4k by 2k or Quad HD...lots of rumors? thoughts? - Page 123 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #3661 of 3692 Old 04-15-2014, 09:38 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,556
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 130 Post(s)
Liked: 56
Quote:
Originally Posted by irkuck View Post

It is not true block upscaling is distortion-free, that is against theory of digital signal processing. Practically it means jaggies at edges as one of the main distortions.
But it's not changing the original 1080p image. ie. it should be almost the same as a 1080p TV of the same size showing the same image, except that the colours should look more consistent on the UHDTV (2160p) TV as it will have 4 times as many pixels which have the correct colour, whereas the 1080p TV (of the same size) would have only 1 pixel for every 4 of the 2160p TV, and for the 1080p TV, with normally 3 subpixels it wouldn't blend the 3 subpixels as well as the UHDTV one (due to the size).

So although there will be 'better' upscaling methods (that should usually give a more accurate version of what should be there - and which could smooth out the jaggies, normally at the expense of blurring), the block upscaling should give an equivalent (though more consistent colours because of more pixels & subpixels) to a 1080p TV of the same size.

If they really wanted to give a more exact replica of a 1080p TV on 2160p TV they could try and simulate the bigger pixels and especially subpixels of the 1080p TV (of the same TV size) with the higher number of pixels in the 2160p TV. smile.gif Though with only twice as many in both directions, it probably wouldn't be a good representation - you'd need more to do a more accurate version - and to be able to match the gaps between subpixels as well as simulate the subpixels themselves.

Though another thing is - do the 1080p TV make the TV to make the subpixels blend well together in a way that doing a simple block upscale wouldn't achieve. So you'd need to use a slightly different algorithm.
--
But the main point was that the someone was claiming that 2160p TVs gave an inferior picture to a 1080p TV (eg. of the same TV size) when displaying 1080p content. So by having an option that replicated (as much as possible) the look of the 1080p TV displaying the 1080p content on the using the available pixels of the 2160p TV of the same size (where block upscaling may be one way - although not exactly), you could have it appear to look (as much as possible) just the same. You could also have better upscaling methods, which reduced jaggies but that also introduce blur/softening. The user could choose which method depending on preference.
Joe Bloggs is offline  
Sponsored Links
Advertisement
 
post #3662 of 3692 Old 04-15-2014, 10:53 AM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by irkuck View Post
 
Quote:
Originally Posted by fafrd View Post

That's nonsense - you can always use a distortion-free 2x2 'block mode' to get 1080p scaled up to 4K with no change in effective pixel output and less SDE and 720p also upscales to 4K in a distortion-free 3x3 block mode which is not possible when it upscales to 1080p.

It is not true block upscaling is distortion-free, that is against theory of digital signal processing. Practically it means jaggies at edges as one of the main distortions.

 

And that is not true.  Those jaggies were there at 1080 on a 1080 display.  Those very same jaggies are there at 1080 2x2 pixel replicated on a 2160 display.  No distortion was added.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3663 of 3692 Old 04-15-2014, 11:13 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,556
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 130 Post(s)
Liked: 56
Quote:
Originally Posted by tgm1024 View Post

And that is not true.  Those jaggies were there at 1080 on a 1080 display.  Those very same jaggies are there at 1080 2x2 pixel replicated on a 2160 display.  No distortion was added.
Except for the higher number subpixels, and won't there be a diffuser in front of the 1080p display that tries to blur the sub-pixels together a bit (and maybe even pixels a bit)? Wouldn't that also cause a slight difference between the same content on the 2 types of TVs (since, if so, there'd be less diffusion over the pixels on the 2160p TV of the same size, since the pixels are smaller)?

Joe Bloggs is offline  
post #3664 of 3692 Old 04-15-2014, 11:15 AM
Member
 
Desk.'s Avatar
 
Join Date: Jan 2006
Posts: 158
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 49 Post(s)
Liked: 57
Here's a native 1080p image, then upscaled to 4K.

I can see how some people might consider the first image to be 'sharper', certainly...





As I understand it, the only options in upscaling a 1080p image, below left, to 4K is with blurring, middle, or haloing, right. Neither is ideal, and I'm interested to know whether I'd be better with a native 1080p display if that'll be the bulk of my viewing...



Desk
Desk. is offline  
post #3665 of 3692 Old 04-15-2014, 12:15 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 227
What many people don't seem to realize is that each pixel in a 1080p video for example represents a sample - it is not a rectangle.
A sample is a point which is (theoretically) infinitesimally small.

Nearest neighbor interpolation will give you a perfect representation of the original 1080p video as shown on a 1080p display.
Proper upscaling algorithms will give you an image which is far closer to the original source.

Upscaling is not simply blurring the image.


However, this typically does not apply to computer generated graphics, which do assume that a pixel is going to be a rectangle of a certain size and shape, so using video upscaling techniques often give poor results.
Chronoptimist is offline  
post #3666 of 3692 Old 04-15-2014, 12:28 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 592
Quote:
Originally Posted by Desk. View Post

Here's a native 1080p image, then upscaled to 4K.

I can see how some people might consider the first image to be 'sharper', certainly...





As I understand it, the only options in upscaling a 1080p image, below left, to 4K is with blurring, middle, or haloing, right. Neither is ideal, and I'm interested to know whether I'd be better with a native 1080p display if that'll be the bulk of my viewing...



Desk

From where have you understood that?

I don't believe that either of those approaches is the same as the approach used in the example in post #3663 by Joe Bloggs.

Run your 1080p image on the left through that algorithm and it is going to look identical to the 1080p image (or slightly better because of the reduced SDE).
fafrd is online now  
post #3667 of 3692 Old 04-15-2014, 01:49 PM
AVS Special Member
 
dsinger's Avatar
 
Join Date: Apr 2003
Posts: 1,727
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 58 Post(s)
Liked: 90
Quote:
Originally Posted by Chronoptimist View Post

What many people don't seem to realize is that each pixel in a 1080p video for example represents a sample - it is not a rectangle.
A sample is a point which is (theoretically) infinitesimally small.

Nearest neighbor interpolation will give you a perfect representation of the original 1080p video as shown on a 1080p display.
Proper upscaling algorithms will give you an image which is far closer to the original source.

Upscaling is not simply blurring the image.


However, this typically does not apply to computer generated graphics, which do assume that a pixel is going to be a rectangle of a certain size and shape, so using video upscaling techniques often give poor results.

Question about the computer generated graphics. Will this same problem occur for internet sources such as photos in a news story? Reason for asking is I plan to buy a cheap 40" or so 4k set to use as a PC monitor and almost all images will be similar to what I am looking at here on AVS. Thanks
dsinger is offline  
post #3668 of 3692 Old 04-15-2014, 02:18 PM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by Chronoptimist View Post

What many people don't seem to realize is that each pixel in a 1080p video for example represents a sample - it is not a rectangle.
A sample is a point which is (theoretically) infinitesimally small.

 

I wouldn't call a sample a "point" because a "point" itself has geometric meaning.  It's like saying "5" is a point.  It is not.  It is a value.  In the case of a value that represents an amount or type of something, it is a measurement.

 

The question about what the shape of a sample "is" has to do with how the sample is generated and how it is displayed.  If generated by bilinear interpolation it is a parallelogram.  As for the destination side, it is how we see it.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3669 of 3692 Old 04-15-2014, 04:33 PM
AVS Special Member
 
8mile13's Avatar
 
Join Date: Nov 2009
Posts: 3,922
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 257 Post(s)
Liked: 218
Quote:
Originally Posted by fafrd View Post

Quote:
Originally Posted by Desk. View Post

Here's a native 1080p image, then upscaled to 4K.

I can see how some people might consider the first image to be 'sharper', certainly...





As I understand it, the only options in upscaling a 1080p image, below left, to 4K is with blurring, middle, or haloing, right. Neither is ideal, and I'm interested to know whether I'd be better with a native 1080p display if that'll be the bulk of my viewing...



Desk

From where have you understood that?

I don't believe that either of those approaches is the same as the approach used in the example in post #3663 by Joe Bloggs.

Run your 1080p image on the left through that algorithm and it is going to look identical to the 1080p image (or slightly better because of the reduced SDE).

Picks are from Red.com. upscaled 1080p vs native 4k


''All interpolation methods incur some combination of (left) blocking, (middle) blurring and (right) halos artifacts.''

I've been reading all those wonderfull eyewitness stories about 1080p Sony upscaling. Red tells a slightly different story.
8mile13 is offline  
post #3670 of 3692 Old 04-15-2014, 10:21 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 227
Quote:
Originally Posted by 8mile13 View Post

Picks are from Red.com. upscaled 1080p vs native 4k
Keep in mind that Red are in the business of selling 4K cameras.

No-one would deny that native 4K content would be better than upscaled content. But it does not mean that upscaled content necessarily looks bad, and certainly not worse than on a 1080p display.
Quote:
Originally Posted by Desk. View Post

Here's a native 1080p image, then upscaled to 4K.
I can see how some people might consider the first image to be 'sharper', certainly...
http://www.avsforum.com/content/type/61/id/413737/width/500/height/700
http://www.avsforum.com/content/type/61/id/413740/width/500/height/700
As I understand it, the only options in upscaling a 1080p image, below left, to 4K is with blurring, middle, or haloing, right. Neither is ideal, and I'm interested to know whether I'd be better with a native 1080p display if that'll be the bulk of my viewing...
The first image is a representation of how it will look on a 1080p display, if it's assumed that is taken from a 1080p source.

Without the original source material, it's difficult to do a comparison - I had to take your original image and downscale it using nearest neighbor scaling for a close approximation.

Here's a better comparison:
Source at 2x (as if 1080p content were displayed 1:1 on a 1080p display)
Bad upscaling (bilinear scaling)
"Standard" upscaling (bicubic scaling)
Good upscaling (NNEDI3 scaling)

You don't gain any detail from upscaling, but you you do get nice smooth edges and retain most of the sharpness.
The sharpness of the "2x" image is not actually real sharpness though - it's artificial sharpness caused by aliasing. The upscaled image is closer to the original source material.

Which would you rather watch?
Chronoptimist is offline  
post #3671 of 3692 Old 05-16-2014, 07:47 AM
AVS Special Member
 
8mile13's Avatar
 
Join Date: Nov 2009
Posts: 3,922
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 257 Post(s)
Liked: 218
Ultra-HD TV shipments top 1m

UHD TV panel shipments amounted to 1.1 million units in March, nearly a threefold increase from 384,300 units a month earlier in February. Contributing to the market's exceptional showing in March were UHD prestocking orders for China's Labor Day holiday at the beginning of May, and the FIFA World Cup in June.
_______________________________________________________________________________________________________________________________


Even though South Korean makers are late to the UHD TV panel market, they are now beginning to launch lower-cost UHD TV displays, including the socalled ''Green'' panels from Samsung and ''G+'' panels LG. The Green panel, for instance, allows the reduction of UHD panel manufacturing costs by using fewer display driver integrated circuits and a different pixel pattern with only three-quarters of true 3180 x 2160 UHD resolution..
8mile13 is offline  
post #3672 of 3692 Old 05-16-2014, 08:01 AM
AVS Special Member
 
KidHorn's Avatar
 
Join Date: Nov 2003
Location: Derwood, Maryland
Posts: 2,890
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 194 Post(s)
Liked: 206
Quote:
Originally Posted by 8mile13 View Post

Ultra-HD TV shipments top 1m

UHD TV panel shipments amounted to 1.1 million units in March, nearly a threefold increase from 384,300 units a month earlier in February. Contributing to the market's exceptional showing in March were UHD prestocking orders for China's Labor Day holiday at the beginning of May, and the FIFA World Cup in June.
_______________________________________________________________________________________________________________________________


Even though South Korean makers are late to the UHD TV panel market, they are now beginning to launch lower-cost UHD TV displays, including the socalled ''Green'' panels from Samsung and ''G+'' panels LG. The Green panel, for instance, allows the reduction of UHD panel manufacturing costs by using fewer display driver integrated circuits and a different pixel pattern with only three-quarters of true 3180 x 2160 UHD resolution..

Good news. 15 million units in 2014. In a few years, regular HDTVs for sale will be hard to find.
KidHorn is offline  
post #3673 of 3692 Old 05-16-2014, 10:56 AM
AVS Special Member
 
Joe Bloggs's Avatar
 
Join Date: Apr 2007
Location: UK
Posts: 2,556
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 130 Post(s)
Liked: 56
Quote:
Originally Posted by 8mile13 View Post

Even though South Korean makers are late to the UHD TV panel market, they are now beginning to launch lower-cost UHD TV displays, including the socalled ''Green'' panels from Samsung and ''G+'' panels LG. The Green panel, for instance, allows the reduction of UHD panel manufacturing costs by using fewer display driver integrated circuits and a different pixel pattern with only three-quarters of true 3180 x 2160 UHD resolution..
Fake UHDTV doesn't sound good. I hope they make it clear to everyone, eg. in advertising, that it isn't full UHD resolution.
Joe Bloggs is offline  
post #3674 of 3692 Old 05-16-2014, 11:07 AM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by 8mile13 View Post

UHD TV panel shipments amounted to 1.1 million units in March, nearly a threefold increase from 384,300 units a month earlier in February.

 

Seems another TV tech is progressing faster than we thought possible a few years ago.

 

I rather like this trend.

 

So long as it doesn't include "curved".


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3675 of 3692 Old 05-16-2014, 11:21 AM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 592
Quote:
Originally Posted by Joe Bloggs View Post

Quote:
Originally Posted by 8mile13 View Post

Even though South Korean makers are late to the UHD TV panel market, they are now beginning to launch lower-cost UHD TV displays, including the socalled ''Green'' panels from Samsung and ''G+'' panels LG. The Green panel, for instance, allows the reduction of UHD panel manufacturing costs by using fewer display driver integrated circuits and a different pixel pattern with only three-quarters of true 3180 x 2160 UHD resolution..
Fake UHDTV doesn't sound good. I hope they make it clear to everyone, eg. in advertising, that it isn't full UHD resolution.

Sounds a lot like Sharps Q+ 1080p+/4K- stuff (but even worse if it is truly only 6M true pixels).

'True' 4K has 24M subpixels and 8M true pixels.

Sharp Q+ 1080p+/4K- has 16M subpixels and 8M true luma pixels, 8M true green pixels, but only 4M true blue and 4M true red pixels

Don't what this Samsung Green and LG G+ stuff is, but it sounds like it has 18M subpixels (75%) organized into 6M pixels in some way...
fafrd is online now  
post #3676 of 3692 Old 05-16-2014, 11:23 AM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 592
Quote:
Originally Posted by KidHorn View Post

Quote:
Originally Posted by 8mile13 View Post

Ultra-HD TV shipments top 1m

UHD TV panel shipments amounted to 1.1 million units in March, nearly a threefold increase from 384,300 units a month earlier in February. Contributing to the market's exceptional showing in March were UHD prestocking orders for China's Labor Day holiday at the beginning of May, and the FIFA World Cup in June.
_______________________________________________________________________________________________________________________________


Even though South Korean makers are late to the UHD TV panel market, they are now beginning to launch lower-cost UHD TV displays, including the socalled ''Green'' panels from Samsung and ''G+'' panels LG. The Green panel, for instance, allows the reduction of UHD panel manufacturing costs by using fewer display driver integrated circuits and a different pixel pattern with only three-quarters of true 3180 x 2160 UHD resolution..

Good news. 15 million units in 2014. In a few years, regular HDTVs for sale will be hard to find.

I know that there are a great number of die-hard plasma fans here on the board sorry to see Panasonic pull the plug, but it looks increasingly clear that they correctly read the writing on the wall and it was the right call...
fafrd is online now  
post #3677 of 3692 Old 05-16-2014, 01:26 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 227
Quote:
Originally Posted by fafrd View Post

Sounds a lot like Sharps Q+ 1080p+/4K- stuff (but even worse if it is truly only 6M true pixels).
'True' 4K has 24M subpixels and 8M true pixels.
Sharp Q+ 1080p+/4K- has 16M subpixels and 8M true luma pixels, 8M true green pixels, but only 4M true blue and 4M true red pixels
Don't what this Samsung Green and LG G+ stuff is, but it sounds like it has 18M subpixels (75%) organized into 6M pixels in some way...
Sharp's Q+ panels are effectively 1920x2160 native displays, with four subpixels per pixel.
Subpixel rendering is used to increase the horizontal resolution beyond 1920 pixels wide.
Because the subpixel structure is RGBY, depending on what is being displayed, this can be addressed as RGB and BYR groups.
So the horizontal resolution is fluid, and can be anywhere between 1920px or 3840px depending on what is being displayed.
It's not true 4K, but it's not as bad as calling it "fake 4K" suggests either.

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.
Chronoptimist is offline  
post #3678 of 3692 Old 05-16-2014, 01:33 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 592
Quote:
Originally Posted by Chronoptimist View Post

Quote:
Originally Posted by fafrd View Post

Sounds a lot like Sharps Q+ 1080p+/4K- stuff (but even worse if it is truly only 6M true pixels).
'True' 4K has 24M subpixels and 8M true pixels.
Sharp Q+ 1080p+/4K- has 16M subpixels and 8M true luma pixels, 8M true green pixels, but only 4M true blue and 4M true red pixels
Don't what this Samsung Green and LG G+ stuff is, but it sounds like it has 18M subpixels (75%) organized into 6M pixels in some way...
Sharp's Q+ panels are effectively 1920x2160 native displays, with four subpixels per pixel.
Subpixel rendering is used to increase the horizontal resolution beyond 1920 pixels wide.
Because the subpixel structure is RGBY, depending on what is being displayed, this can be addressed as RGB and BYR groups.
So the horizontal resolution is fluid, and can be anywhere between 1920px or 3840px depending on what is being displayed.
It's not true 4K, but it's not as bad as calling it "fake 4K" suggests either.

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.

Totally agree. It's basically identical to the real deal for 4K at 4:2:2 and only loses some resolution when displaying 4K at true 4:4:4.
Quote:
Originally Posted by Chronoptimist View Post

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.

What is the pentile structure and how would it result in 2/3 the effective pixel count?

In terms of display of 4K at 4:2:2, how would this compare to the Sharp Q+ stuff?

And same question in terms of display of 4K at true 4:4:4.
fafrd is online now  
post #3679 of 3692 Old 05-16-2014, 01:51 PM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by fafrd View Post
 
Quote:
Originally Posted by Chronoptimist View Post
 
Quote:
Originally Posted by fafrd View Post

Sounds a lot like Sharps Q+ 1080p+/4K- stuff (but even worse if it is truly only 6M true pixels).
'True' 4K has 24M subpixels and 8M true pixels.
Sharp Q+ 1080p+/4K- has 16M subpixels and 8M true luma pixels, 8M true green pixels, but only 4M true blue and 4M true red pixels
Don't what this Samsung Green and LG G+ stuff is, but it sounds like it has 18M subpixels (75%) organized into 6M pixels in some way...
Sharp's Q+ panels are effectively 1920x2160 native displays, with four subpixels per pixel.
Subpixel rendering is used to increase the horizontal resolution beyond 1920 pixels wide.
Because the subpixel structure is RGBY, depending on what is being displayed, this can be addressed as RGB and BYR groups.
So the horizontal resolution is fluid, and can be anywhere between 1920px or 3840px depending on what is being displayed.
It's not true 4K, but it's not as bad as calling it "fake 4K" suggests either.

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.

Totally agree. It's basically identical to the real deal for 4K at 4:2:2 and only loses some resolution when displaying 4K at true 4:4:4.
Quote:
Originally Posted by Chronoptimist View Post

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.

What is the pentile structure and how would it result in 2/3 the effective pixel count?

 

Check out this link: http://www.clivemaxfield.com/diycalculator/popup-h-rgbwnum.shtml for a great visual explanation for pentile.

 

Chron is almost certainly referring to the fact that a pixel is theoretically only 2 sub pixels in size.  It's an inherent kind of sub pixel rendering (in a way).  Consider the diagrams in the link above.  Note: this is for an WRGB pentile arrangement.  RGB pentile arrangements are still a bit of a mystery to me.

 


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3680 of 3692 Old 05-16-2014, 03:41 PM
AVS Special Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 9,955
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 526 Post(s)
Liked: 549
Quote:
Originally Posted by 8mile13 View Post

''All interpolation methods incur some combination of (left) blocking, (middle) blurring and (right) halos artifacts.''
Only if you don't do it right. You can avoid the halo artifacts if you use an anti-ringing filter.

For example, madVR has very good upscaling algorithms that don't do or suffer from what Red is alleging.

Examples:
http://www.avsforum.com/t/1442663/which-hd-would-be-sufficient-for-madvr/100_50#post_23644293
http://madshi.net/madVR/monsters.png
Stereodude is offline  
post #3681 of 3692 Old 05-18-2014, 10:00 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 3,626
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 592
Quote:
Originally Posted by tgm1024 View Post

Quote:
Originally Posted by fafrd View Post

 
Quote:
Originally Posted by Chronoptimist View Post

 
Quote:
Originally Posted by fafrd View Post

Sounds a lot like Sharps Q+ 1080p+/4K- stuff (but even worse if it is truly only 6M true pixels).

'True' 4K has 24M subpixels and 8M true pixels.

Sharp Q+ 1080p+/4K- has 16M subpixels and 8M true luma pixels, 8M true green pixels, but only 4M true blue and 4M true red pixels

Don't what this Samsung Green and LG G+ stuff is, but it sounds like it has 18M subpixels (75%) organized into 6M pixels in some way...
Sharp's Q+ panels are effectively 1920x2160 native displays, with four subpixels per pixel.

Subpixel rendering is used to increase the horizontal resolution beyond 1920 pixels wide.

Because the subpixel structure is RGBY, depending on what is being displayed, this can be addressed as RGB and BYR groups.

So the horizontal resolution is fluid, and can be anywhere between 1920px or 3840px depending on what is being displayed.
It's not true 4K, but it's not as bad as calling it "fake 4K" suggests either.


I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.


Totally agree. It's basically identical to the real deal for 4K at 4:2:2 and only loses some resolution when displaying 4K at true 4:4:4.
Quote:
Originally Posted by Chronoptimist View Post

I suspect that these "G+" panels are going to be using a pentile-like structure, and the report of having 3/4 the pixels of native 4K is a mistake. They probably meant 2/3 the resolution rather than 3/4.


What is the pentile structure and how would it result in 2/3 the effective pixel count?

Check out this link: http://www.clivemaxfield.com/diycalculator/popup-h-rgbwnum.shtml for a great visual explanation for pentile.

Chron is almost certainly referring to the fact that a pixel is theoretically only 2 sub pixels in size.  It's an inherent kind of sub pixel rendering (in a way).  Consider the diagrams in the link above.  Note: this is for an WRGB pentile arrangement.  RGB pentile arrangements are still a bit of a mystery to me.




Interesting - thanks. This seems like something that would be especially well-suited to LGs WOLED (with its white OLED sheet) - is Samsung the only one pursuing this pentile pixel arrangement? Does Samsung have a lock on the patents or something?

This seems lie it would be able to display 4K @ 4:2:2 very effectively, with 50% increase in light output for the same power consumption. Are there downsides I am missing (more complex/expensive to manufacture)?

Grouping the green and white subpixels together (and the red and blue) seems like it would be a better configuration of the most accurate 4:2:2 possible (4 for luma, 4 for green, 2 for red and 2 for blue).
fafrd is online now  
post #3682 of 3692 Old 05-19-2014, 06:15 AM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761

^^^As far as I know, Pentile is an actual trademark of Samsung.  But also, I believe (<---note), the concept predates that.

 

It usually completely rubs the purists (myself included) the wrong way.  At first I looked at it entirely as a kind of cheating or a kind of manufacturing  stop-gap measure.

 

But I slowly warmed up to it when an optics/imaging scientist of mine gave a very complicated explanation of how it defeats a natural aliasing that occurs when trying to match the harshness of a grid to the images attempted to be gathered.  The conversation with him went on by email for pages.  Oye...no way I'm bringing that mess into a forum.

 

And I can buy that position, but only if the sensors gathering that information are themselves pentile, which apparently isn't a requirement to him.  In any case, I can readily see how it defeats the aliasing effect of any straight line.  So I'm still warming up to it.

 

BTW, the OLED with its subpixel arrangements on my Galaxy Note II, no matter how they work, produce absolutely wonderful images.  But at first glance they are absurd:

 


Note to others: Despite continual information to the contrary, notice that the IPS listed here is not showing a chevron shaped subpixel.  I think we can put that little assumption to rest now.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3683 of 3692 Old 05-19-2014, 10:51 AM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by fafrd View Post

Grouping the green and white subpixels together (and the red and blue) seems like it would be a better configuration of the most accurate 4:2:2 possible (4 for luma, 4 for green, 2 for red and 2 for blue).

 

This last part threw me.

 

You might realize this, but the chroma subsampling ratios are actually measurements of Y, Cb, and Cr.  The the Cb and Cr aren't specifically blue and red, they're components of a digitized version of the YUV color model.

 

http://www.pcmag.com/encyclopedia/term/57460/chroma-subsampling.  Note: I can't verify this for sure, but I suspect these guys may have lifted the information off of the wikipedia page for chroma subsampling which is IMO a horrendously bad practice, so consider that if you find any inaccuracies.  I lost my clearer explanation.

 

Y is the luma information normalized at 4, with two additional axes Cb and Cr supplying the entirety of the color information devoid of their luma components.  Color folks will jump down my throat on this one, but I prefer to think of it as just two axes into a color plane with a luma axis extending vertically off of it...it's how I visualize it when I'm just not up for the CIE chromaticity diagram.  LOL.

 

I'm curious: you might be thinking of a very weird bit depth arrangements such as 2/4/2, which was a hokey single byte (8 bits) representation of RGB that I've only actually seen used once, and was, well, gross.  2 bits for red, 4 for green, and 2 for blue for every pixel.  It's a notion of supplying more bits for the colors we were the most sensitive to, but for something that small it had a very limited usage.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3684 of 3692 Old 05-19-2014, 02:02 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 227
Quote:
Originally Posted by tgm1024 View Post

^^^As far as I know, Pentile is an actual trademark of Samsung. But also, I believe (<---note), the concept predates that.
It usually completely rubs the purists (myself included) the wrong way. At first I looked at it entirely as a kind of cheating or a kind of manufacturing stop-gap measure.
But I slowly warmed up to it when an optics/imaging scientist of mine gave a very complicated explanation of how it defeats a natural aliasing that occurs when trying to match the harshness of a grid to the images attempted to be gathered. The conversation with him went on by email for pages. Oye...no way I'm bringing that mess into a forum.
And I can buy that position, but only if the sensors gathering that information are themselves pentile, which apparently isn't a requirement to him. In any case, I can readily see how it defeats the aliasing effect of any straight line. So I'm still warming up to it.
BTW, the OLED with its subpixel arrangements on my Galaxy Note II, no matter how they work, produce absolutely wonderful images. But at first glance they are absurd: http://www.avsforum.com/content/type/61/id/434174/width/1000/height/1000/flags/LL
Note to others: Despite continual information to the contrary, notice that the IPS listed here is not showing a chevron shaped subpixel. I think we can put that little assumption to rest now.

The Galaxy S5 with its new "diamond pixel" layout now seems to have enough resolution/pixel density that it compares very favorably to the traditional RGB stripe layout:
03-100-crop63ol7.jpg
tgm1024 likes this.
Chronoptimist is offline  
post #3685 of 3692 Old 05-19-2014, 02:29 PM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761

Indeed it does, thanks for that pic!!!!  WOW....look at how uniform the effect is on the inner edges of the "s"----no matter the angle, it looks like there's no increase or decrease in artifact-effects!!!!  That looks surprisingly good!

 

They are definitely onto something!


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3686 of 3692 Old 05-19-2014, 02:49 PM
 
vinnie97's Avatar
 
Join Date: Jul 2006
Location: Nunya
Posts: 11,657
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 213 Post(s)
Liked: 1010
Never thought I'd describe a pixel layout as beautiful, but that certainly fits the bill. Too bad the unlocked version of the S5 is $600 or more (and they're only incorporating the technology into a 5" screen). eek.gif
vinnie97 is offline  
post #3687 of 3692 Old 05-19-2014, 03:00 PM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:Quote:
Originally Posted by vinnie97 View Post

Never thought I'd describe a pixel layout as beautiful, but that certainly fits the bill
 
Yeah, perhaps not as long as some of the monsters lurking around here, but I've been in the graphics/imaging world for a long time and I don't think I've seen something pull that off quite so well.  Round of applause to Chron for putting it on our radar.
 
Quote:
Too bad the unlocked version of the S5 is $600 or more (and they're only incorporating the technology into a 5" screen). eek.gif

 

My Galaxy Note II had a list price at BB of ~ $800.  It's a glorified S4 (with a stylus that I forget is there).  But that extra .5"......wow what a difference.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3688 of 3692 Old 05-19-2014, 03:03 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 227
I think the key difference is that the diamond layout means that each subpixel is almost the same distance from the next whether you are drawing horizontal, vertical, or diagonal lines - this is a lot better than the traditional RGB stripe layout where it goes from "touching" on vertical lines to being quite far away on diagonals or horizontal lines.
As long as the display is sufficiently high enough resolution, it doesn't seem to matter that there are fewer red/blue subpixels than green.

While I have long complained about displays which do not use the standard RGB stripe subpixel layout, I think I would be happy with a 4K OLED display using this.
Chronoptimist is offline  
post #3689 of 3692 Old 05-19-2014, 03:09 PM
AVS Special Member
 
tgm1024's Avatar
 
Join Date: Dec 2010
Posts: 6,352
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 761
Quote:
Originally Posted by Chronoptimist View Post

I think the key difference is that the diamond layout means that each subpixel is almost the same distance from the next whether you are drawing horizontal, vertical, or diagonal lines - this is a lot better than the traditional RGB stripe layout where it goes from "touching" on vertical lines to being quite far away on diagonals or horizontal lines.
As long as the display is sufficiently high enough resolution, it doesn't seem to matter that there are fewer red/blue subpixels than green.

While I have long complained about displays which do not use the standard RGB stripe subpixel layout, I think I would be happy with a 4K OLED display using this layout.

 

Yeah, I was noticing that the grouping is almost always the same no matter how a "dark shape" cuts into it.  However look, there's an additional bit of processing going on, and it's got me wondering: throughout the inner perimeter of the "s" (the glyph I'm studying intently), the closest subpixel to the black is dimmer.  Almost as an intentional anti-aliasing.


Grow milkweed. The Monarch Butterfly requires it, and its numbers are dwindling fast.
tgm1024 is offline  
post #3690 of 3692 Old 07-08-2014, 04:47 AM
AVS Special Member
 
8mile13's Avatar
 
Join Date: Nov 2009
Posts: 3,922
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 257 Post(s)
Liked: 218
UHD HD TV Phase 1 broadcast specification approved for Europe by the DVB Steering Board yesterday.
http://www.vodprofessional.com/news/...teering-board/
dsinger and Joe Bloggs like this.
8mile13 is offline  
Reply OLED Technology and Flat Panels General

Tags
Samsung Bd D7000 3d Blu Ray Disc Player Silver , Samsung Pn51d8000 51 Inch 1080p 3d Ready Plasma Hdtv 2011 Model , Displays , Pioneer Pdp 5080 Hd
Gear in this thread

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off