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

[Stacey Spears]

Originally Posted by GREM

Have you had a chance to check Sony's players?

No, I don't currently have a Sony player. I plan to get one in the next month tot test.

[Stacey Spears]

Originally Posted by 5150z

As wxman2003 said in an earlier post, the 2016 LG Oled line has an issue tone-mapping HDR10 content. Which I learned on my 65B6. Forcing DV(on my Oppo 203)on HDR10 content improved things very nicely. It's what I use all the time now. Are these 709 to 2020 upscales and any other anomalies noticeable while viewing content, or only with patterns on your disc? These things can be so frustrating. Thank you for all your wealth of knowledge SS.

These are subtle. The OPPO has alignment issues in various modes, so it is no worse in Dolby, other than the ringing. Someone on here pointed out some real-world content (non-test patterns) where Panasonic's chroma ringing could be seen.

The 6 series has a tiny LUT for tone mapping, which impacts DV too. Even with the issues, Dolby's tone mapping is much better than LGs across all generations.

Do you use PC mode on the 6 series? It seems to stay 10-bit and 4:4:4 chroma vs. the 7-9. It is fixed again on the X. Now they just need to enable Real Cinema (5:5) and we are golden.

[Stacey Spears]

Originally Posted by FilmFreakosaurus

Unless the papers on DV for streaming leaked in the past have been updated, it looked like streaming DV was (and maybe still is) even worse or lower quality given the extreme compression necessary for internet files for Netflix, Amazon, etc., plus they do not do 12 bit Full Enhancement Layer (FEL) encodings for streaming that some 4k discs have (8 or 10 bit only).

Ben was recently posting about lowering the Amazon bitrate down to DVD levels for UHD. I watched the Wilds and it looked like DVD bitrates! The Apple TV service seems to be using higher bitrates than most.

[FilmFreakosaurus]

Originally Posted by Stacey Spears

Ben was recently posting about lowering the Amazon bitrate down to DVD levels for UHD. I watched the Wilds and it looked like DVD bitrates! The Apple TV service seems to be using higher bitrates than most.

Streaming is not technically improving it seems, just getting worse. It's truly sad to think this is our future. Don't get me started on Kaleidascape because it's a rich man's system and totally overpriced and a closed loop, so your collection is vulnerable to studio whims and market changes.

When you mean DV is using Rec 709 upsampling and not Rec 2020... when does this occur? On FEL video layer content? Is it flagged Rec 2020 on the disc (even if the actual gamut content is mostly P3), but during the DV engine stage it gets cut to Rec 709? Could the player and/or display manufacturers be implementing DV incorrectly? Panasonic players seem to clip DV highlights, for instance.

Help us understand the process. The more Dolby gets called out, the more likely they might quietly improve things in the future, for their reputation if anything.

[Philnick]

Originally Posted by Stacey Spears

. . .

However, at this point in time, I think Dolby on BD is a real disappointment.
1. They appear to be using 709, instead of 2020, upsampling so there is a half pixel vertical delay between Y and C. Verified this on both OPPO and Panasonic.
2. The 4:2:2 to 4:4:4 conversion in the Dolby engine, on the display side, is very low quality and lacks chroma resolution compared to HDR10 on the same display.
3. Dolby has chroma ringing in their upsampling, similar to Panasonic. You can see it come and go on the OPPO when you switch between Dolby Vision and HDR10.

Maybe I will invest in a Lumagen or MadVR Envy and use HDR10 on all Dolby titles.

. . .

I've never had a DolbyVision capable display - I have a JVC DLA-RS1000 (a/k/a the N5/NX5), which is the entry-level version of their lamp-based true 4K HDR10 projectors with built-in on-the-fly tone-mapping tuned to the capabilities of the projector.

That seems to be the way to go, instead of DV's pre-cooking its tone-mapping to SDR (709) based on average display capabilities.

[SeeMoreDigital]

Originally Posted by Stacey Spears

Ben was recently posting about lowering the Amazon bitrate down to DVD levels for UHD. I watched the Wilds and it looked like DVD bitrates!

Shame Amazon doesn't offer bit-rate info on their streaming service like Netflix...

[Stacey Spears]

Originally Posted by FilmFreakosaurus

When you mean DV is using Rec 709 upsampling and not Rec 2020... when does this occur? On FEL video layer content? Is it flagged Rec 2020 on the disc (even if the actual gamut content is mostly P3), but during the DV engine stage it gets cut to Rec 709? Could the player and/or display manufacturers be implementing DV incorrectly? Panasonic players seem to clip DV highlights, for instance.

Help us understand the process. The more Dolby gets called out, the more likely they might quietly improve things in the future, for their reputation if anything.

When we talk about 709 vs. 2020, there are three things we could be discussing.
1. The gamut. Think of the triangles on the CIE diagram.
2. The color space conversion matrix. This is a 3x3 matrix that converts between RGB and YCbCr
3. The chroma conversion between 4:2:0 and 4:2:2. This is the up/down sampling/scaling we talk about.

In the context of what I mentioned about DV and BD playback, it is number 3. Numbers 1 and 2 are fine from what I can tell.

Here is the chroma location differences between 709 and 2020. The white circles are the luma samples and the red squares are the chroma samples in 4:2:0. In 709, the chroma is stored in between the luma samples, vertically. In 2020, they are co-sited with luma. Get this wrong and you have a half a pixel shift (YC delay) between luma and chroma. Horizontally they are both co-sited with luma. There is one chroma sample for every four luma samples in 4:2:0.


The Dolby tools that convert 16-bit TIFFs into 10-bit 4:2:0 YUVs for encoding correctly use 2020 chroma down scaling. The BD players are incorrectly using 709 up scaling for Dolby content. For HDR10, the up scaling on Panasonic is fine. The 4:2:0 to 4:2:2 conversion in the player happens in the Dolby block for Dolby Vision. For SDR and HDR10, it uses the players decoder. Panasonic uses a different decoder for different players and I am testing on a UB9000. For OPPO, it is MediaTek.

The content is flagged as 2020. There is no P3 flagging, so all HDR content is placed inside of a 2020 container and flagged as 2020 at encode time. Unless it is 709 HDR, which I have no seen anyone use, but is not impossible. The flagging I am talking about is in the HEVC elementary stream. You can specify the gamut, matrix and transfer function. There is also a chroma_loc flag. You are supposed to use chroma_loc 0 for 709 and chroma_loc 2 for 2020. Nothing stopping you from doing something different, but there is no chroma_loc flag in HDMI, which is a problem for anyone sending 4:2:0 over HDMI. Assumptions are made.

I have not found Panasonic to clip highlights in DV on my test patterns. It matches OPPO and goes up to code value 940 (10,000 cd/m²). Do you have an example of this clipping? Which display are you using? The Z9D, for example, clips Dolby highlights on all players. You can pull down contrast and the clipping goes away. A later generation Sony display did not have this problem.

[FilmFreakosaurus]

Originally Posted by Stacey Spears

When we talk about 709 vs. 2020, there are three things we could be discussing.
1. The gamut. Think of the triangles on the CIE diagram.
2. The color space conversion matrix. This is a 3x3 matrix that converts between RGB and YCbCr
3. The chroma conversion between 4:2:0 and 4:2:2. This is the up/down sampling/scaling we talk about.

In the context of what I mentioned about DV and BD playback, it is number 3. Numbers 1 and 2 are fine from what I can tell.

Here is the chroma location differences between 709 and 2020. The white circles are the luma samples and the red squares are the chroma samples in 4:2:0. In 709, the chroma is stored in between the luma samples, vertically. In 2020, they are co-sited with luma. Get this wrong and you have a half a pixel shift (YC delay) between luma and chroma. Horizontally they are both co-sited with luma. There is one chroma sample for every four luma samples in 4:2:0.


The Dolby tools that convert 16-bit TIFFs into 10-bit 4:2:0 YUVs for encoding correctly use 2020 chroma down scaling. The BD players are incorrectly using 709 up scaling for Dolby content. For HDR10, the up scaling on Panasonic is fine. The 4:2:0 to 4:2:2 conversion in the player happens in the Dolby block for Dolby Vision. For SDR and HDR10, it uses the players decoder. Panasonic uses a different decoder for different players and I am testing on a UB9000. For OPPO, it is MediaTek.

The content is flagged as 2020. There is no P3 flagging, so all HDR content is placed inside of a 2020 container and flagged as 2020 at encode time. Unless it is 709 HDR, which I have no seen anyone use, but is not impossible. The flagging I am talking about is in the HEVC elementary stream. You can specify the gamut, matrix and transfer function. There is also a chroma_loc flag. You are supposed to use chroma_loc 0 for 709 and chroma_loc 2 for 2020. Nothing stopping you from doing something different, but there is no chroma_loc flag in HDMI, which is a problem for anyone sending 4:2:0 over HDMI. Assumptions are made.

I have not found Panasonic to clip highlights in DV on my test patterns. It matches OPPO and goes up to code value 940 (10,000 cd/m²). Do you have an example of this clipping? Which display are you using? The Z9D, for example, clips Dolby highlights on all players. You can pull down contrast and the clipping goes away. A later generation Sony display did not have this problem.

Thanks for the explanation. Trying to wrap my brain around it all.


When I say Panasonic DV clipping, I was talking about prior discussions on this forum about noticing better DV highlight performance of the Oppo players vs the Panasonic players. From what you are saying, it might be attributed to the particular DV display being used and not the players themselves. Correct?

[Stacey Spears]

Originally Posted by FilmFreakosaurus

When I say Panasonic DV clipping, I was talking about prior discussions on this forum about noticing better DV highlight performance of the Oppo players vs the Panasonic players. From what you are saying, it might be attributed to the particular DV display being used and not the players themselves. Correct?

I need a little more of the disc authored and then I can spend more time comparing OPPO and Panasonic in DV and HDR10 to see how they differ. I may find something when I dig deeper.

I did have an issue on my Panasonic where the PQ tracking boxes were not blending. I had confirmed the Optimizer was off. Then I realized one of the other HDR controls was on and maxed out. I had assumed they were all controlled by the optimizer setting. This was when I was seeing how the various settings changed the output a few months back. These controls were impacting Dolby Vision, I believe. I have looked at so many different scenarios over the weekend, I don't recall what was what at this point. :-)

For chroma alignment. When it is off by half a pixel, this impacts vertical edges. It is a slight softening to the edge. If you look at the red stripe on yellow background, one edge becomes a dark red and the other a pink'ish red. I see this type of error on TV all the time with Macys and Target commercials. Their logos make good test patterns.

The 2020 alignment issue is because they decided to change this for 2020 and it is a mention in the doc. There is a reason its not a bad idea, but the trouble it has caused was not worth the change in my opinion. Or rather the gain is not worth the pain.

[FilmFreakosaurus]

Originally Posted by Stacey Spears

I need a little more of the disc authored and then I can spend more time comparing OPPO and Panasonic in DV and HDR10 to see how they differ. I may find something when I dig deeper.

I did have an issue on my Panasonic where the PQ tracking boxes were not blending. I had confirmed the Optimizer was off. Then I realized one of the other HDR controls was on and maxed out. I had assumed they were all controlled by the optimizer setting. This was when I was seeing how the various settings changed the output a few months back. These controls were impacting Dolby Vision, I believe. I have looked at so many different scenarios over the weekend, I don't recall what was what at this point. :-)

For chroma alignment. When it is off by half a pixel, this impacts vertical edges. It is a slight softening to the edge. If you look at the red stripe on yellow background, one edge becomes a dark red and the other a pink'ish red. I see this type of error on TV all the time with Macys and Target commercials. Their logos make good test patterns.

The 2020 alignment issue is because they decided to change this for 2020 and it is a mention in the doc. There is a reason its not a bad idea, but the trouble it has caused was not worth the change in my opinion. Or rather the gain is not worth the pain.

Ugh, it seems like they're always adding more problems than they are fixing. Then it's never mind, we're moving on to 8k with its own set of unresolved issues.

[Sledgehamma]

Originally Posted by Stacey Spears

Confirmed the issue is an OPPO bug and does not occur on Panasonic.

However, at this point in time, I think Dolby on BD is a real disappointment.
1. They appear to be using 709, instead of 2020, upsampling so there is a half pixel vertical delay between Y and C. Verified this on both OPPO and Panasonic.
2. The 4:2:2 to 4:4:4 conversion in the Dolby engine, on the display side, is very low quality and lacks chroma resolution compared to HDR10 on the same display.
3. Dolby has chroma ringing in their upsampling, similar to Panasonic. You can see it come and go on the OPPO when you switch between Dolby Vision and HDR10.

Maybe I will invest in a Lumagen or MadVR Envy and use HDR10 on all Dolby titles.

I also tested player vs. tv-led. Player-led has the same issues where black is crushed more and the top two boxes on the CSE pattern don't blend, all the time! At first I thought the OPPO bug might simply be playing in player-led, but that was not the case as player-led has some other differences as well.

I did find that if I let a pattern play until the player loops it, it actually resolves the bug on OPPO. Not sure why it does but nothing else but a power cycle does.

Excellent find Stacey! If you look at my review of the Zidoo Z9X you can see that in Dolby VS10 mode the chroma upsampling is actually very good (Realter was never good at it).
I don't have much details on how VS10 really works, it looks like its doing much of the decoding (chroma upsampling is better, banding is gone). Just keep in mind that the VS10 just used the HDR10 test pattern and converted it to Dolby Vision.
So it will be extremely interesting how the output will be on a DV test pattern.
However, the Zidoo does not support "full" profile 7 playback. What it is doing for profile 7 MEL is BL+RPU and the same is planned for profile 7 FEL. Thus, the EL is always being disregarded.
Maybe VS10 is based on a later DV SDK, but I'm just guessing.

[Stacey Spears]

Originally Posted by Sledgehamma

Excellent find Stacey! If you look at my review of the Zidoo Z9X you can see that in Dolby VS10 mode the chroma upsampling is actually very good (Realter was never good at it).
I don't have much details on how VS10 really works, it looks like its doing much of the decoding (chroma upsampling is better, banding is gone). Just keep in mind that the VS10 just used the HDR10 test pattern and converted it to Dolby Vision.
So it will be extremely interesting how the output will be on a DV test pattern.
However, the Zidoo does not support "full" profile 7 playback. What it is doing for profile 7 MEL is BL+RPU and the same is planned for profile 7 FEL. Thus, the EL is always being disregarded.
Maybe VS10 is based on a later DV SDK, but I'm just guessing.

Does it support profile 8? If so, that is something we can test once I get the discs done and have the cycles.

I suspect the BD SDK has not been updated since it was released. Certainly not in two years.

[nathan_h]

I am sure I am missing something in this conversation, because my first read through the last several pages would lead to be to believe that all Dolby Vision content (whether on disk or a streaming service) is only using the rec709 color space within the rec2020 container....but that HDR10 content (disk and streaming) is using more (when the content creators choose) of the rec 2020 color space (likely out to the P3 color points in some cases). And that this is not related to bugs in any particular playback device...that this is in the source itself?

Please tell me I am reading this incorrectly.

[Sledgehamma]

Originally Posted by Stacey Spears

Does it support profile 8? If so, that is something we can test once I get the discs done and have the cycles.

I suspect the BD SDK has not been updated since it was released. Certainly not in two years.

Yes it does support profile 8

[Stacey Spears]

Originally Posted by nathan_h

I am sure I am missing something in this conversation, because my first read through the last several pages would lead to be to believe that all Dolby Vision content (whether on disk or a streaming service) is only using the rec709 color space within the rec2020 container....but that HDR10 content (disk and streaming) is using more (when the content creators choose) of the rec 2020 color space (likely out to the P3 color points in some cases). And that this is not related to bugs in any particular playback device...that this is in the source itself?

Please tell me I am reading this incorrectly.

You are reading incorrectly.

Dolby and HDR10 are the same in terms of color volume. The issue here is strictly the conversion from 4:2:0 into 4:2:2 or chroma up scaling. This is a playback bug. Please read my post from 7:42 AM where I mention three different 709 vs. 2020 discussions that are possible and which one we are discussing here. Your confusion is understandable since those that wrote the 2020 doc decided to incorporate 4:2:0 pixel layout into it.

My comments are also just for BD. Since I don't have my patterns on a streaming service, I can't tell you how they perform on the playback side.

There are many different chroma layouts in 4:2:0. DVD and BD are assumed to always use chroma_loc 0. UHD BD is expected to support both chroma_loc 0 and chroma_loc 2. For DVD (601), the progressive layout is what was adopted where chroma is co-sited horizontally and interstitial vertically. This is chroma_loc 0. This did not change when we moved to HD DVD and Blu-ray (709). The fact that they decided to make this change when they introduced 2020 has been a mess. Many older TVs use 709 up scaling for 2020 content if you play a file from USB. BD was the first format to get this right, at least in some players and scenarios.



For the new disc, we have patterns to test this in HDR10, HDR10+ and Dolby Vision (all 2020) for HD and UHD resolutions. (all chroma_loc 2) And SDR in both 709 (chroma_loc 0) and 2020 (chroma_loc 2) in HD and UHD resolutions. There are a lot of ways to get this wrong, which is why we are providing so many versions for people to test with. In my limited testing, the players seem to ignore the chroma_loc flag and instead decide which method to use based on the color space flag.

[Sledgehamma]

Originally Posted by Stacey Spears

My comments are also just for BD. Since I don't have my patterns on a streaming service, I can't tell you how they perform on the playback side.

This can be tested with the Zidoo as well as it supports profile 5. Many TVs should support playback of it via USB as well.

[Balling]

Originally Posted by Stacey Spears

The content is flagged as 2020. There is no P3 flagging, so all HDR content is placed inside of a 2020 container and flagged as 2020 at encode time. Unless it is 709 HDR, which I have no seen anyone use, but is not impossible. The flagging I am talking about is in the HEVC elementary stream. You can specify the gamut, matrix and transfer function. There is also a chroma_loc flag. You are supposed to use chroma_loc 0 for 709 and chroma_loc 2 for 2020. Nothing stopping you from doing something different, but there is no chroma_loc flag in HDMI, which is a problem for anyone sending 4:2:0 over HDMI. Assumptions are made.

Netflix does use DCI-P3 or Display P3 without any BT.2020 container in-house and HDMI 2.0 does have DCI-P3 mode in colorimetry block that can then be further clarified whether it is DCI-P3 or display p3 in HDMI Gamut metadata. Consumers do not have access to that, but if they had it is supported (just like xvYCC extended gamut linited range and sYCC YCBCr format with selectable quantisation, yes, YCbCr can also be full range in HDMI, ha). This is all in Nvidia control panel digital format menu.

Next. You can use any chroma_loc you want. And there are not just 2 of em. HEVC does specify all of them and they are supported by Mediainfo (it writes 4:2:0 (Type 2), for example). We in ffmpeg also added support for all of them. https://github.com/FFmpeg/FFmpeg/com...9674e4a083d43b Dolby Vision does redefine BT.2020 spec's chroma sitting at least for IPTPQc2 color (that is profile 5, not HDR10 compatible and it should be full range, it uses Type 0).

"H.265 (2018-02) requires top-left chroma siting (VUI = 2), if the decoded video is intended
for interpretation according to ITU-R BT.2020-2 or Rec. ITU-R BT.2100-1. Previously, H.265
(2016-12) described the default chroma siting as center left (VUI = 0)."

[David M]

Originally Posted by nathan_h

I am sure I am missing something in this conversation, because my first read through the last several pages would lead to be to believe that all Dolby Vision content (whether on disk or a streaming service) is only using the rec709 color space within the rec2020 container....but that HDR10 content (disk and streaming) is using more (when the content creators choose) of the rec 2020 color space (likely out to the P3 color points in some cases). And that this is not related to bugs in any particular playback device...that this is in the source itself?

Please tell me I am reading this incorrectly.

Stacey is talking about the part of the Rec.709/Rec.2020 standards that describe the exact alignment of the chroma pixel data, not the part of the standards that describes how saturated the colours can be.

[Stacey Spears]

Originally Posted by Balling

Next. You can use any chroma_loc you want. And there are not just 2 of em. HEVC does specify all of them and they are supported by Mediainfo (it writes 4:2:0 (Type 2), for example). We in ffmpeg also added support for all of them.

In my testing, consumer devices have been ignroing the chroma_loc flag and use the color space flags to decide if it should be 0 or 2. And they don't support anything else.

Originally Posted by Balling

Dolby Vision does redefine BT.2020 spec's chroma sitting at least for IPTPQc2 color (that is profile 5, not HDR10 compatible and it should be full range, it uses Type 0).

For Profile 7 (Blu-ray) and Profile 8 Dolby Vision uses HDR10 as the base layer for the primary video and so it must use the same chroma_loc as HDR10. ICtCp is not supported on Blu-ray.

[Stacey Spears]

Originally Posted by Sledgehamma

This can be tested with the Zidoo as well as it supports profile 5. Many TVs should support playback of it via USB as well.

Since profile 5 is based on ICtCp, we don't support it. Our tools generate YCbCr natively in 4:2:0, so we can support profile 7 and 8. If ICtCp gets adopted outside of Dolby, we will consider it. Personally I am a fan.

And as I mentioned, many TVs prior to 2017 were still using 709 style up scaling on 2020 content. It probably takes 18 months to spin a new chip, so there is always lag when changes like this occur. Especially when it was not communicated very well.

We started by using HD resolution, because HD scaled to UHD made it easier to see the error in the case you don't' have a loupe. We then discovered some TVs introduced horizontal alignment issues during scaling. The Sony HDR capable display with the speakers that looked like ears, prior to the Z9D, had this issue. I don't recall the model number.