No problemo in cracking the 700,000 viewership mark with 100,000 views in one month’s time, so a holiday brain teaser for TV technical gurus (Charles?), or those who aspire to be, question is -......
Poynton, are you reading or playing again on twitter; if nobody chimes in I'll have to give out the answer myself.
I really am glad it's not just me who notices these things.
Geoff, good luck today.
Recently, I heard a wise man say…."One of the dangers of the internet is that people can have entirely different realities. They can be cocooned in information that reinforces their current biases."
Bravo Paul , for bravely stepping up when nobody else has and giving it a good shot as the question posed is indeed difficult, geared toward engineers and scientists the likes of Poynton.
But sorry to say, that’s not the answer I’m looking for, which relates more to a signal format being included in 2020, but not making it into the later rec. 2100. Actually, sometime last Spring I think, you previously thanked me for the applicable information in regards to how the human visual system functions, if that rings any memory bells.
PeterTHX, I added the intro video clip to above post #3705 to whet your appetite for perhaps catching the next episode in the series hosted by Joe (his son got him interested in the sport).
not mentioned in the respective next to the last bullet point ^ is that thee upcoming Rose Parade is planned with scripts in HDR (HLG).
Graphics are easier to do with the metadata free approach, as mentioned last summer - https://forum.blu-ray.com/showthread...s#post13745946
Not in HDR, but a behind the scenes look at the preparation….
Curious. Both should be HEVC so I wonder what the difference is beyond compression? Or does he mean just that, they've got to do two different encodes, a high bitrate and a much lower one? Well, that's to be expected if so.
Um, but we know why and I'm surprised that Big John isn't aware of this: the dual layer base + enhancement system is what makes the encoding for disc a LOT more complicated.
Let's say you've got your 12-bit Dolby Vision master. Most streaming delivery methods are a single layer version, that is to say it's comprised of Dolby Vision and that's it. They don't have to provide any backwards compatible streams buried within that DV version if they don't want to, as the service will usually detect what you've got (SD, HD, UHD etc) and output the appropriate version direct from source. So when you're encoding that master for streaming you just select a single layer 8-bit or 10-bit delivery stream (remember, the 12-bit output is reconstructed at the display end) and that's basically it.
But the method for compressing onto UHD disc is very different because of the BDA's mandate that if an optional HDR format is used on a disc (Dolby, Philips) then an HDR10 version MUST also be included. Obviously it's not prudent to do two separate encodes as there's nowhere near enough space, so they devised a method which analyses the 12-bit DV master and the 10-bit HDR10 trim pass (itself directly derived from the DV master) to create a suite of difference data (not just the dynamic metadata but the 4:2:2 colour, extended bit depth etc). This difference data is then encoded within a 1920x1080 enhancement layer, the quarter resolution indicating that it adds a nominal 25% overhead to bitrates, which is compressed to disc alongside the HDR10 layer.
So far so straightforward, some might say. But disc also has other technical issues to consider, like with the unique LTR/HTR zones where a disc can use the maximum 128 Mb/s bitrate for the outer 92% BUT the inner 8% of a 33GB layer must use a lower bitrate so as to not exceed the permitted RPM. Sure, you could pad out the first layer easily enough so that the movie starts on the 92% but a triple layer disc will likely have to contend with two more forays into the LTR zone, one as the second layer returns home (outer to inner) and another as the third layer begins (inner to outer). (Perhaps those inner 8%'s can also be padded out but you'll lose 8% of space on each layer as you do so.)
The point is that it's easy enough to manage for a single layer encode but when you've got one layer riding atop another that adds 25% to the overall bitrate then they both need to be carefully managed throughout the compression process because you don't want to starve the base layer of bits. It's nailing this process which has been at the heart of the problems with getting DV onto disc, even a long-standing authoring house like Deluxe only got certificated by Dolby to handle this specific process very recently.
C’mon folks don't be stingy, assign a few more thank yous to Geoff, as that ^ post is yet another example of the Blu-ray.com rank and file knowing more technologically than the press.
Someday if 12-bit decoders become universally available in consumer electronic devices then Dolby Vision could just use 12-bit ICtCp plus playback metadata therefore making for simpler implementations. The content would already be in the intermediate format leaving the complex tone mapping to the display stage.
Quote:
Originally Posted by Geoff D
even a long-standing authoring house like Deluxe only got certificated by Dolby to handle this specific process very recently.
Tomorrow heading down to Coronado to celebrate the New Year’s and do some mountain biking in the Laguna Mountains so in case I don’t get a chance to check in again, everyone have a Happy New Years.
Because it remains still too revolutionary, what specific parameter value made it into Rec. BT.2020 but not into Rec. BT.2100 (https://forum.blu-ray.com/showthread...post14369414)? The answer is not in that slide series.
Answer –
The same parameter that produced a BIG debate with regards to whether or not to include it in BT.2020, as revealed to fellow member Ron Jones back in April of 2013, namely constant luminance (CL) signal format - https://forum.blu-ray.com/showthread...ce#post7361160
Tomorrow heading down to Coronado to celebrate the New Year’s and do some mountain biking in the Laguna Mountains so in case I don’t get a chance to check in again, everyone have a Happy New Years.
Not in HDR, but since I mentioned it, last chance to watch the award-winning documentary Blood Road for free on Red Bull TV. Available until December 31st.
To view before the no-charge deadline visit: https://www.redbull.tv/video/AP-1RVR...W11/blood-road. Ladies can ride darn well too , they proved that the week of CEDIA 2017 with us guys.
C’mon folks don't be stingy, assign a few more thank yous to Geoff, as that ^ post is yet another example of the Blu-ray.com rank and file knowing more technologically than the press.
I've thanked your post, not to boost my own ego but to thank you outright because a hefty portion of that post (and plenty more posts besides) has come from reading your own missives over the last few years.
IF (a big IF here) you were playing a Dolby Cinema graded file, in a very dark room with a large screen, then yes, a 108 nit peak white is the spec, and a PQ electro optical response curve, and Rec 2020 color space, but coded as XYZ in the file. If you don't know what that all means, look it up.
PQ is NOT tone mapping. Gamma was a simple log curve that turned a signal voltage into a certain level of brightness. A Gamma of 1.0 would be linear, twice the signal would be twice as bright, but even back in the early days of TV, they knew that the human eye was more sensitive to the change in brightness at low levels than at high levels. The early Gamma curve was basically the response of the imagine tube, which fairly well matched the response of a TV picture tube. With the advent of digital recording of video, and non tube imaging devices, and now LED and LCD displays, they had to come up with a map for devices to follow so Gamma became a standard. Rec 709 HDTV uses a gamma somewhere between 2.2 and 2.4 depending on where the content was graded. Even this "standard" is not very standard. Early HDTV was 2.2 but most are now grading in 2.4, so where do we calibrate the display??
PQ replaces Gamma in most HDR systems. It is no longer a single log curve. It is a complex relationship that was developed through extensive testing. If you google it, you will find papers about it. But by carefully mapping how the human eye responds to changes in light intensity, PQ can cover a much wider range of dark to light with steps that the human eye can't see while not needing more bits. 12 bit Gamma will show visible steps, where just 10 bit PQ will not. it is not necessarily more bits, it is using the bits better.
The other thing that was learned through testing is that the size of the image does matter. A 60 foot wide screen at 108 nits will be very bright. Especially in a dark room, and even more so when the image has been dark for a while and the light jumps to 108 nits. Watch the scene in "The Jungle Book" when he comes out of the dark cave into the sun. In the cave, you could still make out the walls and floor, it was not completely black, but it was probably under 2 nits with detail. The sky out of the cave, was close to 100 nits, it is a hug jump and your eyes need to adjust. Trying to duplicate the same sensation on a small screen in a home is very different. Almost no home environment, outside of a dedicated theatre will get your field of view as dark as it was in the theatre, (yes, even with the legal isle lights). To be able to make out the cave walls and the floor etc., it can't be near as dark as it was in the Dolby Vision Cinema version. And having that 65 inch TV go to 100 nits would not be much of a jump. You probably would not even squint like you did in the cinema. And just sitting close to the TV still does not make the same sensation.
Quote:
That's a main factor, not for us that can only follow but for movie studios and movie makers which are the ones that will choose how to market their works in the home market. Do they really need to aim at six stops of highlights in order to convey a cinematic experience? I guess not.
Let's do a little math, if a face is at 20 nits, very typical, then a 6 stop brighter highlight would be 1,280 nits. that is still far less than the real world. In Dolby's white paper, they took a picture of a simple flower in sunlight. They chose a few spots on the flower, and actually used a lab grade color meter to measure the real life image. The dark area behind the flower was still at 145 nits. The dark area at the center of the flow was 188 nits, the darkest red of the petals was 2,300 nits, and the brightest yellow petal was 14,700 nits. This is what you would see walking outside and seeing a real flower during the day. So even a 10,000 nit display needs to compress this every day image. When you are outside on a normal sunny day, your eyes need to adjust and bring everything down to allow you to make it all out. Then you go into a basement and it can take minutes for your eyes to adjust, but then you can once again make out things in the dark shadows. The human eye is an amazing thing. The iris makes very quick adjustments, but can only cover a few stops. I saw in an article, they calculated the effective F of a human eye as a low of 2.1 with the iris fully open to an F of 8.3 with the iris fully closed. This will vary from person to person a bit, but we see a little over 6 stops in the fast iris changes. The retina has a range of about 6 to 8 stops at any given moment in time, if we call it 7 for the retina and 6 for the iris, then you have a contrast range of about 8,000 to 1 for a short term event. Anything brighter makes you squint, and darker crushes into black. But if you are in a dark place for any length of time, or move to a brighter place for a length of time, then your eye will start to adjust the sensitivity of the retina. The short term range will vary some, but it stays close to a 10 to 14 stop range from black crush to white clip even with the iris adjusting, and withing the same image, it can be as little as 6 stops from the darkest to lightest detail you can perceive at any given moment. So why do we even need HDR then? Well, because we are presenting a complex image and the viewer can look at a dark area, or look at a bright area, and they still want to see the details. And when the scene does dive into the dark, it does not take long at all for the eye to adjust and see that a black screen is actually a light grey. With 20 stops available, when it goes from a nice 20 nit face to black, it is very black, and in most cases, the image will go to something else before your eye can adjust and start to see any glow from the projector or an HDR flat screen.
Quote:
That the key issue. The colorist/grader establishes a brightness range and relative levels between pure black and pure white (the "normal range") that shouldn't be messed with NO MATTER what the highlight capabilities of the TV are. In other words, all modern TV sets can replicate the brightness range that the studio chose for the main content of movies, nominally 1-100 nits, and that range should be kept AS MASTERED, in the home viewing environment.
Again, you are mixing up a cinema screen and a Flat screen TV, as well as cinema content vs consumer content. Unless you have a digital cinema server, and the KDM keys for a Dolby Cinema version, you will not be getting a package that is graded for 108 nit peak PQ. The DCI cinema versions will be Gamma graded for peak of 48 nits, but for a large screen in a dark room. ALL HOME VERSIONS ARE GRADED DIFFERENT. This is a big reason I did not want to write this in the first place, and I should probably stay out of it. You can't directly translate a cinema viewing situation into a home. I have setup DCI servers and projectors on 12 to 15 foot wide screens in very nice well light controlled home theatres for some very rich people. This is NOT Dolby Cinema, just DCI mind you, so we are working with 48 nit graded content. In most cases we end up having to run them closer to 55 to 70 nits to look right on the smaller screen, even though, the picture height to distance is most likely closer than an average movie theatre. And when we setup the system to play Blu Ray disks and HDTV, we end up at 80 to 100 nits. A bit lower than a flat TV because the screen is still much bigger. I don't know if there is any well written paper on how the size effects the apparent light, but we see this in room after room, it is real. For a normal size movie theatre, the DCI spec of 14 fl (48 nits) translates pretty well from the color grading suite to the cinema. And for Dolby Cinema, the Grading room to the theatre is also a close match.
The smallest rooms I have seen a Dolby Cinema setup in are still over 30 feet wide. They know to not try to grade on a small screen for a cinema release. The home versions are graded on 40 to 60 inch screens, in a room with some ambient light, to simulate the typical home. In the end, once it goes into the home, there are some who will try to run it at spec, and others who will crank up contrast and color etc. With Rec 709, it is pretty obvious it is limited and some enhancement can look better. The home version of Dolby Vision does the best it can to replicate what the color grader saw while doing the master. The much larger range and the tone mapping between the grading monitor and the chosen display, makes it so it should look as good as the display is capable of.
Quote:
PQ tone-mapping affects the normal range if there is a mismatch between the source data and the TV brightness capabilities. Does that make any sense?
As I tried to explain, PQ has nothing to do with tone mapping to a different TV range. PQ just maps how much light a certain digital code value should produce on the screen. Please, if this does not make sense, do some searches on Gamma vs PQ. The color tone mapping and grey scale mapping is a different situation all together.
I am not sure if I can post a link, but here is the Dolby Vision white paper https://www.dolby.com/us/en/technolo...whitepaper.pdf
If it works, this should start a download of the white paper PDF file from Dolby.
A simplified version of how the consumer setup works is something like this...
They grade the full HDR range version first. On a Dolby Vision spec monitor capable of up to 4,000 nits, full Rec 2020 color space, and great contrast down to something like 0.005 nits. After that version is approved, they then create the rec 709 color space limited version with 100 nits peak white and about 0.12 nits for black. The raised black is the result of the gamma curve counting down from peak white to the lowest code value. If your image goes darker on Rec 709, it is not properly following the Gamma curve.
This is where TONE MAPPING does it's work. The meta data contains the difference between the full HDR and the limited Rec 709 versions. In Dolby Vision, this data can be different for every frame. The content creator decided what part of the image needs to be produced at what level. In one scene, maybe it works just fine to let the highlights peak out and even clip, but the shadow detail is preserved, but in a later scene, the shadows can clip into black, while you can make out the details of light glinting on something bright. So the choice of squeezing the range, clipping only the top, clipping only the bottom, or any combination of them is all up to the colorist, on a scene by scene, or even frame by frame basis. The processor in the TV, knows the range the display can produce, and the range of the original full HDR version. But if it can't produce the full version, it uses the meta data from the limited version to map it down in the same way the original colorist decided was right at any given moment. This is done in both the grey scale brightness range as well as the color saturation. Any colors that exceed Rec 709 will need to be reduced in saturation to fit in the limited version, but what about a Display that can hit P3, or even more? The same process is used. Did the colorist clip the color, reduce the range, or drop it even lower to allow another color to still be more saturated in the next shot?
Without the dynamic data to explain what to do with each frame, it might not look as the colorist intended. You can certainly make a demo clip look amazing for any given HDR display, but if it was not created for that display, how well is it going to translate? This is where Dolby Vision differs from the rest of the pack. Maybe the shadow clipping in one scene does not matter, but what about the scene where you can still see the black gun barrel in the shadow? With fixed meta data, are you going to lift the black of every scene for the one to still work? Are all the colors going to be reduced saturation so the one shot of a red rose can be more vivid? No database in a TV can know what a director is going to do in a movie that is not even made yet. If a display CAN produce the full image of 4,000 nits down to 0.001 nit, accurately follow PQ, and produce the full Rec 2020 color space, then tone mapping and dynamic meta data are not really needed. But then again, this would be if you had the original full range PQ signal, but in most cases, it is not sent like that. To make the system backwards compatible, the signal is actually sent BACKWARDS. The main video signal is Rec 709, so if a display does not even know what Dolby Vision is, you get a properly graded Rec 709 image that looks as good as what current Blu Ray can do. But if the processor in the display understands the Dolby Vision meta data, it can fully recreate the original 4,000 nit Rec 2020 signal by putting it all back together. This is why the display needs the processing power. And why even a full 4,000 nit display needs to have the dynamic meta data to know how each frame needs to be rebuilt. And if the display can't match the original grade, it also knows how the original content creator wants it to be limited.
Now does it make sense?
PLEASE PLEASE at least skim the white paper.
There is a lot more to making a true HDR system than just cranking up the contrast and brightness. Sure, many Rec 709 programs can look better with the brightness and contrast turned up a bit, but it can be hit or miss and ruin the look of some content. Content creators and display companies are going more and more with Dolby Vision because it actually addresses the entire chain from content creation to the viewer.
And please stop trying to directly compare a 60 foot wide cinema screen to a 65 inch diagonal TV. It is like apples to grapes. There may be red versions of both, but that is about the end of the similarities.
It's worth mentioning though that the backwards part of Dolby Vision is dependent on what encoding profile is used, the image is not encoded as 709 in a single-layer HDR or dual-layer UHD Blu-ray version. That's reserved for the vaunted SDR + HDR dual-layer version that Dolby kept talking about even though it appears to be rarely used, probably because the content providers who currently use DV aren't all that bothered about an accurate SDR downconversion, and as we know the BDA made sure that the base layer was HDR10 when it comes to UHD Blu-ray.
Yes, a 709 trim pass is what's used to create the dynamic metadata (the TV's processor knows what specs it can handle so it compares that to the 709 metadata to derive the unique difference) but that pass itself is not encoded into the final file unless it's using that specific SDR + HDR profile. I think.
I've thanked your post, not to boost my own ego but to thank you outright because a hefty portion of that post (and plenty more posts besides) has come from reading your own missives over the last few years.
Good to know.
So far, I’ve only watched the 1st half of the recent match at Selhurst Park and based on that I was fearing you were holding a grudge (Alexis was offside and I truly believe distracted the keeper) and thusly you were not feeling Glad all Over - https://www.youtube.com/watch?v=gRzBNGGb-zc#t=1s
All countries
Argentina
Austria
Belgium
Brazil
Bulgaria
Chile
China
Colombia
Cyprus
Czech Republic
Estonia
Denmark
Finland
Greece
Hong Kong
Hungary
Iceland
Indonesia
Ireland
Israel
India
Latvia
Lithuania
Malaysia
New Zealand
Norway
Philippines
Poland
Portugal
Romania
Russia
Singapore
South Africa
South Korea
Sweden
Switzerland
Taiwan
Thailand
Turkey
Ukraine
United Arab Emirates
in cracking the 700,000 viewership mark with 100,000 views in one month’s time, so a holiday brain teaser for TV technical gurus (Charles?), or those who aspire to be, question is -......
….
and do some mountain biking in the Laguna Mountains so in case I don’t get a chance to check in again, everyone have a Happy New Years. 
can ride darn well too 
This is an interesting post at AVS
you were holding a grudge (Alexis was offside and I truly believe distracted the keeper) and thusly you were not feeling Glad all Over - 
Linear Mode
