Spears & Munsil 4K UHD HDR Benchmark (Test Discs) May 15, 2023

[Stacey Spears]

Originally Posted by TbeRw01

The LLDV bug is not actually a big issue. It's the L2 trim Chroma_Weight that doesn't work but It's almost NEVER used in Movies and TV Shows by colorists.

The Spears and Munsil montage just happens to use it a lot.

Can you dump Love + Death and Robots from Netflix? Chroma weight is used more the brighter you go. We saw artifacts on that show in LL that went away in normal.

Here are the shots that show the artifacts when chroma weight is missing, when used. This is the previous montage. I don't know if the same issue occurs on the new one since it was re-graded.

Here is a shot list of the previous disc and which shots to look at: (again, previous disc!)
https://www.dropbox.com/s/4q44gk2244...pare.xlsx?dl=0

Shot 01 - There is banding as it fades up from black in player-led on LG. Not present in tv-led or HDR10.
Shot 18 - Look at the sun on the right. In player-led, you will see noise around and inside the sun. Not present in tv-led or HDR10.
Shot 23 - Look at that center cloud. In player-led it has a distinct outline. Not present in tv-led or HDR10.
Shot 29 - Look at the top of the honey, it's much noisier in player-led. Not present in tv-led or HDR10.
Shot 30 - Look at the yellow parts of the peacock feather. Lots of dancing noise in player-led. Not present in tv-led or HDR10.
Shot 31 - It is undersaturated in player-led. Not present in tv-led or HDR10.
Shot 32 - Look at the yellow flowers on top of the cactus. Looks posterized in player-led. Not present in tv-led or HDR10.
Shot 34 - Look at the lower right inside of the flower. There is a distinct outline, like clipped color in player-led. Not present in tv-led or HDR10.
Shot 36 - Look to the right of the butterfly, where it is out of focus. Looks like a clipped / flat spot in player-led. Not present in tv-led or HDR10.
Shot 59 - Look on the right lower third edge of the frame. Looks like a lot more visible noise in player-led. Not present in tv-led or HDR10.
Shot 62 - Look at the eagles beak. It's this greenish yellow in player-led. Not present in tv-led or HDR10.
Shot 64 - Look at the scales with more yellow in them. They are clipped in player-led. Not present in tv-led or HDR10.

The shot list has notes in it as well with the shots to look at, in case I missed some.

[TbeRw01]

Originally Posted by Stacey Spears

Can you dump Love + Death and Robots from Netflix?

By default,in resolve the Dolby algo doesn't touch Chroma Weight so it has to be done by the colorist.

Here's the first 3 episodes' plot for NF Love + Death and Robots Season 03, Chroma Weight is not used, at least for these episodes:
https://drive.google.com/drive/folde...Cc?usp=sharing


And here's one recent movie that slightly use it and its only a 400nits grade:

[Geoff D]

Yep, the chroma does clip/show noise in the Montage in LL Cool J (though the ZD9 does much betterer with the banding than Stacey’s LG, not sure why, and the noise in the ‘eye’ of the peacock feather is quite pronounced in HDR10 whereas both DV flavours (MEL and FEL) clean it up nicely in LL on the ZD9) but in real world viewing I’ve yet to notice such issues with colour when comparing to the HDR10 equivalent.

[TbeRw01]

And it looks like something changed on the 2022 LG TVs because in TV-LED DV connected to the x800m2, Chroma Weight doesn't work. It also doesn't work using the TV internal player...

The same player connected to the LG C8, Chroma Weight works just fine.

[Stacey Spears]

The CX is the latest I have tested on and it was okay. I will ask someone who has the G3 to test with their Panasonic. Hopefully it works.

[Scenic Labs]

Then it should certainly alarm you that the publisher (me) works from home most days from a house in the New Jersey Highlands surrounded by wild turkeys, possums, coyotes and mountain lions. And that his teenage daughters are currently packing S&M preorders. Oh, and our business address is a UPS Store. But we’ve had people come to my house to buy bias lights and discs!

In all seriousness, I’ve worked with Ian Smith at Camrose for over 20 years. When I was the Director of Marketing for DVD International, he handled distribution for Digital Video Essentials in the UK.

That said, our goal over the next few months is to unify a number of the international websites and require most dealers to carry our entire range.

Originally Posted by SeeMoreDigital

I'm a bit nervous buying from them. If you click on their 'Contact Us' link and look up their address (ie: 7 Union Street, Carmarthen, SA31 3DE, UK), 'Street View' reveals a small lime green painted terraced house with a red front door!

[RocShemp]

Originally Posted by Geoff D

Got my copy today, thanks Stacey and thanks to all involved for creating such a comprehensive set of test goodies.



I’ve already found the hidden Dolby Vision test pattern and as expected it’s got a great big red X owing to the ZD9’s low latency Dolby.

I was super excited mine shipped today. You got yours already, making you the lucky one.

[mrtickleuk]

Originally Posted by Scenic Labs

Then it should certainly alarm you that the publisher (me) works from home most days from a house in the New Jersey Highlands surrounded by wild turkeys, possums, coyotes and mountain lions. And that his teenage daughters are currently packing S&M preorders. Oh, and our business address is a UPS Store. But we’ve had people come to my house to buy bias lights and discs!

Hehe. Please pass on my thanks to your teenage daughters! And my humble thanks to Stacey, Don and everyone else involved!

[SeeMoreDigital]

Originally Posted by Scenic Labs

In all seriousness, I’ve worked with Ian Smith at Camrose for over 20 years. When I was the Director of Marketing for DVD International, he handled distribution for Digital Video Essentials in the UK.

That said, our goal over the next few months is to unify a number of the international websites and require most dealers to carry our entire range.

Well colour me convinced... I've ordered the disc from the Carmarthen based distributor. Total cost £41.90 (including £2.95 p&p)

[xskip]

Originally Posted by xskip

avail. in EU, not UK. any spec. shop? thank you.

found some in FR. (futureland and hdland) price: 79 or 75€ + p&p
from biaslighting.co.uk with VAT free purchasing and VAT in my country+ postal service tax... all together 65€ :-)

[Robert Zohn]

We fulfilled all of our pre-orders today and still plenty available at the one-time pre-launch promotional price. The pre-launch sale ends Monday, May 15th.

[cheez avenger]

My copy shipped out yesterday.

[INdetectableMAN]

With this new disc, will we be able to verify the EOTF Dolby Vision tracking of our OLED televisions in absolute mode pre/post calibration?

With Calman it is not possible because OLED TVs do not have an iTPG to verify Dolby Vision in absolute mode, so you will need an external DV absolute compatible pattern generator before and after calibrating Dolby with Calman home for LG.

Will it be possible now to verify pre/post with Calman with these patterns through a player like Panasonic?

And with other software like ChromaPure?

Is there reference material using real images that indicate the nits that certain parts of the frame should measure in DV?

I am especially interested in everything related to Dolby Vision.

Is the 2.9 profile of the discs darker or inaccurate than 4.0 due to tone mapping?

I have already ordered the disc from biaslight uk

[RocShemp]

My copy is due to arrive tomorrow. Can't wait to dive into these particular waters. My last calibration disc was the HD Benchmark 2nd Edition. Looking forward to see all that is new.

[Stacey Spears]

Originally Posted by INdetectableMAN

With this new disc, will we be able to verify the EOTF Dolby Vision tracking of our OLED televisions in absolute mode pre/post calibration?

With Calman it is not possible because OLED TVs do not have an iTPG to verify Dolby Vision in absolute mode, so you will need an external DV absolute compatible pattern generator before and after calibrating Dolby with Calman home for LG.

Will it be possible now to verify pre/post with Calman with these patterns through a player like Panasonic?

And with other software like ChromaPure?

Is there reference material using real images that indicate the nits that certain parts of the frame should measure in DV?

I am especially interested in everything related to Dolby Vision.

Is the 2.9 profile of the discs darker or inaccurate than 4.0 due to tone mapping?

I have already ordered the disc from biaslight uk

Absolute patterns seem to work, at least on Panasonic 9000/820 as they share the same decoder. Lower end models use a different part. On OPPO, bringing up pop-up help or config breaks absolute mode. You have to switch to perceptual and back to get it to work again.

Relative mode does not work correctly on any player.

The HDR Analyzer montage shows you the avg and peak for a given frame on the 10K master. It can't show you what the tone mapping should make it since you would need one version for every display on the planet and those change every year too.

The sample image in the analyzer has been tone mapped down to 1K for viewing.

There are two parts to DV being darker than HDR10.
1. Not setting TMax in the config file. Out of the box, it is below the capability of the display. e.g. I have a C9. My C9 measures 830 nites. The config file has 700 by default for that display.
2. 2.9 uses a metric to test that causes it to be darker. This was removed as a test in 4.0 to reduce that issue.
3. Everyone uses LG as the client monitor to review DV. This is what the content creators sign off on. If it is darker, it is still their intended look.

Keep in mind that 2000, 1000 and 600 HDR10 montages were tone mapped from 10K using Dolby Vision.

[INdetectableMAN]

Originally Posted by Stacey Spears

Absolute patterns seem to work, at least on Panasonic 9000/820 as they share the same decoder. Lower end models use a different part. On OPPO, bringing up pop-up help or config breaks absolute mode. You have to switch to perceptual and back to get it to work again.

Relative mode does not work correctly on any player.

The HDR Analyzer montage shows you the avg and peak for a given frame on the 10K master. It can't show you what the tone mapping should make it since you would need one version for every display on the planet and those change every year too.

The sample image in the analyzer has been tone mapped down to 1K for viewing.

There are two parts to DV being darker than HDR10.
1. Not setting TMax in the config file. Out of the box, it is below the capability of the display. e.g. I have a C9. My C9 measures 830 nites. The config file has 700 by default for that display.
2. 2.9 uses a metric to test that causes it to be darker. This was removed as a test in 4.0 to reduce that issue.
3. Everyone uses LG as the client monitor to review DV. This is what the content creators sign off on. If it is darker, it is still their intended look.

Keep in mind that 2000, 1000 and 600 HDR10 montages were tone mapped from 10K using Dolby Vision.

Thank you very much Stacey for the answer.

Just to point out that she believed that what made the OOTB picture darker in DV was a consequence of a default maximum brightness in the settings file that was overly conservative, ie high.

What we do to get a brighter image is to reduce the Tmax value of the configuration file. For example, Cinema Homa uses a low value of around 350 nits to be brighter.

I understand what you mean about the profile metric of 2.9 obscuring the image. In that case, even after having calibrated the screen, UHD bluray content based on 2.9 due to the tone mapping itself and its meterica would darken the image in certain situations more than in HDR10, is that it?

I think that on my LG screen I will end up disabling DV from the Panasonic players and I will see my UHD bluray in HDR10 format.

[Geoff D]

I ran some HCFR calibration with the DV patterns and on the ZD9 the DV is literally halving the light output vs HDR10 at certain steps of the signal! It's not terrible up to about 50% of the signal, still dimmer but only by a few nits, which is fine for the average level of most HDR content as it doesn't get crazy bright, but up from there it starts to roll off significantly, at 75% you should be getting 1000 nits but it barely gets over 600 in DV and the brightest point the ZD9 can manage - 1800 nits - is down to 900 nits. The PQ EOTF curve is also greatly linearised into a smooth upwards slope (much like what the Pannysonic HDR Optimiser does: https://forum.blu-ray.com/showpost.p...postcount=6551) rather than the steeply logarithmic 'cliff edge' that the PQ EOTF should present.

What's weirder is that when I ran a DV greyscale plot and adjusted it to show the proper colour temp in the measures (as the DV white point on the ZD9 is different from SDR and HDR10), it still looked greenier than my calibrated HDR10 which was showing an essentially identical plot. So even though it was measuring the same, it was different! I know Stacey said about a green tint in some shots in LLDV but shirley my colourimeter would've picked up on it. So I've just eyeballed it to match...which is what I was doing before anyway and thought that these DV patterns would finally allow me to nail down the DV colour temp once and for all. Ah, technology

[Stacey Spears]

If you play the HDR10 10K and the DV on an LG, the LG will show crushing of highlight detail on the horses in snow with HDR10.

Just a guess, but the DV tone mapping may be such that it pulls more overall down in order to show delineation between the levels above a displays capability.

This is also just on the consumer side. The 2000, 1000 and 600 montages don't have this as they were done using the professional Dolby algorithms. Those also work in high precision float vs. integer, but that will show in other ways.

Just speculation.

The green comment was just around the eagles beak where it goes green when chroma weight is not working and is yellow otherwise. I have not seen anything else like that.

On LG, when you calibrate DV, you disable tone mapping, so it does not get in the way. Absolute mode does this as you can see looking at the tone mapping pattern under the Specialty menu. Absolute and relative don't work at all in DVLL, which is by design.

[TbeRw01]

Originally Posted by Stacey Spears

2. 2.9 uses a metric to test that causes it to be darker. This was removed as a test in 4.0 to reduce that issue.

Speaking only of the content mapping metadata (L1), do you have more information on the very old cm v2.9 algo low range that was more dynamic than the recent ones (2.9 and 4.0)? Was the older algo raising letterbox black levels? Was it too aggressive? Even HDR10plus doesn't have any of these restrictions:

Old CM v2.9 algo rules:
doesn't seem to have any restrictions.

Newer CM v2.9 algo L1 rules (regardless of the content brightness):
average_pq floor is 2.5nits
max_pq floor is 100nits
min_pq max is 0.00262nits

CM v4.0 algo L1 rules(regardless of the content brightness):
average_pq floor is 10nits
max_pq floor is 100nits
min_pq max is 0.00262nits

Example:

[Show spoiler]
https://slow.pics/c/WlGwchoE

Regardless of the trim passes (most cmv4.0 rpu don't have any), IMO, the biggest improvement in CM v4.0 is the L3 offset metadata that works with L1.
FYI, all the movie DV plot can be found here: https://drive.google.com/drive/u/1/f...DV8QKFJwFxnDt-

[Stacey Spears]

2.9, on the display side, only existed on VIZIO I believe. If it were on LG, it would have been replaced with a FW update. Displays use 3.1 to playback 2.9 content. There was never a 3.1 on the creation side and 2.9 on the display side outside of VIZIO.

I don't know what the black bar issue was, never asked Dolby about it.