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

[DisplayCalNoob]

Originally Posted by Stacey Spears

We would not use it on disc, we would offer it as a download to play from a USB stick, for example.

Well, I'm all in. I just need to get confirmation on DV usb support.

Is it to late to get car footage or city traffic footage in the montage for the second disc?

On the current disc, is there a backlight performance test available. Wondering if the Hz went up or down on my display or would you say the montage is the presentation considering its real world content.

[Stacey Spears]

Originally Posted by DisplayCalNoob

Well, I'm all in. I just need to get confirmation on DV usb support.

Is it to late to get car footage or city traffic footage in the montage for the second disc?

On the current disc, is there a backlight performance test available. Wondering if the Hz went up or down on my display or would you say the montage is the presentation considering its real world content.

Yes, too late. There are red tail lights on the current montage. Its a time lapse so they are stretched. I provided a more in-depth answer on why you may see orange tail lights or stop lights where you asked on another forum.

There a FALD zone counter on the current disc. We are adding a set of patterns to measure backlight resolution on the new disc. Does require a meter to measure the display.

Not sure what you are asking about the Hz going up or down and the rest. Can you be more specific?

[Stacey Spears]

Originally Posted by DisplayCalNoob

Is it to late to get car footage or city traffic footage in the montage for the second disc?

I did forget to mention that I was actually looking for a location that had a red blinking stop light hanging from a wire at a 4-way stop as I thought it would make a nice shot. I wanted the roads wet and late at night so the light would be properly exposed. Sadly we never found a location that worked for what I had envisioned, so what you wanted was on our minds last year.

It would have been a shot with the camera on the ground pointing up in terms of angle / point of view. I was also trying the find the thread on reduser from years ago where the color of tail / traffic lights was discussed as they were complaining about it being orange, which is why I have specific experience with that topic. I had actually shot the traffic light near my house as test material to have content to help improve the camera color science at the time.

[DisplayCalNoob]

Originally Posted by Stacey Spears

Yes, too late. There are red tail lights on the current montage. Its a time lapse so they are stretched. I provided a more in-depth answer on why you may see orange tail lights or stop lights where you asked on another forum.

There a FALD zone counter on the current disc. We are adding a set of patterns to measure backlight resolution on the new disc. Does require a meter to measure the display.

Not sure what you are asking about the Hz going up or down and the rest. Can you be more specific?

I'm thinking of the speed or frequency the backlights are able to transition between the different levels of luminance.

[Stacey Spears]

Originally Posted by DisplayCalNoob

I'm thinking of the speed or frequency the backlights are able to transition between the different levels of luminance.

One of the aspect of our upcoming peak luminance pattern is that it also has a rise time test.

It was meant to be used with a meter like the Klein K10 with an app that has that specific functionality. You can measure how fast it transitions from black to peak luminance.

[Stacey Spears]

Here is a Dolby Vision 2.9 version of our dynamic range low pattern. This is profile 7 in an mp4, which is probably not really a supported scenario. However, it works for my testing and development. I know it works from a USB stick directly into an LG as well as through an OPPO. Playing through the Panasonic UB9000 just results in the HDR10 base layer playing. I mention profile 7 and support only because the montage will hang in certain spots on the OPPO but plays on the LG directly.

This may not play on the LG CX or at least a report from someone with a CX and C9 says it does not work on their CX but does work on their C9. It works on my C9 and Tyler's C7.

The reason I am sharing this is that there seems to be some belief that after a Calman calibration everything below 3% is clipped or crushed and that is not true at all. This goes up to code value 100. Code value 99 is 4%. Code value 82 is be 2%. The pattern goes from code value 4 through 100. If I turn brightness up on the LG C9, I can see it goes to 64, which is 0%.

Hopefully everyone finds this useful.

[wxman2003]

Originally Posted by Stacey Spears

Here is a Dolby Vision 2.9 version of our dynamic range low pattern. This is profile 7 in an mp4, which is probably not really a supported scenario. However, it works for my testing and development. I know it works from a USB stick directly into an LG as well as through an OPPO. Playing through the Panasonic UB9000 just results in the HDR10 base layer playing. I mention profile 7 and support only because the montage will hang in certain spots on the OPPO but plays on the LG directly.

This may not play on the LG CX or at least a report from someone with a CX and C9 says it does not work on their CX but does work on their C9. It works on my C9 and Tyler's C7.

The reason I am sharing this is that there seems to be some belief that after a Calman calibration everything below 3% is clipped or crushed and that is not true at all. This goes up to code value 100. Code value 99 is 4%. Code value 82 is be 2%. The pattern goes from code value 4 through 100. If I turn brightness up on the LG C9, I can see it goes to 64, which is 0%.

Hopefully everyone finds this useful.

This file played fine on my E6 and through the Oppo 203. DV was triggered in both cases. can I use this file to set brightness correctly with DV on my E6, or should I use your UHD disc, and force the 203 to play DV, and set the brightness using the brightness pattern on the disc?

[Stacey Spears]

Originally Posted by wxman2003

This file played fine on my E6 and through the Oppo 203. DV was triggered in both cases. can I use this file to set brightness correctly with DV on my E6, or should I use your UHD disc, and force the 203 to play DV, and set the brightness using the brightness pattern on the disc?

This pattern can be used for setting brightness in DV. Code value 82 is 2% and code value 99 is 4% on our normal brightness pattern, if that helps. 64 would be 0%.

My C9 is correct the default 50 for brightness. The 6 series was their first for DV? Pre auto-cal days as I recall. Or at least different.

[wxman2003]

Originally Posted by Stacey Spears

This pattern can be used for setting brightness in DV. Code value 82 is 2% and code value 99 is 4% on our normal brightness pattern, if that helps. 64 would be 0%.

My C9 is correct the default 50 for brightness. The 6 series was their first for DV? Pre auto-cal days as I recall. Or at least different.

Correct with the 6 series. DV can not be calibrated on the 2016 modes. HDR10 can be calibrated.

[wxman2003]

Originally Posted by Stacey Spears

This pattern can be used for setting brightness in DV. Code value 82 is 2% and code value 99 is 4% on our normal brightness pattern, if that helps. 64 would be 0%.

My C9 is correct the default 50 for brightness. The 6 series was their first for DV? Pre auto-cal days as I recall. Or at least different.

On my E6 with brightness at 50, values 64 thru 67 remain black, with 68 and above flashing. I guess that's the best one can do with an E6 since it can't be calibrated. Is code 67 close to 0.5%?

[mrtickleuk]

Originally Posted by wxman2003

On my E6 with brightness at 50, values 64 thru 67 remain black, with 68 and above flashing. I guess that's the best one can do with an E6 since it can't be calibrated. Is code 67 close to 0.5%?

Stacey explained here the formula.

Putting the maths into excel I got (and I apologise massively in advance if this is wrong!)

Code value : Percentage (of the range of [940-64=876] values)
 64 : 0.000%
 65 : 0.114%
 66 : 0.228%
 67 : 0.342%
 68 : 0.457%
 69 : 0.571%
 70 : 0.685%
 71 : 0.799%
 72 : 0.913%
 73 : 1.027%
 74 : 1.142%
 75 : 1.256%
 76 : 1.370%
 77 : 1.484%
 78 : 1.598%
 79 : 1.712%
 80 : 1.826%
 81 : 1.941%
 82 : 2.055%
 83 : 2.169%
 84 : 2.283%
 85 : 2.397%
 86 : 2.511%
 87 : 2.626%
 88 : 2.740%
 89 : 2.854%
 90 : 2.968%
 91 : 3.082%
 92 : 3.196%
 93 : 3.311%
 94 : 3.425%
 95 : 3.539%
 96 : 3.653%
 97 : 3.767%
 98 : 3.881%
 99 : 3.995%
100 : 4.110%

As to how the rounding rules are written, no idea. I know that for the 8-bit SDR world Calman has one way, and maths has another way which has caused issues with pattern discs in the past.

[Stacey Spears]

Originally Posted by wxman2003

On my E6 with brightness at 50, values 64 thru 67 remain black, with 68 and above flashing. I guess that's the best one can do with an E6 since it can't be calibrated. Is code 67 close to 0.5%?

I guess the question is are you looking for % of code value, which is what mrtickleuk provided above. (Which looks correct) Or do you want to know what nit level code 67 is? When you look at our window pattern section where we have 5, 10, 15, 20, etc... Those are in % of code value. In the analog days we called them IRE on a WFM. Some may even call them % stimulus.

BTW, until your post, I forgot just how much that the 6 series crushes black. I thought, at one point, everything 2% and below was gone on our brightness pattern. I need to re-check next time I am at Don's house as he has the C6. Its the last 3D display and I have to hear about no one making 3D displays, blahh, blahh, blahh. :-) I tease him about this, but I really do understand his frustration. Lets not tell him though, its more fun to tease him about it.

Originally Posted by mrtickleuk

As to how the rounding rules are written, no idea. I know that for the 8-bit SDR world Calman has one way, and maths has another way which has caused issues with pattern discs in the past.

There are no rules on which way to round in video. Calman itself has standardized on how to round internally. Even then, it was debated early on which way to round. At the moment, I don't remember what ended up being used as the internal standard, but I just remember that I had to learn about bankers rounding. We are fans of dither. What we do is more noise-shaping / error diffusion that we then randomize to deal with fixed patterns.

I posted these pictures years ago, but will share again. There will be four images.

10 to 8-bit using the noise shaping / error diffusion that is randomized.


10 to ~2-bit using rounding. If you look closely, you may see some banding.


Same but with an RGB histogram.



10 to ~2-bit using random noise dither.


Same but with an RGB histogram.


10 to ~2-bit using the noise shaping / error diffusion that is randomized.


Same but with an RGB histogram.


For nits, code value 67 is 0.000342874 nits.

For reference
509 - 99.91279849 nits. Usually just round to 100.
674 - 600.2933939 nits. Usually just round to 600.
723 - 1004.191904 nits. Usually just round to 1000.
789 - 2003.692695 nits. Usually just round to 2000.
855 - 4014.717702 nits. Usually just round to 4000.

The reason I chose the values above is because those are the nit levels we target on disc. Some patterns use that code value and others dither to get a closer nit value. Depends how it is constructed. I will use 1000 nits as an example:

10-bit code value.
722 - 993.7307592
723 -1004.191904
724 -1014.762441

You can see the large change with just 10-bit precision and why we use dither to try and target 1000 using multiple code values. PQ was designed with 12-bit in mind. Where each code value was less than one JND.

Whenever Don and I have a discussion and I say I want something at 10%. He asks if that is 10% linear, 10% of code value, 10% out of my mind, etc....

We made some changes recently to SDR labels. On the UHD disc we used nits for SDR too. Well, the 100 nit SDR is only true if the TV is set to 100 nits since it is a relative system. So we changed SDR on the next disc to be % linear. So 50% linear would be 50 nits if the display really is at 100 nits for SDR. But if it is at 200 nits, then 50% linear would be 100 nits. Hope that makes sense. We are always trying to improve what we do. There is that fine line between being correct and being pedantic.

The one area we are still sloppy is when writing RGB vs. R'G'B'. The prime donation means its in non-linear space. When most of us talk about RGB, we are really talking about R'G'B', its just a pain to write like that all the time. In a real SMPTE paper, we would be more strict and use the prime when talking about non-linear. We debate this in our disc articles and menus. How pedantic / correct should we be?

The other day I read an email exchange between two people. One was using RGB to mean R'G'B' the other really assumed they meant linear RGB and kept trying to correct his logic, which was correct, just assumptions and notations were the problem.

[Stacey Spears]

Originally Posted by DisplayCalNoob

Well, I'm all in. I just need to get confirmation on DV usb support.

Is it to late to get car footage or city traffic footage in the montage for the second disc?

On the current disc, is there a backlight performance test available. Wondering if the Hz went up or down on my display or would you say the montage is the presentation considering its real world content.

I found the traffic light I had shot, which was in 2014. Here is the same shot with three different color options. These were decode options when converting the RAW image into RGB. These are in the order the color processing had evolved. The last one was in 2014. If I re-processed with today's algorithms, I am sure they would be better. (fingers crossed)

One


Two


Three

[Stacey Spears]

Below are scaling examples. The first was the 480p scaled to HD, I believe.

They need to be viewed at native size to avoid other scaling issues and to see the differences.

1. Split screen of both styles


2. Traditional scaling


3. Post smoothing algorithm


Here is the original inspiration for the spiral in our test pattern.

1. Original image unscaled


2. Traditional Scaling


3. Post Smoothing


1. Traditional Scailng


2. Smart Edge Scaling

[Stacey Spears]

Originally Posted by wxman2003

On my E6 with brightness at 50, values 64 thru 67 remain black, with 68 and above flashing. I guess that's the best one can do with an E6 since it can't be calibrated. Is code 67 close to 0.5%?

When rounded to 8-bit, 67 would be 16.75, so it may round to 17. 63-66 all will most likely round to 16, which is black or 0%.

Something to keep in mind.

[gkolb]

Hi Stacey,

On my iPhone the first traffic light photo is displayed, the 2nd and 3rd are just little placeholder boxes. Any idea why 2 and 3 don’t display? FYI I’m rocking’ an old 6S, if that matters.

[wxman2003]

Originally Posted by Stacey Spears

I guess the question is are you looking for % of code value, which is what mrtickleuk provided above. (Which looks correct) Or do you want to know what nit level code 67 is? When you look at our window pattern section where we have 5, 10, 15, 20, etc... Those are in % of code value. In the analog days we called them IRE on a WFM. Some may even call them % stimulus.

BTW, until your post, I forgot just how much that the 6 series crushes black. I thought, at one point, everything 2% and below was gone on our brightness pattern. I need to re-check next time I am at Don's house as he has the C6. Its the last 3D display and I have to hear about no one making 3D displays, blahh, blahh, blahh. :-) I tease him about this, but I really do understand his frustration. Lets not tell him though, its more fun to tease him about it.

I was trying to figure out how much my E6 was clipping on the low end with DV. Apparently it really does not crush that much with DV if I can see code level 68 flashing in a dark room. Definitely not 2%.

[Stacey Spears]

Originally Posted by gkolb

Hi Stacey,

On my iPhone the first traffic light photo is displayed, the 2nd and 3rd are just little placeholder boxes. Any idea why 2 and 3 don’t display? FYI I’m rocking’ an old 6S, if that matters.

Sorry, not sure. Does the same thing happen with the other groups of images?

I see them in Chrome on Windows. Have not even tried on an iPad yet.

I will say they are magical and amazing and if you don't see them you are clearly missing out! ;-) Kidding of course. The 2nd image was over saturated to point of color clipping. The 3rd is a balance in between the first two. More saturated but not clipped out.

[p5browne]

Originally Posted by Stacey Spears

I found the traffic light I had shot, which was in 2014. Here is the same shot with three different color options. These were decode options when converting the RAW image into RGB. These are in the order the color processing had evolved. The last one was in 2014. If I re-processed with today's algorithms, I am sure they would be better. (fingers crossed)

One


Two


Three

Same here on my home desktop PC. Main picture, but placements only for the other 2 pics.

[Stacey Spears]

Once we get the podcasts rolling, I was thinking maybe once every 4th or so podcasts, we do a customer episode. It can be a single customer or multiple. Idea is you get to ask all the questions you want. Is this something any of you are interested in?

You would probably send a list of rough questions in advance so we can prepare for them. We would not restrict questions only to those, but for complex ones, it would give us time ahead to prepare for anything super complicated.

In fact, if people can start listing questions you would like to see answered or subjects you would like to see covered. Actually, maybe before we record a podcast with customers, whomever joins that week can solicit question on the forum to make sure we cover those topics for those that either can't, or don't, want to be on camera.