4:4 chroma (at 4K119 to avoid the motion/cadence bug with f/w 14xx), but the results are equally excellent as my HDR calibration with FMM is very good too.

The maximum error I have for megenta is under 2 in dE2000 and under 6 in dE ITP.

So really I have no idea what your problem is with FMM, or how you can get a dE of 27 for magenta (doesn't matter if it's 2000 or ITP at that stage, I hope it's the latter though) for magenta at 100% sat. THIS IS NOT NORMAL, EVEN WITH 4:2:2!

Until you correct that, there is no point in discussing FMM. You're using the wrong settings or something is badly broken in your set.

[flossy_cake]

I have no idea why or how you're getting these results

It could be a regional thing, I had Panasonic OLED FZ950U version sold in Australia for very cheap during a sale and it had a massive chroma bug that looked shit on any graphics with blue on red background or cyan on orange background. It was noticeable at 3.5m viewing distance on any content, broadcast TV with STB sending YUV 4:4:4 and WiiU RGB 4:4:4 (like red of mario's cap against blue sky background). I took photos and sent it to Vincent's twitter account but he couldn't reproduce it. In the end I gave up and sold it on gumtree, such a weird thing to happen. Different regions do sometimes get different hardware revisions and we know the firmware is different per region too. The "localisation" team might have stuffed something up. One thing we always get with our onboard DVB-T tuners is massive blurring and low pass filtering cause that's what our AU localisation team thinks we want. The blurring is so bad you would really laugh if you saw it, some years it even got worse and it was already shockingly bad, they just decided to turn up the blur filter for the DVB-T tuner even harder.

[Manni]

It could be a regional thing, I had Panasonic OLED FZ950U version sold in Australia for very cheap during a sale and it had a massive chroma bug that looked shit on any graphics with blue on red background or cyan on orange background. It was noticeable at 3.5m viewing distance on any content, broadcast TV with STB sending YUV 4:4:4 and WiiU RGB 4:4:4 (like red of mario's cap against blue sky background). I took photos and sent it to Vincent's twitter account but he couldn't reproduce it. In the end I gave up and sold it on gumtree, such a weird thing to happen. Different regions do sometimes get different hardware revisions and we know the firmware is different per region too. The "localisation" team might have stuffed something up. One thing we always get with our onboard DVB-T tuners is massive blurring and low pass filtering cause that's what our AU localisation team thinks we want. The blurring is so bad you would really laugh if you saw it, some years it even got worse and it was already shockingly bad, they just decided to turn up the blur filter for the DVB-T tuner even harder.

That's not what Huhn's problem is. Look at his magenta tracking. It's completely wonky. The chroma subsampling thing is a red herring, I took all the measurements above in OS SDR, so 4:2:2 chroma. If there was such an issue with FMM in any country, it would have been reported in the owner's thread or settings thread on AVS, as there are very knowledgeable users from all over the world and we're all getting the same results. I'm in the UK, I assume Huhn is in Germany, I don't think there would be significant differences between our sets, and there is NO WAY FMM properly configured would measure like that, even on a HTPC with madTPG (which is what I use).

Huhn doesn't provide any FMM settings (especially colorspace) or confirm he's using my recommended settings, doesn't confirm that he's disabled BOTH game mode AND PC Mode, so who knows. All I know is that this isn't how a S90C with a properly configured FMM should measure, so he has to find out what's wrong in his setup.

Anyway, last post here, it's off topic unless his madTPG/madVR is wrongly configured and it should be discussed on AVS instead. He could use a pGen to rule out a HTPC issue specific to his set-up.

[flossy_cake]

The chroma subsampling thing is a red herring

Yeah I didn't mean to say that huhn's problem is necessarily related to chroma subsampling, only to give an anecdote of regional differences as a possible culprit.


there is NO WAY FMM properly configured would measure like that, even on a HTPC

I guess he should try outputting YCbCr 4:4:4 vs YCbCr 4:2:2 vs RGB 4:4:4 at full range vs limited range and confirm the issue persists in all those modes before moving onto the next step of perhaps trying a different GPU (onboard GPU might be convenient) and then finally a different OS (could just do a clean install of Windows 10 to a spare drive for testing purposes).

I took all the measurements above in OS SDR, so 4:2:2 chroma

Out of curiosity why do you use 4:2:2 and not 4:4:4? Aren't Windows graphical assets like icons and such all stored internally as 4:4:4?

I output YCbCr 4:4:4 but my Samsung plasma downscales internally to 4:2:2, it's not too bad, slightly annoying but not a problem at normal viewing distances.

[Manni]

Out of curiosity why do you use 4:2:2 and not 4:4:4? Aren't Windows graphical assets like icons and such all stored internally as 4:4:4?


I don't use 4:2:2. I use RGB (4:4:4) full, but in OS SDR the samsung downscales this to 4:2:2. This doesn't happen with OS HDR, where you get the full chroma back.

So I took all my measurements in OSD SDR (which shows 4:2:2 instead of 4:4:4) to show that the chroma bug had nothing to do with the horrific measurments on magenta posted by @huhn.

My normal setup is 4K119 RGB 4:4:4 with OS HDR, so I get:

- 119p refresh rate and full 4:4:4 for games.
- no HDMI sync at all when playing 23p video content
- JRVR automatically maps SDR rec-709 content to HDR BT2020
- There is no motion/cadence bug cause by the 14xx f/w on the Samsung, as motion processing is disabled at 100+ refresh rate.

[flossy_cake]

I don't use 4:2:2. I use RGB (4:4:4) full, but in OS SDR the samsung downscales this to 4:2:2.

What do you mean exactly by OS SDR? Operating System set to SDR mode?

And are you saying that in 2024 Samsung is still needlessly downscaling to 4:2:2 just like my 2013 Samsung plasma does?

[Manni]

What do you mean exactly by OS SDR? Operating System set to SDR mode?

And are you saying that in 2024 Samsung is still needlessly downscaling to 4:2:2 just like my 2013 Samsung plasma does?

OS SDR means that the display settings in Windows have HDR turned off.

My Samsung is a 2023 S90C, and yes in RGB 4:4:4 with an nVidia GPU it downscales to 4:2:2 if the OS is set to SDR (HDR off).

If you enable HDR (OS HDR on, i.e. HDR enabled in the display properties), you get full 4:4:4 chroma as expected, at least at 4K100+ with PC mode disabled and game mode disabled using FMM, which is what I'm using.

[huhn]

That's not what Huhn's problem is. Look at his magenta tracking. It's completely wonky. The chroma subsampling thing is a red herring, I took all the measurements above in OS SDR, so 4:2:2 chroma. If there was such an issue with FMM in any country, it would have been reported in the owner's thread or settings thread on AVS, as there are very knowledgeable users from all over the world and we're all getting the same results. I'm in the UK, I assume Huhn is in Germany, I don't think there would be significant differences between our sets, and there is NO WAY FMM properly configured would measure like that, even on a HTPC with madTPG (which is what I use).

Huhn doesn't provide any FMM settings (especially colorspace) or confirm he's using my recommended settings, doesn't confirm that he's disabled BOTH game mode AND PC Mode, so who knows. All I know is that this isn't how a S90C with a properly configured FMM should measure, so he has to find out what's wrong in his setup.

Anyway, last post here, it's off topic unless his madTPG/madVR is wrongly configured and it should be discussed on AVS instead. He could use a pGen to rule out a HTPC issue specific to his set-up.

while it is 100% that the FMM subsampling is a no brainer compared to other issue my system has.

with the magenta i was not measuring pattern i was measuring a mix of red and blue. ignoring the bug graphic work fine and the color was "correct" but fmm subsampling was creating a new darker color. just to show that subsampling actually changes colors and makes even if FMM would work not useful for me. FMM is ~10.6 nit graphic is ~23.6 nits. the DE is not fair i can not get a correct DE. i was literally measuring the 4:4:4 test image.

while i have not 100% ruled out a source issue.
the driver was DDU and the TV was factory reset.
the fact that game mode works makes it very unlikely to be a source issue.

thanks for this: https://mega.nz/file/djlVWYSb#drLwG32StzYRjFuA8rlLwCW6AXyW-lA0wFZhI2F1JdA
so this is not unique to my TV.
i encounter red 255 all the time. let's leave it at that.
What do you mean exactly by OS SDR? Operating System set to SDR mode?

And are you saying that in 2024 Samsung is still needlessly downscaling to 4:2:2 just like my 2013 Samsung plasma does?
all do.
for samsung you needed PC mode in the past now is a bit more complicated and there are modes that work without.
sony needs the preset graphic or game (use graphic trust me).
LG needs PC mode similar to samsung.

he get's 4:4:4 in FMM HDR not in SDR.
so he uses HDR to work around that FMM SDR is subsampling by using JRVR which can tone map SDR to HDR. i personally would not do that but i also never verified it to make a 100% proper statement.
we on madVR need a different mode to avoid subsampling.

madshi doesn't seem to care about this and really like the idea of tone mapping to SDR.

[Manni]

while it is 100% that the FMM subsampling is a no brainer compared to other issue my system has.

with the magenta i was not measuring pattern i was measuring a mix of red and blue. ignoring the bug graphic work fine and the color was "correct" but fmm subsampling was creating a new darker color. just to show that subsampling actually changes colors and makes even if FMM would work not useful for me. FMM is ~10.6 nit graphic is ~23.6 nits. the DE is not fair i can not get a correct DE. i was literally measuring the 4:4:4 test image.

while i have not 100% ruled out a source issue.
the driver was DDU and the TV was factory reset.
the fact that game mode works makes it very unlikely to be a source issue.

thanks for this: https://mega.nz/file/djlVWYSb#drLwG32StzYRjFuA8rlLwCW6AXyW-lA0wFZhI2F1JdA
so this is not unique to my TV.
i encounter red 255 all the time. let's leave it at that.

all do.
for samsung you needed PC mode in the past now is a bit more complicated and there are modes that work without.
sony needs the preset graphic or game (use graphic trust me).
LG needs PC mode similar to samsung.

he get's 4:4:4 in FMM HDR not in SDR.
so he uses HDR to work around that FMM SDR is subsampling by using JRVR which can tone map SDR to HDR. i personally would not do that but i also never verified it to make a 100% proper statement.
we on madVR need a different mode to avoid subsampling.

madshi doesn't seem to care about this and really like the idea of tone mapping to SDR.

You're not providing any setting or answering any of the questions I asked. If you reset the TV, the FMM settings are completely wrong. You need to use the recommended settings.

As I said, you can easily correct red at 100% sat with the CMS if you want. I don't see the point in doing this given that 1) it would wreck the lower saturation points, which are absolutely fine and 2) there is no issue with real content, given that there is almost never 100% saturated red in real content, and lower saturation points work fine. Again, zero issue when watching content with multiple sources, both HTPC and non-HTPC (satbox, Oppo 203 clone, Dune mediaplayer, ATV 4K, etc).

Anyway, you don't seem to want to resolve this, you're only making ridiculous statements, so I've corrected your misinformation and other people can see how "bad" it actually is when the source and TV are properly configured.

Good luck!

[huhn]

You're not providing any setting or answering any of the questions I asked. If you reset the TV, the FMM settings are completely wrong. You need to use the recommended settings.

As I said, you can easily correct red at 100% sat with the CMS if you want. I don't see the point in doing this given that 1) it would wreck the lower saturation points, which are absolutely fine and 2) there is no issue with real content, given that there is almost never 100% saturated red in real content, and lower saturation points work fine. Again, zero issue when watching content with multiple sources, both HTPC and non-HTPC (satbox, Oppo 203 clone, Dune mediaplayer, ATV 4K, etc).

Anyway, you don't seem to want to resolve this, you're only making ridiculous statements, so I've corrected your misinformation and other people can see how "bad" it actually is when the source and TV are properly configured.

Good luck!

first of all i through we where done about. fixing fmm over the inet is just disrespectful of other peoples time.

i provided information of the setting i used fmm was preset reset, warm2 and auto colorspace. this will give you a normal gamut but errors in color gamma and so.

at least on my TV i can only use graphic or entertain in PC mode or additionally game which now allows another 8+ preset no one ask about and cares about.

i will now go to 60 hz use FMM and dail the usually settings in and share just out of respect:
https://i.postimg.cc/sDPwBNpJ/image.png

in my absolute honest opinion:
a de00 of over 2 for red is bad.
the de for ictcp you show are hands down terrible. there is an brown in there with over 12.
again my opinion and most people will still say below 2 and de00 is reference it is not to me and that is also my problem.
i leave the red 255 matters to me a lot and leave it at that this is not worth fighting over.

this is also not to down talk these shared setting people get great gamma response out of this TV. and i can't share better ones with proper red blue and green and more...
and just to agree yes i could go in the CMS and get proper red 255 and i did but i messed everything else up and that why i think this TV is terrible.

i currently run game mode i don't want to use with a de00 of 1.5 for red and 3 for ictcp and still think the results are bad.

[Nicholas Steel]

A couple weeks after updating my display driver, MadVR stopped rendering video for every video I've tried (only audio plays). I use MadVR with MPC-BE and have already tried reinstalling MadVR as well as deleting its Settings file.

Windows 10 22H2. Was using Nvidia 528.49 driver, am now using 560.70.

Here is a copy of the error message that immediately appears when attempting playback of a video with MadVR selected: https://justpaste.it/b1569

I'm also operating under the assumption MadVR doesn't have some kinda kill code that breaks old versions in order to push people to stay up to date/not report issues for old versions. I'm sticking with MadVR 205 currently because 206 causes every other frame to be solid white for some of my video files.

[huhn]

he got you good there.
your madVR version timed out.
https://www.videohelp.com/software/madVR/old-versions
206 has broken smooth motion if that is currently active disable it before it is flickering like crazy and some people really can't handle that.
supposedly it works with d3d11 decode which breaks other stuff.

[Nicholas Steel]

That uh, that really sucks, so no way around that huh?

I could understand blocking say, more than one version older than current, but I don't understand why you'd block the 2nd latest just in case something like what I'm experiencing happens...

Thank you for pointing out Smooth Motion as the culprit, still though, being forced to update to broken version... that's worse than a lot of game DRM's & anti piracy measures.

[huhn]

indeed.
yes it does!

they have a life expectancy of x month i don't remember and because there are literally no new beta in a while there is a good chance 206 times out. if you could go back in time they would work again huhn out.

mpv smooth motion is not the same so i can not point you somewhere else.

[Sunspark]

Just match the refresh rate..

[huhn]

here you go: http://madshi.net/madVRhdrMeasure206b.zip
someone with 1 post random posted this.

this is madshi server should be fine...

[tp4tissue]

I don't use 4:2:2. I use RGB (4:4:4) full, but in OS SDR the samsung downscales this to 4:2:2. This doesn't happen with OS HDR, where you get the full chroma back.


Headscratch. What??

What do you mean it downscales to 422. In pc graphics mode 444 works for both hdr and sdr. :scared:

It only turns into 422 when you run 23/24 hz, where pc graphics mode is disabled.

[Manni]

Headscratch. What??

What do you mean it downscales to 422. In pc graphics mode 444 works for both hdr and sdr. :scared:

It only turns into 422 when you run 23/24 hz, where pc graphics mode is disabled.

I’m talking about FMM/Movie mode, not PC graphics mode.

I have other sources, I have zero interest in PC-only modes. Plus only FMM and Movie mode are supported by Calman autocal, which works very well for both 2-pt white balance and 20-pt gamma.

I don’t use SDR, I only use HDR, which has 4:4:4 chroma unlike SDR. JRVR tonemaps SDR to HDR with very good results

EDIT: Plus I just checked and PC Graphics also downscales to 4:2:2 in SDR (at 4K119 10bits RGB 4:4:4 with a 3090), at least according to madshi's chromares pattern. That's in 4K119, which is the default mode I use. If you select below 60hz, you do get 4:4:4 in SDR but it reverts to FMM automatically. This allows to have different calibrations for game and video, but doesn't apply to me as I use 100+ hz only as a workaround to the cadence/motion issue in Samsung f/w 14xx.

[ryrynz]

I'm also operating under the assumption MadVR doesn't have some kinda kill code that breaks old versions in order to push people to stay up to date/not report issues for old versions.

Dat assumption..

Happened to me as well, first time.. I thought my content was borked for a hot minute.

Get a black screen upon playback after pausing and a bunch of white flashing on 206 & 206b that I never had on 205. Smooth Motion wasn't active, but I've disabled it now to test. Looks like it fixed it.

[Manni]

While I was at it, and this is definitely my last post on that topic here, I also (re)measured PC Graphics. I had ruled it out as it's unusable in HDR for video content as it doesn't track PQ, it's way too bright over the whole scale, and this isn't correctable as even with gamma = -3 (the minimum) it's still way over targets, so this could only be corrected in the service menu (which I'm not interested in):
https://mega.nz/file/R6NFEBKa#NQ_K9ogzoma-9jYvTyWHNZwgJsSxBTnv2F38RBfWFnc

SDR (Auto/rec-709) doesn't have the slight oversaturation on red at 100% it has in FMM, but it's badly undersaturated so you have to pump color up to 31 to get decent tracking. Overall, the results are far worse than with FMM (I only measured dE2000, that's bad enough):
https://mega.nz/file/xul0GL4J#fo_63BxRd5e1GLNllmuGb3Xrp6KHtnlHogLpSdLDzgA

So I really don't see the attraction for PC Graphics mode, given that it also downscales chroma in SDR at 4K100+ and reverts to FMM anyway below 60hz. It's a bear to calibrate at 4K100+ unless you're happy to use the service menu or use a 3D LUT.

Over and out.

[flossy_cake]

Happened to me as well, first time.. I thought my content was borked for a hot minute.


Should at least put a text message on screen saying it's expired or something.

The non-beta final 0.92.17 is unlimited usage right? Right? :eek: :scared: :confused:

[strumf666]

It should be, but since it is so old you might have problems with HDR switching, and possibly other issues too.
I think madVRhdrMeasure170 doesn't have time limit and is much more recent than the 0.92.17
Displaymode switching seems to be broken on 206 and 206b at least on my setup (win11 6900xt with latest drivers and LG g3), I get invalid signal/no picture but if I disable it, then it works even if the display switches from SDR into HDR.

edit: 170 doesn't work for me, so it looks like it's time limited too, not sure where I saw that it isn't.
edit2: Looks like updating lav filters to the latest nightly fixed the madvr display mode switching... Works normally with LAVFilters-0.79.2-18 and madvr 206b in MPCBE 1.7.3 x64