MadVR in use with LG OLED Thread

[huhn]

forget the vega 64 the 5700 is in the same price range and the same performance range.

RX 5700 driver support is at best broken at the moment.

madshi recommends turing card he didn't do it here but he did it at other places before far more then one time.

[Gadgety]

Thank you for the heads up, huhn! Wondering if it might be so that the MadVR Pro offering that I've since discovered may utilize the RTX specific benefits.

[chros]

I got LG 55EG920V a couple years ago only to watch movies. In the store the TV was demoed with a 4k opera, and the image was breathtaking, and it looked like HDR. All reviews I saw highlighted the possibility for HDR images. Quite a while late I learned that the HDR was only possible when feeding the images via a USB stick (!) not via HDMI, and that it was caused by a software flaw that LG had no intention of correcting. HDMI input gets overridden, I understand.
Are you sure that it is the case? Did LG do it with a firmware upgrade?
Strange, I quickly checked this forum (https://www.avforums.com/threads/lg-55eg920v-4k-hdr-oled.1951100/page-79) and there are guys who used HDR via PC (meaning using HDMI).

I've got two questions, first, does this influence the usage with MadVR?
You want to output HDR to the TV with madvr: it looks way better (https://www.avsforum.com/forum/37-video-processors/3053602-madvr-envy-anticipation-thread-46.html#post58580332) than SDR converted image (with madvr tonemapping).
So, try to sort this out, either with this very TV or (if you can afford) upgrade to a newer one. :)

Would you know, this being a 55 inch TV, any point in doing NGU Very high, or is there no discernible difference between that and NGU high? I have a lot of 720 material.
Id' say there's not much difference on an 55" screen (especially if you view it from ~3m). E.g. I quadroupling with NGU AA Low and downscaling ssimd with 720p and it looks good.

It'd be really interesting to see image comparisons for various chroma levels from ngu aa medium and going down the range, just to see what I can expect (and ultimately what kind of card to go for). Perhaps such a comparison already exists?
Take a look at this thread (https://forum.doom9.org/showthread.php?t=175429) to view some "recent" image comparisons (including chroma upscaling as well).

[Gadgety]

Thank you, chros, both for the links and the replies!

As for "Are you sure that it is the case? Did LG do it with a firmware upgrade?"

I have never seen a confirmation that LG ever upgraded the HDMI HDR capability in firmware (Edit: I have now). I have seen attempts at petitioning them, with very few takers. I haven't got a 4k capable card, nor UHD bluray player to test either. Which is why I ask here. Before investing money in a GPU.

"Strange, I quickly checked this forum and there are guys who used HDR via PC (meaning using HDMI)."

So it seems, but it's kind of inconclusive, and some confusion, as per usual on forums, and the thread ends in 2017.

"You want to output HDR to the TV with madvr: it looks way better than SDR converted image (with madvr tonemapping)."


That's surprising. So far what I had seen seemed to suggest SDR out. I'll remember to try if I get that far.

[Charky]

That's surprising. So far what I had seen seemed to suggest SDR out. I'll remember to try if I get that far.
I don't get why anyone would want to convert HDR to SDR on a HDR-capable display.

I mean, HDR to SDR conversion is useful if your display don't support HDR (I watch a lot of HDR content on my Sony HW40ES thats been converted to SDR thanks to madvr excellent dynamic tone mapping), but if you own a HDR display and have HDR sources... just watch them as they are supposed to be :D

[j82k]

On a display that can archive its peak brightness and its full color gamut in SDR, HDR to SDR would be the better method as that would make certain there is no tone mapping involved on the display end.

The problem is just with Oled TVs which have a restricted brightness output in SDR mode, if that wasn't the case HDR to SDR would be the better choice...

[Gadgety]

I don't get why anyone would want to convert HDR to SDR on a HDR-capable display.

I mean, HDR to SDR conversion is useful if your display don't support HDR (I watch a lot of HDR content on my Sony HW40ES thats been converted to SDR thanks to madvr excellent dynamic tone mapping), but if you own a HDR display and have HDR sources... just watch them as they are supposed to be :D

Right, astoundingly apparently it wasn't functioning well, and LG didn't correct it. There was a petition back in the day:


"Our model has the capability to produce a stunning HDR experience. All HDR videos from Demo UHD-3D - Ultra-HD / 4K / HDR / 3D can verify this fact, running either from an USB stick or an external UHD HDR Player. But when we play "real" HDR content (movies or series) the experience is utterly ruined. The image is dull and the colours are washed out. The problem is that our Contrast and OLED Light are locked when playing HDR Content. The reason is they should automatically be set to maximum for HDR Content, but they are not. The problem is merely a small bug, but it has huge impact on our experience of watching proper HDR content. So please give us an update, LG!"

https://www.change.org/p/lg-hdr-firmware-update-lg-eg920-models

Others were complaining it didn't work from HDR players either. I never had 4k sources so I couldn't test it, and there wasn't any confirmation from LG or anyone else that it had been addressed.

[chros]

I never had 4k sources so I couldn't test it, and there wasn't any confirmation from LG or anyone else that it had been addressed.
Since there's nobody here with your model as it seems, you have 2 options:
- try it for yourself (that's preferable)
- sign up for those forums and ask other members about this, maybe it's changed

[Gadgety]

I got a PM reply. From a GPU it works. Now what remains is selecting a GPU that can handle upscaling from 480 to 4k. It may not be worth the extra cost for a card that can handle it. A question, unrelated to LG TVs, I understand part of the challenge with upscaling is the time it takes for the card to do it. Is it possible to record the MadVR output from a GPU, sort of "remasterization" of the movie file?

[chros]

I got a PM reply. From a GPU it works. Now what remains is selecting a GPU that can handle upscaling from 480 to 4k.
Good news! :) Take a look at the this post (https://www.avsforum.com/forum/26-home-theater-computers/2364113-guide-building-4k-htpc-madvr-114.html#post58596206) and the link in it, hope it helps. 1 more thing: if you can't or don't want to use PC mode on the TV (to get chroma 4:4:4) then you can go down to bicubic60 chroma upscaling that requires way less resource.

[Gadgety]

Thank you, chros! The link in the post shows the table containing this piece of information: if (srcHeight < 577) and (deintFps < 26) "0576p25".

Prior to that it says:

-d3d11 mode in LAV video (madvr image processing is much faster compared to the other modes)
- hdr passthrough to 4k hdr display
- in nVidia panel
-- power management set to "Adaptive"
-- output: RGB, FULL, 12bit


I read that as when inputting below 577 in height your 1060 is still able to output RGB, FULL, 12 bit. Is my understanding correct?

[chros]

Of course :) But instead of guessing, wait for your new card and then you can try out everything :)

[Gadgety]

OK, wow, so the 1060 has enough power to take SD to 4k on a 55 inch screen, and it looks like image and chroma upscaling, NGU Sharp, High Quality, at 25 frames a second, and even faster rates, is my understanding correct? Of course my understanding is rather coarse. Are there auxiliary settings you set lower in order to accomplish this? I think I saw somewhere that HDR tone mapping requires at least a GTX1660 6GB, or from the RTX2060 and up. So, if I forego the RTX series, a 1660 would suffice?

[chros]

OK, wow, so the 1060 has enough power to take SD to 4k on a 55 inch screen, and it looks like image and chroma upscaling, NGU Sharp, High Quality, at 25 frames a second, and even faster rates, is my understanding correct?
Yes, link (https://forum.doom9.org/showthread.php?p=1881568#post1881568), although you don't want to use Sharp with <=720p content, but AA (maybe Standard).

Are there auxiliary settings you set lower in order to accomplish this?
No, but you can see it on those screenshots.

I think I saw somewhere that HDR tone mapping requires at least a GTX1660 6GB, or from the RTX2060 and up. So, if I forego the RTX series, a 1660 would suffice?
I don't know about 1660, I think it doesn't have any tensor core, so if that's true then no point of buying it.
But IMO there's no point of using madvr's HDR tone mapping either with a HDR capable display (like yours) but simply using passthrough. You can read more about it from here through 5 pages (https://www.avsforum.com/forum/37-video-processors/3053602-madvr-envy-anticipation-thread-46.html#post58578684).

[Gadgety]

Thank you very much, chros!

[chros]

Yea, the PC mode flaws are annoying. I have a huge macro in my remote to rename the input every single time I watch HDR... then rename it back after I am done.
They really need to add a easier solution to switch in/out of PC mode.
:D :D :D

Hum no, they need to fix 4:4:4 RGB support so it doesn't suffer worse image quality than 4:2:2 YCC.
+1 (but probably it won't happen anytime soon :( )

I hope they'll start focusing on the bugs PC mode has.
We can only hope and pray :D

Why HDR playback should be affected at all in this mode? PC mode was designed for general desktop usage and gaming.
Because of the increased amount of data to process (4:4:4 vs 4:2:2). Ted thinks that all is happening because of the WeobOS UI.

They're using a lower quality image processing in PC mode compared to other modes, so in practice you have to choose between 4:4:4 RGB and good image quality.
"good image quality" :D

Out of curiosity, how does these two pictures look on an LG OLED in PC Mode when displaying it at 100% size (1:1 Pixel)?
http://www.ozone3d.net/public/jegx/201412/chroma-subsampling-test-4k-tv.png
What @chros was referring to was a statement made by Ted over at AVS, claiming that the LGs don't do a perfect 4:4:4, but very close. Nevertheless, chroma patterns are displayed without (noticeable) sub-sampling.
That's not true: "this can be visible on the yellow and the bottom row text of Quick brown fox image", just look closer and pay attention (SDR or HDR mode, doesn't matter) and you'll notice red small artifacts (I also missed it at first). So, yes, it's visible, but not from 3 metres :)

Only LG OLEDs are affected or also LCD based TVs from LG? Is it year depending?
There is no difference between models and years, as far as we can tell.
Yes, that's right, all of them are the same, and it seems that every LG TV that uses WebOS UI (see above), but I'm not sure about the latter.

But that's just 1 issue out of who knows how many (not necessarily just LG related), e.g.:
- chrominance overshoot (aka "macroblock flashing")
- can't profile/calibrate HDR mode
- near black banding
- tinting
- at least 3 ABL algos
- and this:

... my LG has issues in PC mode but these started quote late into the updates, probably only last year, it was fine initially
As it turned out LG tend to screw up things in later firmwares and they just leave it there as it is for half year/forever: e.g. gamma screw up (black crush) in B8/C8/E8 firmwares. And they (smartly) removed the possibility to revert a firmware in the latest ones! :D

How about Panasonic OLED ... TVs?
They don't have 1 thing that I definitely need: judder free 23p playback in PC mode. @huhn told us this back in the day but I didn't *want* to believe it :)
If the above is not important then this year models can be interesting since finally they also have Dolby Vision.

@jk82 tested it (https://www.avsforum.com/forum/40-oled-technology-flat-panels-general/2896737-oled-screen-uniformity-discussion-banding-vignetting-255.html#post58546778) on GZ2000:
- fixed chrominance overshoot
- chroma 4:4:4 in PC mode
- better near black banding than LG

But if you think that finally a good manufacturer then think again: he has sent back 2 sets (https://www.avsforum.com/forum/40-oled-technology-flat-panels-general/2896737-oled-screen-uniformity-discussion-banding-vignetting-258.html#post58629240) in 3 weeks and that was the pricey flagship model! :)

[huhn]

They don't have 1 thing that I definitely need: judder free 23p playback in PC mode. @huhn told us this back in the day but I didn't *want* to believe it

it's been a couple of years it's just sad to hear that this is still the case.

they have an far above avg user support her in german if you don't contact them using the normal contacts i tried it may self the technician instantly knew what i was talking about but could only point the TV may not have to be able to do this "correctly".

there is one company that at least tries and that's sony are they perfect no far from it but it's not hard to be better then the rest.

processing power limitation is just none sense proper 4:4:4 processing could be done atleast 10 years ago no much longer and now we have big number x times processing power in these TVs. the only reason they don't change is people will still buy there TVs so why care.

[quietvoid]

With test build v103, I can't see any detail loss with "don't desaturate" for DCI-P3 and 760 peak nits on my LG C8.
All is good :)

[RXP]

Quietvoid - could you post what settings you're seeing good results with? I've got it setup for my JVC x9000 but want to see how I can tweak the settings for my OLED.

[chros]

With test build v103, I can't see any detail loss with "don't desaturate" for DCI-P3 and 760 peak nits on my LG C8.
All is good :)
Nice to hear it, I saw that @SamuriHL stated something similar as well :)
As @RXP asked it, can you post your full hdr settings (on avsforum is easier I think), and:
- do you still see flicker / brightness changes during normal content?
- which content did you watch?
- have you compared it to passthrough?

Thanks!

[quietvoid]

These are my settings: https://i.ibb.co/Gx1bQ1g/madvr.png (https://ibb.co/Gx1bQ1g)

I have not tested regular content, just watched the S&M UHD Benchmark demos once while comparing on/off.
There's a significant saturation boost for the Eagle scene, the beak keeps all the yellow instead of being white.

With HSTM, there's much more depth compared to static tonemapping. Passthrough probably looks closer to static tonemapped (dynamic target nits unchecked).

[chros]

I have not tested regular content, just watched the S&M UHD Benchmark demos once while comparing on/off.
There's a significant saturation boost for the Eagle scene, the beak keeps all the yellow instead of being white.

With HSTM, there's much more depth compared to static tonemapping. Passthrough probably looks closer to static tonemapped (dynamic target nits unchecked).
Thanks, then I'll wait for a usable version instead :) Keep us updated ...

Retesting now with GotG.
713 for passthrough, 669 with 760 nit tone mapping for GotG.
Thanks for testing, exactly what I got on my B8 (well, apart from the exact code value, since they're different between the B8 vs C8).

I tried to rephrase it in this post (https://www.avsforum.com/forum/139-display-calibration/2962814-2018-lg-oled-calibration-user-settings-no-price-talk-119.html#post58711138), if you still don't understand what we want to test/prove with it, then just ask :)

[SamuriHL]

With test build v103, I can't see any detail loss with "don't desaturate" for DCI-P3 and 760 peak nits on my LG C8.
All is good :)

It's slightly oversaturated but really close. I watched Dark Knight Rises with it yesterday and just WOW what a fantastic job. I think we need some place in between don't desaturate and 1 and we'll have that nailed.

Nice to hear it, I saw that @SamuriHL stated something similar as well :)
As @RXP asked it, can you post your full hdr settings (on avsforum is easier I think), and:
- do you still see flicker / brightness changes during normal content?
- which content did you watch?
- have you compared it to passthrough?

Thanks!

Yea, they've been doing remarkable work in the past few builds. It's really come a long way. And I want to point out something....this is with the live algo. :D We're not using the measurement files at this point. No flickering anymore with a setting of 1000 on build 103. I've not checked to see if there's been any updates today as I've been stuck at work all day.

I watched Dark Knight Rises as mentioned and I don't know what you mean by "compare it to passthrough" cause passthrough doesn't even come close. LOL

[SamuriHL]

Thanks for testing, exactly what I got on my B8 (well, apart from the exact code value, since they're different between the B8 vs C8).

I tried to rephrase it in this post (https://www.avsforum.com/forum/139-display-calibration/2962814-2018-lg-oled-calibration-user-settings-no-price-talk-119.html#post58711138), if you still don't understand what we want to test/prove with it, then just ask :)

I see what you want to do but I have no way of testing it. Would be nice if I had an HD Fury cause i could throw it on the UB820 but that isn't possible. I did read the later discussion, too, about it not being a super accurate calculation due to the possibility of 1 to several bright pixels being left in a frame. madshi calculates a 99% value for it that should help with that. I know soulnight's tool that works on the measurement files allow you to specify what you want to use which is rather handy. I don't have the time right now but you might want to check using an older madvr test build (measurement files are currently messed up afaik) and using soulnight's tool to specify the value you want to use. See if that changes anything. It may not.

[Warner306]

These are my settings: https://i.ibb.co/Gx1bQ1g/madvr.png (https://ibb.co/Gx1bQ1g)

I have not tested regular content, just watched the S&M UHD Benchmark demos once while comparing on/off.
There's a significant saturation boost for the Eagle scene, the beak keeps all the yellow instead of being white.

With HSTM, there's much more depth compared to static tonemapping. Passthrough probably looks closer to static tonemapped (dynamic target nits unchecked).

Does Histogram Shaped Tone Mapping work with HDR output? There is no need to scale the target nits with passthrough. It should always be 760 nits PQ 1:1. HSTM is supposed to scale with the gamma curve.

[SamuriHL]

It definitely works with HDR output. Very well, in fact. I don't use passthrough at all.

Sent from my SM-G975U using Tapatalk

[Warner306]

I don't understand what it is doing to the HDR output. The dynamic target nits takes the tone mapped BT.2390 PQ tone curve and scales the entire source range to fit the entire display gamma curve. This shouldn't be necessary with HDR output to a display using a fixed PQ curve.

For example, if the target nits is 1,000 nits and the display nits is 100 nits, the entire image is compressed at a ratio of 10:1. So the image is 10x darker than the original source values. If the target nits drops to 200 nits, then the entire image is scaled to 100 nits at a 2:1 ratio. Why would you want to scale the entire source range when using HDR output?

HDR output should only be compressing the specular highlights with a straightforward PQ tone mapping curve. There is no need to expand some parts of the image and compress others after the roll-off is applied.

[chros]

I watched Dark Knight Rises as mentioned and I don't know what you mean by "compare it to passthrough" cause passthrough doesn't even come close. LOL
:D I'm glad to hear it!

No flickering anymore with a setting of 1000 on build 103.
Although I've read about more issues, so I'll wait until it will be fully usable.

Does Histogram Shaped Tone Mapping work with HDR output? There is no need to scale the target nits with passthrough. It should always be 760 nits PQ 1:1. HSTM is supposed to scale with the gamma curve.
Hm, interesting thought.

I see what you want to do but I have no way of testing it.
...
you might want to check using an older madvr test build (measurement files are currently messed up afaik) and using soulnight's tool to specify the value you want to use. See if that changes anything. It may not.
:) There's no need for it. Let me try to explain (https://www.avsforum.com/forum/139-display-calibration/2962814-2018-lg-oled-calibration-user-settings-no-price-talk-119.html#post58715418) one more (and last) time :)

[SamuriHL]

I don't understand what it is doing to the HDR output. The dynamic target nits takes the tone mapped BT.2390 PQ tone curve and scales the entire source range to fit the entire display gamma curve. This shouldn't be necessary with HDR output to a display using a fixed PQ curve.

For example, if the target nits is 1,000 nits and the display nits is 100 nits, the entire image is compressed at a ratio of 10:1. So the image is 10x darker than the original source values. If the target nits drops to 200 nits, then the entire image is scaled to 100 nits at a 2:1 ratio. Why would you want to scale the entire source range when using HDR output?

HDR output should only be compressing the specular highlights with a straightforward PQ tone mapping curve. There is no need to expand some parts of the image and compress others after the roll-off is applied.

You would have to ask madshi on this one. What I do know is it makes a difference, even with HDR output. We were losing shadow detail and these latest builds have brought that back. I can't claim to know the technical details on this at all. But I can clearly see a very large improvement in PQ with these latest builds.

[chros]

I don't understand what it is doing to the HDR output. The dynamic target nits takes the tone mapped BT.2390 PQ tone curve and scales the entire source range to fit the entire display gamma curve. This shouldn't be necessary with HDR output to a display using a fixed PQ curve.
I don't know it either, ask madshi, you made me/us curious.
This is how I can imagine: compressing the range happens before applying the gamma curve on it.

HDR output should only be compressing the specular highlights with a straightforward PQ tone mapping curve. There is no need to expand some parts of the image and compress others after the roll-off is applied.
Maybe it happens before the roll-off.
Well, the current HTSM algo touches everything not just the highlights, and we use 760 nits as real peak nits. :)

[Warner306]

I'd ask him yourself. I don't even have an HDR display to test. If it is expanding or compressing the entire source range, it must be doing something odd to the values below the knee point of the tone mapping curve. Aren't they always supposed to be untouched PQ values with 1:1 luminance?

[chros]

I'd ask him yourself.
I don't dare to ask him anymore, I always feel I just upset him with my potential issues/bugs :)

I don't even have an HDR display to test.
It doesn't matter, since it's almost impossible to test these stuff with real content :D
So, if you are curious, just ask him.

If it is expanding or compressing the entire source range, it must be doing something odd to the values below the knee point of the tone mapping curve. Aren't they always supposed to be untouched PQ values with 1:1 luminance?
Well, this is a hard question, since there's no standard to follow when you tonemap to <=150 nits.

Not to mention, I always feel that HDR10 only (!) a compatibility layer of Dolby Vision for devices that don't support it. What I mean is that we don't have the capability of custom mapping based on scene-by-scene bases. And since almost every BR disc has DV, that's definitely not the creators' intent.

That's why personally I don't mind modifying the curve (e.g. with enabling LG's DTM).

But again, testing these stuff is not that easy: you can't just stop the playback and jump to a frame and take a look, because these algos (including LG's DTM as well) behave differently during normal playback (e.g. depending on the last 2 secs data).

[Warner306]

Well, this is a hard question, since there's no standard to follow when you tonemap to <=150 nits.

Yes, I agree with that, and for displays that must tone map 0-100 nits like virtually every projector HSTM tone mapping seems to work really well. I like it. I would call it a sophisticated form of local tone mapping.

Histogram Shaped Tone Mapping works like a local contrast enhancer that attempts to tone map each part of the image independently. I like how it makes an intelligent decision on what to enhance in a frame and what to compress based on the composition of the histogram.

FYI, a brightness histogram looks like this:

https://photographylife.com/understanding-histograms-in-photography

The histogram shows the distribution of pixels in the image. HSTM is looking at the average distribution of pixels in the image and dividing each scene into pixels that are above the average height of the mean and those that below the average height of the mean.

Parts of the image that are above the average distribution of pixels are expanded to be brighter and more colorful. And parts of the image that are below the average distribution of pixels are compressed more aggressively to appear darker and less colorful.

The end result is that you get contrast enhancement in bright HDR scenes and shadow detail restoration in darker HDR scenes or mixed brightness HDR scenes where shadow detail would otherwise be lost due to linear compression of the entire image.

But for a true HDR display that doesn't have to tone map 90% of the image 90% of the time, it shouldn't be necessary to tone map 0-100 nits (reference white) at all. Any loss of shadow detail or issues related to the accuracy of reference white are the fault of the default display calibration and not the tone mapping of the TV.

I might ask if using HSTM with HDR output is intended or simply hasn't been tested yet at some point when testing is complete. It definitely doesn't make sense to expand or compress ALL of the source values in the video for most scenes to fit a display representing HDR using a PQ curve and not a relative SDR gamma curve. HDR to SDR is expanding and compressing the entire gamma curve like an accordian as the video plays.

This expansion and compression of the image steps is meant for displays that have a very narrow range of contrast and have no choice but to compress almost the entire frame to balance the shadows, midtones and highlights to have the right amount of HDR contrast. Most bright HDR TVs don't have this issue with a lack of dynamic range to work with, save ABL and the brightest specular highlights that have to be compressed.

[chros]

I might ask if using HSTM with HDR output is intended or simply hasn't been tested yet at some point when testing is complete.
Thanks for the explanation above.
Definitely do it, because I'm pretty sure that no one will properly test HSTM with "output HDR format" :)
We are 4-5 of us who follow that topic and have HDR capable TVs, and apart from some small testing no one did any proper testing so far.
And I'm also pretty sure that none of us knows what's going on and why :)

It definitely doesn't make sense to expand or compress ALL of the source values in the video for most scenes to fit a display representing HDR using a PQ curve and not a relative SDR gamma curve. HDR to SDR is expanding and compressing the entire gamma curve like an accordian as the video plays.

This expansion and compression of the image steps is meant for displays that have a very narrow range of contrast and have no choice but to compress almost the entire frame to balance the shadows, midtones and highlights to have the right amount of HDR contrast. Most bright HDR TVs don't have this issue with a lack of dynamic range to work with, save ABL and the brightest specular highlights that have to be compressed.
I'd say I agree with you *if* I wouldn't feel like this:

... HDR10 only (!) a compatibility layer of Dolby Vision for devices that don't support it. What I mean is that we don't have the capability of custom mapping based on scene-by-scene bases. And since almost every BR disc has DV, that's definitely not the creators' intent.

That's why personally I don't mind modifying the curve (e.g. with enabling LG's DTM).

But definitely interesting thoughts, thanks!

[quietvoid]

But for a true HDR display that doesn't have to tone map 90% of the image 90% of the time, it shouldn't be necessary to tone map 0-100 nits (reference white) at all. Any loss of shadow detail or issues related to the accuracy of reference white are the fault of the default display calibration and not the tone mapping of the TV.For the little I understand, because the madVR peak nits is set high (760 nits) and the color gamut is DCI-P3, there's rarely tonemapping involved for most content that is under 1000 nits peak brightness. The TV maps the brightness according to its internal curves.

Now I don't know exactly if HSTM affects shadow detail when the source nits are under the target nits, but I'd think it does not.

So essentially, for HDR displays, madVR is just useful to tonemap over 1000 nits down to the specified target nits, so that the TV can track PQ as usual.
And in these cases, it does better than just leaving the TV tonemap it on its own.

[chros]

Now I don't know exactly if HSTM affects shadow detail when the source nits are under the target nits, but I'd think it does not.
It does, that's what I reported here (https://www.avsforum.com/forum/24-digital-hi-end-projectors-3-000-usd-msrp/2954506-improving-madvr-hdr-sdr-mapping-projector-224.html#post58683312).
I went through the usual 8 test files, and in around 5% there was a clear difference. Whether it's useful or not is another question :D

No offense (and I like the way you test things!) but that's also proves my point :D :
apart from some small testing no one did any proper testing so far

[quietvoid]

Yeah, my eyes aren't good enough to be testing small differences that HSTM causes.
Especially at a distance on a relatively small display. :)

[chros]

:) There are lot of settings that result in tons of combination, so probably that's the case :)

[quietvoid]

I've played with version 107b and don't see any shadows getting affected by HSTM when under 760 nits.
Only changing the HSTM strength from 0 to 50.
With high brightness demos, it darkens the "dark" areas to give more depth. In actual content, it does nothing when under 760 nits.

Also the saturation setting looks the same with all 3 options now, for some reason.

With these settings
https://i.ibb.co/YNJFwvw/v107b.png (https://ibb.co/YNJFwvw)

[chros]

I've played with version 107b and don't see any shadows getting affected by HSTM when under 760 nits.
Only changing the HSTM strength from 0 to 50.
With high brightness demos, it darkens the "dark" areas to give more depth. In actual content, it does nothing when under 760 nits.
Thanks, that's what I saw as well after watching 3 titles.
I'm not sure what it exactly does with titles below 760 nits. But for us to test it we definitely need way brighter real titles (e.g. The Meg and similar ones).

What do you think about the new HSTM params? :)

Btw, I have a small HTML-Js file as a remote control (as a complementary to this (https://forum.doom9.org/showthread.php?t=175007)) for madvr (that can be used on any mobile/tablet) to easily switch between passtrough and madvr DTM (with the help of madvr profiles). (It only works with mpc-hc/be)
If you want I can upload it somewhere ...

[chros]

I just remembered that the C8 shows different 20 Point WB adjustments points when it receives 1000/4000 nits metadata.

I assumed a different WB value shown means that the TV uses a different tone mapping curve but I could be wrong. Looking at HDR white clipping pattern I can't detect a change when going from 760 to 761 in madVR even though that causes the TV to show a different WB value.
You were absolutely right: the 19. point values of WB have nothing to do with the used tone curve.
Actually the different 19. point values of WB are triggered by the following MDL (!) values:
- 1.: 669 (C8) , 669 (B8) : 100 - 760
- 2.: 696 (C8) , 692 (B8) : 761 - 2500
- 3.: 713 (C8) , 705 (B8) : 2501 - 10000

Honestly, I don't know what these are used for :)

I tested the PQ of my 65B8 with different metadata values (https://www.avsforum.com/forum/139-display-calibration/2962814-2018-lg-oled-calibration-user-settings-no-price-talk-120.html#post58739264).

There are interesting things there:
- 3. : maxFALL: it seems that the TV doesn't use it at all *or* it's used by ABL algo(s)
- 4. a) : if maxCLL 10000 and MDL is large (>=4000), then PQ gets really weird :)
-- that's why the 10000 S&M demo looks like crap via passthrough!
- 4. b) : Something strange is happening with maxCLL at 501 nits (not below this value!) and it's like it tries to correct this with higher values
-- do you remember when I stated (https://forum.doom9.org/showthread.php?p=1882982#post1882982) that pixelshader mainly results in slightly brighter image with passthrough, but not (!) with titles that has smaller maxCLL? Now we know why :) :
-- the image gets brighter between 501 and (let's say 670), like with Guardians of the Galaxy (2014)

And these measurements make me think that which are the 3 tone curves exactly that people used to talk about regarding to LG Oleds. :)

Lastly, since now we are *sure* that 19. point values of WB have nothing to do with tone curves, take a look at point 6. :
- take out your meter from the closet and measure your peak brightness with 1% window (~700 with B8, it should be + ~200 higher with C8/E8)
- then measure diffferent PQ curves around the resulted value to see which tracks PQ/rolloff the best

[quietvoid]

I really can't tell any difference between the different HSTM settings/curves on demos.
There's just a difference between strength 0 and 50. Maybe I'm doing something wrong?

So for actual content, I don't know if HSTM affects us high nits HDR users at all..

[chros]

I really can't tell any difference between the different HSTM settings/curves on demos.
There's just a difference between strength 0 and 50. Maybe I'm doing something wrong?
:D I don't think so. I didn't have time to check them yet, maybe during the weekend, hopefully they will come up an agreement about this until then :)

So for actual content, I don't know if HSTM affects us high nits HDR users at all..
I don't even know whether the rest of the options affect us :D
E.g. what about the dynamic tuning value? (I use 50 for it).

[quietvoid]

I don't even know whether the rest of the options affect us :D
E.g. what about the dynamic tuning value? (I use 50 for it).
Yes, that one works for sure with high brightness scenes. You get a more contrasty image.
I used to have it at 100, but realized it lowered overall brightness a little too much so lowered to 75.

Last time I played with it was watching Pacific Rim, so that's on the top brightest titles. Not much difference with "normal" content.

[SamuriHL]

If a scene's brightness is below your real display nit, it shouldn't have any impact at all. It's only if there are elements in a scene that are above the real display nit that the tone mapping settings are being applied. So if you're not seeing much difference on "normal" content then that normal content may not be above your real display nits or not far above it.

[chros]

If a scene's brightness is below your real display nit, it shouldn't have any impact at all.
Are you talking about all of the settings or just dynamic tuning?

[chros]

I really can't tell any difference between the different HSTM settings/curves on demos.
There's just a difference between strength 0 and 50. Maybe I'm doing something wrong?

So for actual content, I don't know if HSTM affects us high nits HDR users at all..
Try out these settings (v109):
https://thumbs2.imgbox.com/50/13/BJbEn3YO_t.png (http://imgbox.com/BJbEn3YO)

- top row: Neo-XP
- middle: Manni
- bottom (selected): madshi's (from couple of days ago)

Try out with John Wick sample, e.g. @00:10 sec: madshi's is extremely strong compared to the other 2 :D

And "desaturate 1" option is working again, e.g. The Meg sample.

Let me know what you think.

[SamuriHL]

Are you talking about all of the settings or just dynamic tuning?

All, including dynamic tuning. The whole point of dynamic tone mapping is to only tone map what needs to be compressed and leave the rest alone. Static tone mapping is a whole other world.

[SamuriHL]

Try out these settings (v109):
https://thumbs2.imgbox.com/50/13/BJbEn3YO_t.png (http://imgbox.com/BJbEn3YO)

- top row: Neo-XP
- middle: Manni
- bottom (selected): madshi's (from couple of days ago)

Try out with John Wick sample, e.g. @00:10 sec: madshi's is extremely strong compared to the other 2 :D

And "desaturate 1" option is working again, e.g. The Meg sample.

Let me know what you think.

I honestly can't wait til they revisit saturation. I'm with soulnight in that I believe somewhere between no desaturation and desaturation 1 exists a sweet spot.

[chros]

All, including dynamic tuning. The whole point of dynamic tone mapping is to only tone map what needs to be compressed and leave the rest alone. Static tone mapping is a whole other world.
Hm, you are right! :)
Although I hoped that's not the case :)
I did a little testing (https://www.avsforum.com/forum/24-digital-hi-end-projectors-3-000-usd-msrp/2954506-improving-madvr-hdr-sdr-mapping-projector-250.html#post58767112) as well, read it through.
I attached there the settings I use for now, let me know guys if you dislike something in it.

I honestly can't wait til they revisit saturation. I'm with soulnight in that I believe somewhere between no desaturation and desaturation 1 exists a sweet spot.
Well, saturation setting doesn't do anything, again, with titles below 700 :)

Btw, did you guys measure your C8 and adjust your real peak nits setting? Or do you still use 760?