HDR-issue with 2017 LG OLEDs or with the Vero?

75% on the 1000 nits pattern on my C9.

1 Like

Just did a complete reinstall of OSMC on my Vero 4K as per the Wiki instructions (OSMC_TGT_vero3_20190808.img). After the launch I did nothing except enable “Adjust display refresh rate: On start/stop” in the settings and added a network share with my files. Still 4000nits on everything.

Then ran the update in “My OSMC” to update to the October update. Still 4000nits.

Then tried reinstalling with the OSMC_TGT_vero3_20190105.img to see if the last Kodi 17 build already had that issue. No luck. Seems this issue has been around for a while :neutral_face:

@sam_nazarko If you want, I can try installing an imagine with the new unreleased kernel. Maybe the HDR changes in there somehow fix the issue and we don’t need to bother investigating this issue any further.

For what it’s worth on a Samsung Q9FN 11. HDR Clipping Grayscale 1,000 nit_HDR10.mp4 shows at about 74-75% too when played back from a Vero 4K+ through a Denon AVR-X4500H (which is showing as 4K:24hz in/out, HDR10 in/out, YCbCr 4:4:4 in/out, BT.2020 in/out, and 10bit in/out for pixel depth)


Same config here q9fn and denon x4500h, output 4K:24hz in/out, HDR10 in/out, BT.2020 in/out, and 10bit in/out for pixel depth

flashes at 82%. Played the file from 1st post. 11. HDR Clipping Grayscale 1,000 nit_HDR10.mp4

1 Like

I’m not familiar with Samsung TVs. Are you using a preset that is close to reference (“Movie” with “Warm2” I guess) and doesn’t do any post-processing, especially something like dynamic tone mapping or dynamic contrast?

Just normal settings, just factory.

My Q9FN basically follows this Samsung Q9FN/Q9/Q9F QLED 2018 Calibration Settings - RTINGS.com if that’s of use?

1 Like

I think factory is something crazy like “Normal” or “Dynamic” which has post processing enabled. Can you try again to play the file but change the mode to “Movie” and make sure “Color tone: Warm 2” and “Contrast Enhance: off”. Otherwise the TV will do its own tonemapping.

At your service. Movie, warm2 contrast enhance off

Now it flashes 75%

1 Like

According to some people on here, you should not be playing back test patterns as they are not real video and nobody would ever notice a difference even though you just spent 1000’s on a top end tv and calibration equipment :slight_smile:

Anyways, there is definitely picture degradation when HDR MaxXXX is not passed through - its obvious. Vero 4K+ does not currently pass these but will do soon. Looking fwd to checking the test patterns - oops ….

My Odroid N2 does pass HDR metadata correctly and the difference is obvious. Same with my Oppo Bluray streamer vs Vero 4K+. PS Odroid N2 has potentially slightly better pic quality then the Oppo as the new Mali GPU in Amlogic S922 devices has better chroma upscaling than the Oppo Mediatek. Once again, easy to spot on Spears & Munsil UHD benchmark disk. But you know what they say about these things ………

Edit : meant to say, I would just wait for the MaxXXX fix and take it from there.


This is NOT about the missing MaxCLL/MaxFALL information.
This is about the basic Mastering Display Luminance and all movies being tonemapped to 4000nits by the Vero 4K(+) on, apparently, all the LG OLED 2017 models. Adding MaxCLL/MaxFALL in the upcoming update isn’t gonna do anything if the entire HDR-information over HDMI is being ignored/not received by the TV.
The pattern is just there to make it easy for everyone else to test.

No, just no. During my testing trying to isolate the issue the past few days I’ve looked at a lot of samples from movies and compared them to the internal player and the FireTV Kodi. It’s very noticable for some movies, especially if they’re dark to begin with (Underworld Blood Wars for example looks like a black crush-gallery on the Vero 4K).
Tonemapping to 4000nits instead of 1000nits (let alone something like 300nits from the missing MaxCLL) these TVs begin their rolloff way too early. So you not only end up with an overall dimmer image, but the shadows and midtones are too dark while the highlights are still bright, essentially blinding you more than they should.

Having said that, I’ve underestimated how much of a difference missing MaxCLL/MaxFALL can make. Aquaman for example has a MaxCLL of 1506cd/m² while the max Mastering Display Luminance is 4000cd/m². The underwater scenes look really great on the internal player and the FireTV Kodi, it looks very bland and boring on the Vero 4K in comparison. So I’m very excited for the MaxCLL-update and hopefully a working Mastering Display Luminance-signal.

1 Like

I was being a tad sarcastic re the difference :slight_smile:

Of course there is a massive difference and I have plenty of discussions on AVSForum Calibration threads around this.


Just to be clear, Vero4k(+) does not do any tonemapping, nor should it unless converting to SDR.

We send a signal encoded as BT2020/smpte2084 as decoded from the source, together with the metadata to enable the display to do its best to reproduce what the colourist saw on his/her master display, taking account of the viewer’s preferences relating to their viewing environment.

Exactly how display manufacturers do their tonemapping is unknown - including how they treat max lumi and MaxCLL and what they do if either are absent. Published recommendations are scarce and I suppose those algos are proprietary. But the fact that Panasonic have introduced variable tonemaps into their BD players suggests the display makers haven’t got it right yet. It wouldn’t surprise me if LG (and others) were on a learning curve in 2016/7 and probably still are.

1 Like

I know. I could’ve chosen my words more carefully in that sentence. But my other posts should make it clear what I mean as I heave repeated them at least 5 times and this was the only sentence that could’ve been taken out of context and interpreted another way.

I don’t understand the rest of your post though and what it has to do with what this thread is about. Either the max luminance data ends up somehow malformed when connected to an LG OLED 2017 or not sent in the first place because of a handshake failure or I don’t know what (min luminance probably too). I already proved this with several test files and movies, which I described in detail above. All of which had no MaxCLL/MaxFALL data, this is strictly about Mastering Display Luminance.
LG OLED 2017 models always tonemap to 4000nits when connected to the Vero 4K(+), regardless of metadata, and have been doing so since at least Vero’s December 2018 version. 4000nits is LG’s default fallback value for missing mastering display luminance data (probably for “wrong” data too).

I’ve done everything I can to isolate the issue with my limited abilities. And it’s some sort of incompatibility between these two devices.

I’m open to test and try anything you want to help you narrow this down. Like I already mentioned in an earlier comment, maybe I can install an image with the newer unreleased kernel, maybe those new HDR changes somehow fixed the issue altogether and this is not worth pursuing any further. Or some build with better debug information/log that I can send you.

From what I remember about their HDR Optimizer, it’s about giving the user a choice. There’s several ways how to handle highlights that go beyond the display’s capabilities, not every display manufacturer chose the same method. Each one has its own advantages and disadvantages, I wouldn’t agree it makes one way more right than the other.

Heyo. Just chiming in here.

Just tested the 1,000 nit file on my Vero 4K+ connected to LG C7 65” via my Sony HT-ST5000 sound bar. The bars flash clearly up to 75 which is very clear to see. I can still see a slight flashing at 76 but it’s very faint and I have to sit about a foot from the TV to see it. At normal viewing distance of around 10 feet I don’t see it past 75 and maybe 76 if I convince myself.

The whole length of the white bar pulses slightly as the viewable areas flash, which is separate from this issue and is related to the luminance limiter all OLEDs have to prevent burn-in.

Picture setting is Cinema Home (User), with all dynamic stuff turned off).

I’m guessing my findings throw a wrench in the works.

1 Like

Interesting. The bars past 75 flash very clearly on my screen.
Are you on the latest firmware (5.80.35 for me in my region) and the latest OSMC? If I use Cinema Home and disable all its stupid dynamic contrast and dynamic color options, it’ll still flash way past 2000nits.
What else could be different that causes this for me and others?

Yeah, it’s certainly interesting :slight_smile:

What I’m trying to say is we could be shooting for a moving target and probably the only answer is user controls a la Panasonic (something that’s on my wishlist).

I just ran your test vids on my (2017) Panasonic and that too overcooks the tonecurve. Using the (almost ready) 4.9 kernel, the 1000nit test flashes up to the 2000nit level and the 2000nit test flashes up to nearly 100%. Mode normal but same result on True Cinema.

Could it be that the Firestick knows something about the quirks of various TVs and compensates?

1 Like

This is why AML have their register mapped SQLite DB for presets in AOSP. We can do the same and adapt this eventually, but the focus would be getting 4.9 out the door where we know the HDR implementation is sound.

That’d be interesting. This would not be part of Kodi I suppose?
I tried a movie scene and manually changed it to several different MaxDisplayLuminance (even quirky ones like 557cd/m²) but always without any MaxCLL data. The TV displayed them all accordingly through the FireTV but not once correctly through the Vero 4K, it would always just do its 4000cd/m² thing :neutral_face: