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

[David M]

In that case you’d be raising Contrast, but yes. You might find that a lot of displays don’t let you clip it. Either way it’s mostly theoretical, the difference in practice is probably slight, so don’t worry about it.

[Deciazulado]

If your panel has a contrast of 1000:1 and you use it for 0-120 nits you have about ~800:1 from 0-100 nits, so raising the 100nits signal to the max, gives you the full 1000:1 from 0-100 nits and you get a third of an f/stop more shadow detail/the min black looking slightly blacker compared to mids and highs, from sacrificing the usually small excursions above the 100 nit standard max white.

The more native contrast a panel has the less this might be used/needed.

[DisplayCalNoob]

I think it all depends on the source and the display, I lowered contrast and got better dark color performance, as well as more detail. Pre-calibration measurements were surprisingly closer to reference.

[Stacey Spears]

Originally Posted by Geoff D

I was gonna say it's not so much personal advertising that's the issue, it's if you're advertising things offsite that directly counter what this site does - reviews and whatnot - that they get the hump about it. But you're a content creator, not reviewer, so shill away! And if the site gets a few shekels from people clicking through their links to buy the disc from amazon or whatever then I'm quite sure they won't mind.

In any case: ME WANTEE

Funny, I did not actually make the thread. A mod must have made it for me.

I guess I had better read all of the pages and start answering questions.

[Stacey Spears]

Originally Posted by Mobe1969

Yes look forward to seeing it. And I hope available widely, not just US!

Yes, it will be available in various countries, it is being shipped to them now. They arrive in the US today from the UK, where they were packaged. We will work on updating our website now that they are coming in.

[Stacey Spears]

Originally Posted by avs commenter

I have a Vizio and was under the impression the calibrated settings already on the TV are the most accurate for HDR and messing with them can really throw things off

We advice against touching contrast as the tone mapping may be impacted. But brightness, color/tint (usually correct in the cinema/theater mode), sharpness, white balance, and CMS still need work.

White balance and CMS need a colorimeter and software like SDR.

HDR is much more difficult due to tone mapping. You end up fighting it on most displays. An ideal system would turn off tone / gamut mapping and let you measure the panel native response and then create a 1D/3D LUT to fix up. Only one manufacturer offers this ability today.

We have over 60 years of understanding SD, HD and SDR calibration. We have about 4 years of HDR knowledge on the same subject. Suffice to stay, we are at the beginning of the journey.

I think I shared images on another thread of 422 vs 444 at 1080p and UHD from a player into an HDR display and some really ugly artifacts. Have not had time to figure out if it is the OPPO or LG in that case. I need to test another player into that display and that player into another display.

[Stacey Spears]

Originally Posted by multiformous

Yes. With Contrast at 50 in HDR, my 2017 Vizio M clips at around 1 000 nits. Dialing back to 47 preserves details to 4 000 nits, and 45 will get you all the way to 10 000.

Touching contrast, in HDR, is a big unknown at this point. On some displays, such as the Z9D, you can bring in detail all the way up to to 10,000 nits if you pull contrast down far enough. Is that a good thing? Not sure.

Using the Z9D as an example, it seems to clip everything about ~1600 nits regardless of metadata. You can see this on the tone mapping ramps. What is worse is that there is massive clipping on some shots of the montage IN Dolby Vision, which should not happen. Contrast will bring it back, but also impact HDRness.

This disc will probably be more important for CE companies at this point to have some tools to improve their products.

[Stacey Spears]

Originally Posted by Scarriere

When the link goes live, could Amazon.ca be an option?

Yes. The distributor is receiving the discs today, if all goes well, and he will then ship to Amazon.com and .ca shortly.

Today you can order directly from the distributors website.

As soon as dealers start to receive them, we will update our website.

[Stacey Spears]

Originally Posted by meglos

Will this disc be useful if the user doesn't have test equipment and is adjusting by eye?

Like the current disc, some things need test instruments and somethings don't.

Like there current disc there are lots of things to show you what is wrong that you can't fix, which are intended for CE companies.

Basically the disc has to serve multiple markets. Consumers, reviewers and CE companies. It would be expensive to make custom discs for each market. This is why we have articles for a few basic things only.

With that said, we plan to create a 3rd edition section for articles. We will update the choosing a color space article first, followed by brightness, then write a sharpness article. We are also looking at hiring a 3rd party to write an electronic manual. All of this will take time.

Today, you can start with 2nd edition articles and use this forum to ask us questions. We are using questions from the press and customers to focus documentation efforts.

[Stacey Spears]

Originally Posted by monroef

Will there be a downloadable version of this? I would like to run this through my Apple TV.

Not for a while. We are looking at USB stick for 8K for an add-on project next year.

Streaming is trickier. Amazon has expressed interest in the past. Problem is putting up patterns with no animated instructions might not work so well.

Apple TV would probably be the most difficult to get on to.

[Stacey Spears]

Originally Posted by DJR662

Interesting.

I was always under the impression you would like to have as many bars on the contrast pattern as possible. So basically you're saying it would be alright to lower contrast for SDR until it hits 235 and slightly lowering the backlight as well then?

For SDR, you really want it to go up to 253 on the bars. The background itself is 254.

For HDR, have five versions of every HDR pattern. (10000, 4000, 2000, 1000 and 600 nits) Each one of those contrast patterns limits the boxes to the code value at that nit level. I will post that info shortly.

[Stacey Spears]

Here are some rough notes that I have written so far. It is the start of an outline for our documentation.

We use the the label cd/m² in place of nits. The mean the same thing, nits is sort of the slang term used as it is easier to say.

cd/m²(nits) to 10-bit code values:
10,000 cd/m² – 940
4,000 cd/m² – 855
2,000 cd/m² – 789
1,000 cd/m² – 723
600 cd/m² – 674

Contrast / Dynamic Range High:
Each version of the contrast pattern is unique. The background white, behind the blinking bars, is the max cd/m² value. E.g. For the 10,000 version of the pattern, the background is 940 and the highest blinking box is one less, 939.

The ramp in the middle goes from code value 64 (black) to the code value for the version of the pattern. E.g. The 10,000 version goes to 940 in the center.

In the center of each ramp is a modulation ramp. This ramps down and then up 1% from the original code value in order to make it visually easier to see where clipping occurs.

There is a marker on the ramp at code value 64 and 940 where it applies.

Brightness:
The brightness pattern has been updated to include a ramp at the top and bottom of the pattern that goes from -4% to 4%. The numbers along the ramp are the code values. Those code values also line up with the bars (-4%, -2%, 2% and 4%).

In the center of each ramp is a modulation ramp. This ramps up and then down -1% from the original code value in order to make it visually easier to see where clipping occurs.

Color Space Evaluation:
Each version of the contrast pattern is unique. The Y ramp goes from 64 to the cd/m² value that represents the pattern version. E.g. The 10,000 version goes to 940.

The ramp in the middle goes from code value 64 (black) to the code value for the version of the pattern. E.g. The 10,000 version goes to 940 in the center.

The frequency-based patches, on this pattern, are also limited by the cd/m² version of the pattern.

There are six boxes at the top of the image with a blinking checkerboard pattern in them. These are to spot check the tracking of the transfer function. For HDR this is ST2084 (PQ) and for SDR this is BT.1886 (Power 2.4). If you don’t see blinking in these boxes, then you know that it is correctly tracking at those points.

The color space conversion has two ramps. If the blinking boxes inside clip at the same point for red and green, then it is probably color conversion. As of today, this works on the SDR version. The HDR version will not work until we have displays that are true BT.2020 and 10,000 cd/m².

Resolution Patterns:
There is a unique version of each resolution pattern for each brightness level. Each version ramps from 64 of to the code value mentioned above based on cd/m² value. While this is the true brightness level for luma, it is not for chroma, however we use them anyway, except for 10,000. We use 960 for chroma on the 10,000 versions.

Ramps:
There are three types of ramps. Stimulus, saturation and tone mapping. There are versions in white, red, green, blue, cyan, magenta and yellow. The ramps are located to the center of the screen. We limited the size of the ramps to cover the same area (%) as the window patterns used for calibration.

In the center of each ramp is a modulation ramp. This ramps down and then up 1% from the original code value in order to make it visually easier to see where clipping occurs.

There are three styles of ramps. S&M, round and dither. The primary colors include all three types while the secondary colors only include round and dither.

Stimulus Ramps:
The stimulus ramps show the actual cd/m² value on the top of the ramp. Only white goes to 10,000. The other colors show the proper values. E.g. red goes to 2627 and blue goes to 593.

Saturation Ramps:
...

Tone Mapping Ramps:
These ramps show the same cd/m² value on all versions instead of showing the correct values since it serves a different purpose.

[Stacey Spears]

We need to point out what is NOT included in the disc at this time.
1. No blue filter included. We are testing all of the filter material available, including dichroic glass. So far, nothing works on all displays. e.g. we have some material for blue and green that do work on an LG OLED, but we have found nothing that works on the Sony Z9D.
2. No manual at this point.
3. No audio calibration outside of A/V sync. We are planning an add-on disc early next year that will have Atmos and DTS:X. We have started that planning now.

The primary reason for the add-on is Dolby Vision 4.0. We will also add BT.2020 SDR. The current disc has BT.709 SDR today. Anyone who gets the current disc should only have to pay the price difference for the add-on. (Should be less than $10 if all goes well) Jason may kill me for saying that. At the time the add-on comes out, the dual disc set will go up to $47.95. If we can make a filter work, then it would also come with the add-on. While we are targeting early next year, we are waiting on Dolby 4.0 encoder tools that support profile 7.

I had posted in another thread that the first person that could guess why we would price at $47.95, we would send a free copy. I don't think anyone ever figured out it. It is kind of meta. The first disc was $23.98, the second $29.97 and the 3rd (when we have a second disc) will be $47.95. All are video frame rates. So if we every have a super bundle, we will have to go for $59.94.

[DisplayCalNoob]

Originally Posted by Stacey Spears

Touching contrast, in HDR, is a big unknown at this point. On some displays, such as the Z9D, you can bring in detail all the way up to to 10,000 nits if you pull contrast down far enough. Is that a good thing? Not sure.

Using the Z9D as an example, it seems to clip everything about ~1600 nits regardless of metadata. You can see this on the tone mapping ramps. What is worse is that there is massive clipping on some shots of the montage IN Dolby Vision, which should not happen. Contrast will bring it back, but also impact HDRness.

This disc will probably be more important for CE companies at this point to have some tools to improve their products.

On my display, lowering contrast was all positive. DV content has never looked better. Actually, I feel the tone mapping is performing in the manner it is supposed to, especially when it concerns DV. I mean, tone mapping is not doing its job if the detail that it is supposed to retain isn't visible. I increased the backlight setting as well, personally I don't think any of the settings on these consumer displays are tied to data from Dolby.

Picture quality is far more impressive than it has ever been.

[ROSS.T.G.]

I think it’s fantastic this disc is arriving and will likely order but I’m not sure how much I’ll use it since my B7 settings shouldn’t be touched. I am interested in greyscale calibration but need to replace my meter since it’s drifted quite a bit. Is there DV greyscale test patterns on the disc?

[Stacey Spears]

Originally Posted by ROSS.T.G.

I think it’s fantastic this disc is arriving and will likely order but I’m not sure how much I’ll use it since my B7 settings shouldn’t be touched. I am interested in greyscale calibration but need to replace my meter since it’s drifted quite a bit. Is there DV greyscale test patterns on the disc?

The only DV content is the montage. All test patterns are HDR10. Right now it is a tools issue to create DV test patterns.

On your B7, be sure to set edge enhancement to on as that means off. You also want to be in cinema and not cinema home. The B7 does not do the proper color conversion for SDR UHD content. You can see this using the colour filter in the menu and viewing the HD color bars pattern in green and red only. In HDR, it is correct.

You also want to be sure to send 444 10-bit UHD into the B7. Anything else will mess up content scaled from HD to UHD. You can test this and see the differences with the disc on the scailng HD pattern.

[Geoff D]

Originally Posted by Stacey Spears

Touching contrast, in HDR, is a big unknown at this point. On some displays, such as the Z9D, you can bring in detail all the way up to to 10,000 nits if you pull contrast down far enough. Is that a good thing? Not sure.

Using the Z9D as an example, it seems to clip everything about ~1600 nits regardless of metadata. You can see this on the tone mapping ramps. What is worse is that there is massive clipping on some shots of the montage IN Dolby Vision, which should not happen. Contrast will bring it back, but also impact HDRness.

This disc will probably be more important for CE companies at this point to have some tools to improve their products.

With Dobly at the default 90 on the ZD9 then it resolves at least 4100 nits worth of information, that's if The Meg is any indication (using the OPPO of course, not a DV-enabled Panny player which will hard clip DV somewhere between 2000 and 2500 nits). If it's nuking stuff above that then it's not ideal, but as the more pertinent information in most 'real world' content on UHD disc rarely lives beyond 4000 nits then I can live with that. And if a display that hits 500 or 1000 or 1500 nits 'native' brightness is mapping down the full ten-thou then that's going to kill the brightness of said scene, shirley? Dynamic mapping is great but it's still caught between how it balances brightness and range, there's no getting away from that.

But, for what this is worth, a while back I measured what lowering the contrast was doing on the ZD9 with a variety of contrast levels (using Ryan's excellent patterns) and it preserves the proper PQ curve albeit at a lower luminance output, it's why I run what I call a '4000 nit' picture mode on my telly for HDR10 that resolves 4k nits of information with a global reduction of about a third in brightness. But when you've already got a light cannon that's being viewed in the dark (as I ALWAYS do for HDR content) then this reduction in luminance isn't anywhere near as damaging as it may sound. IIRC Tyler Pruitt of Spectracal said much the same thing at some event or other.

Something like the Panny's Optimiser preserves the proper brightness curve until it has to map, whereupon it flattens the upper end into a straighter curve that meets its own needs. It's funny but it actually makes the image look less dynamic to me than my reduced-luminance '4000 nit mode', that by the Optimiser keeping the normal APL but turning down the upper end it's "squashing" the range, not in terms of resolvable information but in terms of brightness from low to high. So even though my mode has less brightness it still preserves more of a gap between light and dark, if that makes sense.

Basically Stacey, I can't wait to get dat disc!!! But wait - it doesn't actually have DV test patterns, only the montage? DV patterns are literally the one thing I wanted it most for...

[ROSS.T.G.]

Originally Posted by Stacey Spears

The only DV content is the montage. All test patterns are HDR10. Right now it is a tools issue to create DV test patterns.

On your B7, be sure to set edge enhancement to on as that means off. You also want to be in cinema and not cinema home. The B7 does not do the proper color conversion for SDR UHD content. You can see this using the colour filter in the menu and viewing the HD color bars pattern in green and red only. In HDR, it is correct.

You also want to be sure to send 444 10-bit UHD into the B7. Anything else will mess up content scaled from HD to UHD. You can test this and see the differences with the disc on the scailng HD pattern.

Yep. I do use EE and Cinema. I’ll use Cinema Home for daytime viewing of Netflix or Prime.

I always thought leaving color space on auto on my Oppo was correct. Am I correct that auto is correct for UHD but change when viewing BD upscaled?

[Geoff D]

Auto will send 4:4:4 anyway, unless it's 4K60 which maxes out at 4:2:2 in the current HDMI spec. Changing the bit depth to 10-bit is what you're after.

[ROSS.T.G.]

Originally Posted by Geoff D

Auto will send 4:4:4 anyway, unless it's 4K60 which maxes out at 4:2:2 in the current HDMI spec. Changing the bit depth to 10-bit is what you're after.

You’re the man Geoff.