[opendtv] Re: Sparkle

  • From: cooleman@xxxxxx
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Sat, 03 Dec 2016 14:36:38 +0100

HDR can be coded in 8 bits as well as 10, 12, or more. BBC did experiments,though their production is 10 bits so HLG was a good fit in its standard production environment. Why would it be derived from the 10 bits version, straight from post is more likely. As for distribution Dolby sticks to 12 Bits, but its PQ is used in 10 bits (SMPTE ST2084/ITU BT.2100), practically it is all focussed on 10 Bits distribution, but in autoconversion at display and perhaps ahead of Mike's distribution boxes it starts and thus ends with 8 bit material.

Craig Birkmaier schreef op 03-12-2016 14:03:


A point of clarification.

I'm not able to find any formal definition of HDR8. I assume that this
is 8 bit luma and color samples derived from 10 bit HDR source. Is
this correct?

If yes, then you are saying that a good mapping of HDR source into 8
bits can look almost as good as HDR10?


On Dec 2, 2016, at 10:35 AM, Mike Tsinberg <Mike@xxxxxxxxxxxxxx> wrote:


We compared very carefully HDR10 and HDR8 on the same content and same screens. Yes, if you know where to look you can see regular differences between 8bit and 10bit pictures with or without HDR. However, the differences between HDR8 and HDR10 is extremely hard to see.

Best Regards,
Mike Tsinberg

-----Original Message-----
From: opendtv-bounce@xxxxxxxxxxxxx [mailto:opendtv-bounce@xxxxxxxxxxxxx] On Behalf Of Craig Birkmaier
Sent: Friday, December 02, 2016 7:19 AM
To: opendtv@xxxxxxxxxxxxx
Subject: [opendtv] Re: Sparkle

On Dec 1, 2016, at 10:43 PM, Manfredi, Albert E <albert.e.manfredi@xxxxxxxxxx> wrote:

Good for you, Craig. But it proves nothing. I will explain below what it takes to claim you have an HDR display. You stated that smartphones and tablets are available with HDR displays. All you have to do is support your claim. Or just say ooops, my bad. Should be simple enough.

Ooops, my bad...

There, does that make you feel better Bert?

The reality is that displays are moving beyond the old 8 bit definition of luminance and color optimized for CRT displays. Do they meet the new definitions of what an LCD or OLED HDR display should offer?


Did every NTSC display fully support the capabilities of NTSC? Does every HD display fully support the capabilities of HD (pick a definition as there are several)?

We are in a transitional period. There are several existing/proposed standards for HDR; and several for WCG. There are many sets being sold today as HDR capable that do not meet all of these specs. All one needs to do is look at LCD and an OLED displays that claim to offer HDR side by side to understand that there will be a range of products offering varying levels of performance. And then there is the minor matter that it may not matter based on the viewing conditions where the display is installed.

I wrote:

What you are missing is that there is a continuum between today's 8
bit luminance representations and the specifications that Mark was
citing for HDR.

Except that there are specific definitions at play now. This first, very recent article, only tells half the story, but it's a start.

Yup. And there are specific products being sold that claim to offer HDR that are very different.


It states that SDR displays, at their brightest, provide 300 to 500 nits. Compare that to the displays of smartphones and tablets, to see if they are meaningfully better. It also states:

A meaningless definition. There are SDR displays for outdoor use that have more than 8500 nits of brightness.

Clearly modern mobile devices have displays with HD or better pixel resolutions. Some support improved color gamuts; some offer more than 500 nits brightness. You cannot say what a display is today based on definitions. What you can say is that displays now have the ability to conform to multiple standards.

"In April, the UHD Alliance - an inter-industry group made up of companies like Samsung, LG, Sony, Panasonic, Dolby, and many others - announced the Ultra HD Premium certification for UHD Blu-ray players. This benchmark sets some baseline goals for HDR, like the ability to display up to 1,000 nits of brightness and feature a minimum of 10-bit color depth. Both HDR10 and Dolby Vision meet the standards set by the certification, but how they go about it varies greatly between the two."

Sounds like a bunch of technophiles arguing about the future. Oh wait!

That's what's happening. Maybe I'll come up with a spec for HDR.

They should also have provided an idea of the dimmest possible content, but at least you have the 1000 nits parameter to work with. The wider color gamut might be associated, but you can have WCG without HDR, in principle. Getting the super bright output from your small hand held devices is a much harder nut to crack.

But that nut is being cracked. The nits requirement is related to the display technology, and is heavily influenced by the amount of ambient light where the display is used. Here's one clue that may help Bert over this misunderstanding:

Why is the peak brightness for OLED based HDR only 500 nits, while for LCD it is more than 1000?

I guess it is not related to the actual screen brightness or the requirements of the human visual system. If 0 to 500 nits is good enough for HDR on an OLED display, why do we need twice as much peak brightness for LCD?

Explain this Bert?

This article is better. It quantifies the range of brightness needed for HDR:


It says that SDR covers a range of nits of 3 orders of magnitude, whereas HDR provides 5 orders of magnitude. So, something like 0.3 to 300 nits is SDR. To qualify as HDR, you would need to be able to display a range 0.01 to 1000 nits, for example, or 0.005 to 500 nits. This is what I'm talking about. Now you have the information needed to prove whether smartphone and tablet displays can achieve this range, rather than just barely meet the same specs as SDR displays.

The Sony article is informative in that it places all of this in the context of what the human visual system can perceive under various ambient conditions. The nits requirements above are related to delivering HDR with various display technologies under various ambient conditions.

There is no single point stand here Bert; there is a continuum that is related to source, emission coding, and display.

What is abundantly clear is that you need at least 10 bit luminance to deliver the dynamic range needed for HDR. The FACT that this visual information can be displayed at varying light levels based on the display technology is what you should be focused on.

A mobile device that is capable of displaying 10 bit luminance and color - I'm talking about the internal processing paths and standards supported by the device, not the display - can deliver an improved viewing experience. The quality of that experience will depend on the display technology, the calibration of that display, and the ambient lighting conditions.

The same is true for any TV.

I wrote:

I would note that there is little difference between "real" HDR image
acquisition and the three exposure technique used in many smartphone
cameras - both acquire the HDR information.

This is what convinces me you have yet to understand what HDR display is. Funny thing is, I have never noticed any confusion on dynamic range when it comes to audio.

I'm not talking about HDR display Bert. I'm talking about capturing an image then optimizing it for any display. Take another look at that Sony article Bert. It does an excellent job of explaining (with pictures too!) how we process images captured with High Dynamic Range, then map them into limited representations for emission. As I said, there is little difference between what a video camera and a smartphone camera do in the real world. In both cases we can acquire more dynamic range than a device with an 8 bit display can deliver. In both cases a human or an algorithm determine how the higher dynamic range data is mapped into a more limited "display space."

In principle, if you play the little game of taking three images on an SDR camera, IN PRINCIPLE (I repeat), you can process the images to send to HDR displays. But that is **not** what these hand held appliances are doing. The fact that you don't see the difference is truly weird, because the processing these devices *are doing* is exactly the opposite of what it takes to send images to HDR displays.

Correct. It is more closely related to what a processing system that starts with HDR data does with this data in order to display it on a display with 8 bit luminance.

Let me repeat this: HDR mode in a smartphone camera is used to capture high dynamic range in a *scene*, and then compress that high dynamic range so it can better be displayed on an SDR display. The bright parts are dimmed, the dim parts are brightened. There is no way you can pretend this is HDR display.

Yup. Just like EVERYTHING we watch on our fancy HDTVs Bert. You are showing your ignorance of how imaging systems are used to create an optimal product for use across a range of displays with varying capabilities.

This is a very complex subject with MANY, MANY variables. You cannot neatly fit this into a box and say "this is what it takes to be HDR."

Clearly display technology has evolved and improved. And it will keep improving. What is really in play here is how do we deliver better quality images with HDR and WCG. And how do we do this in a manner that is compatible with billions of screens with varying capabilities.


You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line.

You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line.

You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.

You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: