[opendtv] Re: Push for Ultra HD 4K forces picture quality tradeoffs - CNET

  • From: donald.koeleman@xxxxxxxxxx
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Mon, 26 May 2014 23:00:51 +0200

Hi Jeroen.

Let's not forget about motion portrayal, sharpness is not just spatial resolution plus contrast, but temporal resolution as well, both reproduction and production, high frame rate and no deliberate blurring please, and HFR isn't a measly 48 or 60 frames per second as the movie folks seem to think.

Remember we were at some English HDR demo at IBC, according to the demonstrator, they had to choose between HDR and proper motion and proper colour they could not do all three, just one. Improper colour and associated contrast made for an unsharp picture, so was the improper motion portrayal.

5000 Nits now we're talking;-). All the BT and YUV CB CR talk made me glaze over, but it sounds good.

How wide is that 'visible spectrum' ITU.R 2020, or even wider?

It has been years since I heard you refer to multiple primary colourspaces/systems. Good to see Philips is again focussing on technology over pure design to warrant its pricepremium, as a Philips SVP? explitely told the Picnic conference, what five or six years ago.

The Sony OLED 56"4K demonstrator at IBC had some spectacular highlights, but there definitly was room for intra-image contrast to match the brightness. The content was all gold from the Brasil Carnival parade, so easily bright, but 'ton sur ton' is a problem for contrast. I didn't get to see the Dolby private demo.

Donald

Quoting Jeroen Stessen <jeroen.stessen@xxxxxxxxxxx>:

Hello all,

On 2014-05-23 13:04, Craig Birkmaier wrote:
Push for Ultra HD 4K forces picture quality tradeoffs - CNET

http://www.cnet.com/news/push-for-ultra-hd-4k-forces-picture-quality-tradeoffs/

*(...) *

We all want better picture quality. As I laid out in my article, Three TV improvements more worthwhile than Ultra HD 4K <http://www.cnet.com/news/three-tv-improvements-more-worthwhile-than-ultra-hd-4k/>, there are other aspects to the picture more important than resolution.


I like that part, because currently I am working on the contrast and color
aspects. Contrast as in High Dynamic Range signals (our aim is 5000 nits),
and color as in Wide Color Gamut (our aim is the entire visible gamut).

We are increasingly becoming convinced that the traditional Y'CbCr 4:2:0
format is limiting to both, so we are introducing a new format that we
call Y"u"v" (4:2:0). The two chromaticity components are from UCS 1976,
meaning that all the dynamic range is only in the Y" signal, and all the
color gamut is only in the u"v" signals. Why didn't we think of this before ?

u"v" quantizes more efficiently than CbCr (10 bits each instead of 12,
and less correlation with Y"), and the quality can come closer to 4:4:4.
Plus, there is no artificial color gamut limit from RGB color primaries.
Color space conversion to any display color gamut is straightforward.

We use double primes to indicate that there are some new non-linear
transfer functions involved, but that is only of secondary importance.
If Y' is quantized according to a traditional gamma function (i.e. BT.1886)
then Y" is quantized according to a new perceptual quantizer: at the
bottom it resembles a gamma function, and at the top a log function.
(Not invented by us, but based on work by the late Dr. Barten who was
one of us.) This allows far higher brightness from the same signal range.
All non-linearities are compensated by their inverse in the receiver.

More brightness means less compression (clipping) of the highlights,
this gives more sparkle (specular highlights) and more details in bright
areas. This does not come out of the dark areas, so overall there will be
more detail visible. That enhances sharpness more than 4K alone does.
Mark Schubin will tell you that sharpness = resolution x contrast.

The kicker is that you need an HDR capable display, and at the moment
those are very power hungry. But the visible effect is wowwwww.

When we are ready to publish more details I'll let you know.

Best regards,
-- Jeroen





----------------------------------------------------------------------
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: