[opendtv] Re: Google: Upload High Dynamic Range (HDR) videos (also streaming with Chromecast)

From: Craig Birkmaier <brewmastercraig@xxxxxxxxxx>
To: opendtv@xxxxxxxxxxxxx
Date: Sun, 20 Nov 2016 09:33:46 -0500

Since Bert won't let go...

On Nov 18, 2016, at 9:50 PM, Manfredi, Albert E
<albert.e.manfredi@xxxxxxxxxx> wrote:

Broadcasters participated, but this is irrelevant. The fact remains, you are
obsessing over nonsense. Broadcasters cannot change the standard at the drop
of a hat, simply because they know that the receivers out there will go dark.
So you're just wasting words, and time, by insisting on this idea that
broadcasters should change the standard. It takes a big effort, a la ATSC 3.0.

Bull. I even explained how they can do it, providing a legacy ATSC 1.0
multiplex.

Everything broadcasters need to move to h.264 compression, while
retaining ATSC 1.0 modulation and transport, is already included
in most new TVs.

That's too preposterous even for Craig (maybe not).

Maybe because it's true. I'm not saying that sets are now built with the right
"connection paths," but all the required components are there. You need ATSC
1.0 demod and MPEG-2 TS support (both are there thanks to the ATSC tuner
mandate); you need a h.264 decoder (there with smart TV implementations).

And obviously you can build these capabilities into an external device
connected via HDMI.

You still have to define the new container, Craig.

Nope. The standard already exists.

2014
A/72 Part 1:2015, “Video System Characteristics of AVC in the ATSC Digital
Television System”
19 May 2015
This Part describes the video coding constraints on ITU-T Rec. H.264 | ISO/IEC
14496-10 (“AVC”) video compression in the ATSC Digital Television System. The
Transport Stream constraints for AVC are described in Part 2 of this standard.

A/72 Part 2:2014, “AVC Video Transport Subsystem Characteristics”
18 February 2014
This Part describes the transport of ITU-T Rec. H.264 | ISO/IEC 14496-10
(“AVC”) as constrained by ATSC A/72 Part 1 for the ATSC Digital Television
System. The syntax and semantics of this specification conform to ATSC A/53
Part 3, with additional constraints specified in this Standard.

Get real. Just because a TV set *might* also have an IP front end, and an
awful lot out there **do not**, it doesn't mean that the H.264 decoder can be
reached from an ATSC over MPEG-2 TS stream, just like that. Besides, you
obsess over old codecs. We have to get to H.265 or equivalent, for 4K.

This is real Bert. It would be trivially easy to create a spec for integrated
tuners in new TVs and for external dongles or boxes.

It always was!

Not really. The government cheese boxes were large and expensive. It can be
done now with a few chips that already exist or a purpose built chip with
existing cores.

So what's your point? You would still alienate a lot of TV users. And as soon
as a new standard exists, manufacturers build what's needed into new sets.
You're just repeating your old mantra from a couple of decades ago, and it
was just as simplistic back then.

First, about 80% of home would not even notice. For those still using antennas
they could still access the SD multiplex(s) with legacy receivers and add a
dongle/box to access the new h.264 HD streams.

More important, this is EXACTLY what broadcasters will need to do if they move
to ATSC 3.0, unless they do a hard cut-over that would obsolete existing
receivers.

What have CDNs got to do with this Bert.

Everything. They make the IP streams available in multiple formats, for the
different devices that use them. Unicast allows a receiver to negotiate with
the edge server, Craig. One-way broadcast does not. The only way to satisfy
multiple incompatible devices with broadcast is to simulcast.

Wrong. The servers for the service you are accessing provide these streams. If
the service CHOOSES to use CDNs, then these versions are mirrored on the CDN
sites. In either case, the device negotiates the stream type based on its
capabilities and the bandwidth available (or allowed). Remember, some wireless
carriers are limiting the resolution available for video streams.

Meaning that moving the ATSC 3.0 requires a lot of new infrastructure.

Not sure why you felt obliged to write all those words. Was this ever in
dispute?

No. what is "in dispute," or more accurately under discussion, is whether
broadcasters will adopt ATSC 3.0. Some are happy with what they've got - i.e.
ATSC 1.0. Some would be happy to keep ATSC 1.0 modulation and transport, and
simply upgrade the compression component.

You don't need a new TV to watch Netflix.

You need a new front end, though, unless you own a connected TV that allows
Netflix streaming. Again, are you just writing words for not good reason?

I said this for GOOD REASON Bert.

The challenge faced by Netflix is no different than that broadcasters would
face if they upgrade to h.264 as I described. The only devices Netflix could
reach initially were PCs (actually DVD players before they started the
streaming service). Now they can reach every iOS and Android device, smart TVs
and TVs with a range of connected TV devices ( including devices that support
4K streams).

The whole point in making this claim is that Netflix did not need the FCC to
mandate receivers, or for the TV manufacturers to build smart TVs. The reality
is that ANOTHER industry developed the Internet standards and devices to use
them. The TV manufacturers saw the opportunity to build these capabilities into
their TVs. Broadcasters can do the same.

Wrong on all accounts. The camera DOES acquire HDR images using the
multiple exposure/image processing technique. And the new displays DO
deliver enhanced dynamic range under appropriate viewing conditions.

Aaargh! Insistent cluelessness. The camera's three exposures are going to
limited range displays.

In some cases this is true. But in some cases the displays Support HDR and WCG.
When I take a HDR photo on my iPhone 6 it is shared with my other devices
through the cloud - my Mac, iPads, etc. I do not currently own an Apple device
that has an HDR/WCG display, a situation that will change when I buy a new iPad
Pro on Black Friday.

But you just can't get it through your thick head that Apple is shipping
multiple products that support HDR and WCG: the 5K iMac; the iPhone 7's; the
new MacBook Pros, and the 9.7" iPad Pro I'm buying Friday. And external HDR/WCG
displays have been available from Apple and others for several years; these
displays can be connected to the Mac Pro, iMacs and MacBook Pros.

And then there is the minor problem you are having with the HDR camera exposure
technique. You are wrong about not seeing a difference on devices with limited
displays. It gets back to the benefits of oversampling.

What happens when you process the raw samples from a modern 4K video camera?
The camera is capturing at least 10 bit samples, perhaps 12 bit or more. Those
samples are optimized for a specific mapping to luminance and color gamut.
Ultimately the samples will conform to one of the HDR/WCG mappings currently
contending to be the new standard (Dolby, Phillips, etc).

As you are well aware these standards also support mappings to displays with
restricted luminance and color gamut. If you compare a current generation HD
camera (1920 x 1080 with 8 bit samples and the output of the 4K HDR/WCG camera
down converted to 1920 x 1080 8 bit samples, the down converted image will look
significantly better. In particular the luminance will have more detail because
the original samples are processed to take advantage of the improved dynamic
range.

The same thing happens with the triple exposure technique. The images are
processed to create an optimized 8 bit image with more information in the
darker and brighter areas of the image. On an 8 bit display you see a better
image than is possible with a single exposure; on a HDR/WCG display you will
see even more detail.

When I get my new iPAD I will be able to do a side by side comparison. You can
be sure I'll report the results.

That's why they have to fill the display with a combination of the three
images, Craig. Read the explanations, confound it.

The dark parts of the scene, when sent to the display, are filled with the
**overexposed** camera content. So the dark parts of the scene don't look
black on the display.

The bright parts of the scene, when sent to the display, are filled with the
**underexposed** camera content. This is so the bright parts don't look
washed out, on the display.

**This is called "compressing the dynamic range."** The dynamic range of a
scene is reduced, to display a more pleasing, even if distorted, image. Read
the literature. Do not insist on things you don't get.

You were doing great until the last paragraph. The processing algorithms take
three 8 bit images and create an optimized 10 or 12 bit image with more detail
in the whites and blacks. An optimized 8 bit image is then created for 8 bit
displays.

The resulting image is NOT distorted. It is MORE ACCURATE. Yes, the dynamic
range is compressed, but the samples are more accurate than those acquired with
a single 8 bit exposure. What's more, the information from all three images is
available for further processing so you can use image processing software to
optimize the image if you don't like the automated mapping.

If a 4K display sells for no more than an HD display, then people will start
buying 4K. It's that trivially simple. Same happened with HD. It became
cheaper than much of the old analog fuzzy stuff, so obviously there was no
point getting stuck on dogma.

But that's not happening, nor is it reality. The price differences are getting
smaller, but HD sets keep getting cheaper too. It's one thing to build an LCD
display with 4,000 8 bit pixels and a basic LED backlight configuration.

It's another story if you increase the samples to 10 bits, significantly
increase display brightness, and implement regional backlight damping to
improve detail in dark regions of the image.

And OLED has its own issues. It is difficult to achieve high levels of display
brightness (this may improve as the OLED technology matures). You get great
blacks, but to see the detail in the whites you need to limit ambient lighting
- i.e. a home theater room or blackout curtains...

So the reality is:
- we can buy cheap 4K displays that do not support HDR and WCG;
- under normal viewing conditions we cannot see the extra detail;
- only a limited amount of 4K content is available, and it requires 25/3
broadband to stream it.

Same with your "high res audio." My first response to you was that eventually
it doesn't help much, so why bother. But you didn't understand that.

Sorry, but YOU did not understand it. You turned the discussion to audio
transducers and the fact that people are willing to pay more for better
transducers for standard quality digital audio. You took forever to acknowledge
that hardly anyone is buying high resolution audio gear.

The point remains that if high res audio sold for the same price as CDs, then
of course everyone would start buying it, EVEN IF the difference was
inaudible.

That's your opinion. Not to mention that 4K without HDR and WCG is mostly
meaningless. On the other hand, adding HDR and WCG to HD will make a huge
difference.

When the opportunity for a new audio player comes up, you just buy whatever
is best and still affordable. Sam with 4K sets, Craig. People don't care
about your dogma, get it? They buy a 4K set that is easily affordable, they
see a smoother image even if way up close, and they're happy. All depends on
cost and perhaps size. And neither are problems anymore.

Your opinion. Perhaps you should wait and see what really happens.

That said, I believe that 4K with HDR and WCG will eventually become
commonplace, but TV entertainment is not going to be the driving force. The
driving force will be the non-Nyquist limited Internet and improved image
quality for photography and EVERY other imaging application.

Can we drop this thread now?

Regards
Craig

References:
- [opendtv] Re: Google: Upload High Dynamic Range (HDR) videos (also streaming with Chromecast)
  - From: Manfredi, Albert E

[opendtv] Re: Google: Upload High Dynamic Range (HDR) videos (also streaming with Chromecast)

Other related posts: