Samsung to use Sub-Pixel VGA Screens

pdawerks writes "Samsung Electronics has developed a new graphics chip that will allow half VGA screens to produce VGA resolution. The novelty is specially aimed at future mobiles with VGA screens that will be less than 2.4 inches. It generates color using an entirely new driving method called sub-pixel unit driving methodology." Not sure if I think it is exactly new or not, but it's nifty.

Comments

More Information

[Temporal Outcast]

More details can be found at Deisgntechnica.

Geekzone also has a similar article.

Re:More Information

[Stanhenge]

Refer to the patent US#5193008 for a technique that increases resolution of a raster device.

DP-Tek developed this for laser printer devices, but the idea applies to other technologies. Basically, you can place a physical line between adjacent laser scan lines, using the analog memory of the OPC drum.

Driving color???

[c0p0n]

It generates color using an entirely new driving method called sub-pixel unit driving methodology

I suppose I got my driver license from the wrong place...

Anyone?

[wankledot]

This sounds exactly like sub-pixel antialiasing, which is the basic for lots of things, including OS X's font smoothing on LCDs, and Microsoft's type technology... I forget its name.

Is it really as simple as that? because that's been around for at least 25+ years in theory, a bit less in practice.

Re:Anyone?

[GreyPoopon]

This sounds exactly like sub-pixel antialiasing,

Not exactly. Cleartype and OS X font smoothing use subpixel rendering to increase the horizontal resolution. This technique seems to work on the vertical resolution.

"Contrary to existing color display methods that express color pixel by pixel, this new method creates color at the sub-pixel level representing more than two data lines from the same pixel."

Maybe they accomplish this by rotating the orientation of the pixels so that it impacts the vertical rather than horizontal? Or maybe this is just a big hoax? Anybody have more information?

I'm Confused

[jmulvey]

the title suggests that "VGA" indicates a default screen size (like 4" by 6"), but my understanding is that VGA says nothing about the size of the display, only the number of pixels (you can display VGA resolution of 640 x 480 on a 10" screen or a 30" screen, and its still VGA).

So isn't the whole term "half VGA screen" kinda dumb? Or is it just me?

Re:I'm Confused

[ivan256]

Isn't it about time we depricated the use of those silly acronyms we've bastardized to not mean what they originally meant anymore anyway? Wasn't VGA 640x480 at a mere 256 colors? And didn't it imply a particular ISA bus interface as well? Plus, who can keep track of what WUXGA and QWVGA and UHDWMRXGA all mean? Was somebody just leaning on the keyboard, or did they mean to say something anybody could understand like "1600x1200"? Tell us the resolution in a way that doesn't require a lookup in a massive acronym table please. That way it will be easy to compare displays to each other.

Well it makes sense, and saves battery power...

[arashiakari]

Double the resolution, and blend the colors of neighboring pixels together to fit on a lower res. screen. Sounds like a new way of saying "anti-aliasing" ...

And the window washers are now "corporate vision enhancers!"

Nothing is impossible

[uncl_bob]

The new driver IC has overcome the physically impossible VGA-class and higher resolution images on small size TFT-LCD panels of less than 2.4 inches

Why is it physically impossible to design VGA displays less than 2.4 inches? Too small pixels?

Re:Nothing is impossible

[stonecypher]

Because current LCD pixels require six lead lines, and we can't make lead lines small enough to shrink the pixels any further. The article phrases this badly: it's not that pixels can't be made smaller. It's that TFT LCD pixels' lead lines take all of the available current space, and there is no current technique on the horizon to solve this. Other monitor types do not have this particular problem; this is peculiar to LCD and OLED.

MS Cleartype

[darkmeridian]

The article is really short, but it says that the screen will use sub-pixel technology to allow a half-VGA screen to render VGA resolution. MS Cleartype also uses sub-pixel technology, though to make text sharper.

A linkie with information about sub-pixels in general (though it's on grc.com, whatever.) http://grc.com/cleartype.htm

All new technology? Unlikely

[ikewillis]

Subpixel rendering has been around for quite a long time. Two things that I can think of right off the bat are Microsoft's ClearType and FreeType, both of which have hinting engines which support subpixel rendering.

Subpixel rendering takes into account the physical position of the red, green, and blue subpixels of an LCD display, and can therefore provide up to 3X the horizontal resolution of a typical display (with distortion, of course)

Here's a nice writeup

ClearType?

[theGreater]

Sounds basically like cleartype, right? I mean, all THAT is is using the RGB (or CYM) sub-pixels to smoothe out lines and curves, correct? Err, so what's the BFD?

-theGreater Muller.

Re:Sounds Like Interlacing?

[DeepFried]

Interlacing does not double lines. It is just a process that brings the lines up in an alternating (odd/even) sequence. This is now being joined by progressive scan which brings the lines on in order from top to bottom.

Progressive or interlaced, can each scale in lines of resolution to HiDef. 1080i and 720p respectively. (i=interlaced p=progressive)

Interlacing is used to reduce flicker

[GillBates0]

and does not have anything to do with the resolution. Infact, interlacing is sometimes called "interlace scanning", because the gun in the CRT draws alternate lines across the screen to reduce the visible flicker arising due to the time required to move the gun from top to bottom.

As usual, Wikipedia has a good article. To quote:

Interlacing is a method of displaying images on a raster-scanned display, such as a cathode ray tube (CRT), that results in less visible flickering than non-interlaced methods. The display draws first the even-numbered lines, then the odd numbered lines of each picture.

Re:Interlacing is used to reduce flicker

[shirai]

Interlacing is used to reduce flickering? I think not. It used to be used to reduce *bandwidth*.

An interlaced image refreshing at 60Hz (30 full fields per second divided by 2) is going to have the same flicker as a non-interlaced image refreshing at 60Hz.

This is actually a very complex subject to do with how people view images, resolution vs fields per second, what type of images you are viewing, movement vs. still images, etc. but in terms of reducing flicker, I would say, at the very least, the statement is deceptive.

In fact, one of the major problems with old Amigas running in interlaced mode was the annoying (you got it) flicker. This is because a horizontal line that was exactly 1 pixel would turn on and off every 60th of a second. So in this case, it would depend on how you defined the world flicker too.

To be fair, I think what you meant to say was that given the same bandwidth on a non-digitally compressed transmission and without digitally upconverting the signal, you can get 60 fields per second (at 30 frames per second) instead of 30 fields per second (at 30 frames per second) meaning that you will probably get less inter-frame flicker. But even this is deceptive because if you built televisions specifically for 30 frames per second, you could simply reformulate the glow on the screen to last an extra 1/60th of a second longer. But perhaps this is (a) hard to do and (b) back then they wanted the extra fields per second for smoother motion. By the way, a lot of the bandwidth savings doesn't apply to digital due to the way that digital compression works. This was a controversial point during the discussions on HDTV resolutions.

Fudge. I'm trying to cover all my bases here so I don't get flamed for not knowing what I'm talking about. Suffice it to say, interlacing and reduction of flicker do NOT walk hand in hand. It is simply one factor, of many, that comes into play.

Re:Interlacing is used to reduce flicker

[Anonymous Coward]

>Suffice it to say, interlacing and reduction of flicker do NOT walk hand in hand.

When the TV was invented, it was noticed that a phosphor did not remain lit long enough for the beam to make a complete pass at 29.9fps, therefore there would be significant "flicker" in the picture. The inventor(s) decided to interlace so you'd get a more uniform brighness to the picture and eliminate the flicker. This problem has long since been solved in other ways.

During the VGA days, however, the reasons were entirely different, as you suggest.

Re:Interlacing is used to reduce flicker

[mercuryresearch]

Parent reply post is on the mark here.

Where the confusion comes up is in the old days where interlacing was first used -- black and white television. Interlaced TV signals drove black and white CRTs... and by selecting the right phosphor, the displays had persistence, where the image would continue to glow into the next frame even after being drawn. High persistance phosphor DID cut down on flicker, and was necessary because of the interlacing.

If you want to get into really obscure stuff... Radar display tubes (and some oscilloscopes) often had incredible persistence -- often lasting several to even tens of seconds. Surplus radar tubes were popular among ham radio operators doing slow-scan television, as they could maintain the TV display at the insanely low frame rates used at the time (I think it was about 1 frame per 8 seconds or so, memory is hazy here.)

Anyway, think of the displays long ago as a giant analog visual low-pass filter -- you could throw pretty much any signal at them regarless of flicker and get a flicker-free display, so the bandwidth savings from interlacing was an obvious choice. Not so with digital technology today.

As they say in Germany

[Bender Unit 22]

As they say in Germany "ich habe gemüse in das leiderhosen". Which means that it might be looking like new fancy things but it is still the same old clothes.
Kinda like the Swedish "min trusse lugter af tis",, it's new but then again, it's not.

Is it a case of someone applying existing technologies like smoothing to the hardware layer if you look into what's really going on?

It's not subpixel as with ClearType !

[GrAfFiT]

The article suggests that they added "White pixels". Additionally, the problem of dark screen due to the increased pixel density on high resolution panels has been solved using 4-color (R-G-B-W) rendering algorithm, improving the brightness of TFT-LCD panels. That's radicaly different than ClearType. ClearType uses the normalized RVB subpixels arrangement to triple the "perceived" resolution. That's because the humain eye is more sensitive to luminance than to chrominance (try to recognize colors in the dark, you can't, but you can still read B&W text). The problem here is not text aesthetics. It's global luminosity, as your backlight often has to battle with sunlignt. They add more "white pixels" to enhance the luminosity. In percentage, the number of "color" pixels are lower in this system. But the eye won't actually see the difference.

Re:It's not subpixel as with ClearType !

[shirai]

Note that white pixels aren't a magic bullet. You get some brightness but give up saturation. It works like this:

Given four pixels of RGBW, you can get your brightest color by having all four pixels on. This would result in total brightness of:

1 white pixel for every combination of RGB and

1 white pixel for every white pixel.

So you get the equivalent of 2 white pixels for every 4 pixels or a factor of 1/2 let's say.

In regular RGB, you get a factor of 1/3 because you get the equivalent of 1 white pixel for every set of RGB pixels.

Looking at this, you get 50% more maximum brightness from RGBW vs RGB.

It's not a magic bullet because you lose saturation. For example, if you want a fully saturated red, in the RGBW format, you get 1 full red pixel for every four pixels. In RGB, you get 1 full red pixel for every three pixels. So RGBW gives a factor of 1/4 while RGB gives a factor of 1/3 for a fulls aturated red. This is a reduction in brightness of a full saturation red of 25%.

In other words, your brightest color is 50% higher in RGBW but you brightest red (at full saturation) is 25% less which means you have to fudge around with values to get a picture that seems to make sense or you get a bright picture with dark spots with a lot of saturation in them. So you might, programatically (and this is probably what samsung is doing) increase full saturation red to include white in it. This makes the color brighter but also reduces the saturation.

A lot of projectors with a white component have two modes. A dimmer mode that doesn't use the "W" pixel at all but has richer colors (used for movie viewing) and a presentation mode that does use the "W" when brightness is a factor such as in a meeting (e.g. the room may have light leaking in from windows).

Not saying it is good or bad. Just that a RGBW is not a magic bullet.

READ THE DAMN ARTICLE

[boomgopher]

Quote:
By composing a new pixel with the sub-pixel on the adjacent scanning line, 480*640 VGA resolution can be attained from a 240*640 half-VGA panel.

Drop all the "MacOS does this", "ClearType does this", etc. shit please.

Re:READ THE DAMN ARTICLE

[Greger47]

Err, why? It's the same thing as ClearType, they just rotated the display 90 degrees and are doing subpixles vertically instead of horizontally.

My guess is that someone read that MS patent really carefully and concluded that it only covers horizontal subpixels. :)

The novelty would be that it's implemented in the display driver chip thus I guess it can move any pixel around, not only when rendering fonts.

/greger

Re:READ THE DAMN ARTICLE

[hattig]

As far as I can tell, they are not doing that in any shape or form.

For a start, Cleartype is for text and increases the horizontal resolution of text because the subpixel resolution of a 640x480 screen is actually 1920x480

This is RGBW ... and I am guessing that it is laid out in a

RG
BW

format, i.e., a 640x480 screen would have a subpixel resolution of 1280x960. Cleartype wouldn't work on this screen as it is currently implemented.

What they are doing is taking a 640x240 "Double Height" screen (i.e., 4:3 with tall pixels) and using this to get a subpixel resolution of 1280x480.

So it looks like they are kinda then using a cunning but easy to work out algorithm to spread a 480 pixel high display over a 240 pixel high RG/BW display. I.e., Even Lines contribute 50% to RGBW square, odd lines contribute 50% to BWRG square that is offset a little below.

It certainly isn't perfect. But it sounds easier to fit 1280 subpixels in a small display than 1920 doesn't it?

Re:All new technology? Unlikely

[Monkelectric]

Actually its been around a lot longer then you think. The Apple II used a form of sub pixel rendering written by steve wozniak himself.

Is this really something new?

[gotr00t]

I have seen this methodology used in many applications where the screen was just too small to accomodate a purpose. Take Tezxas, for example, a ZX Spectrum emulator for the TI-89 (and 92+, but that dosn't apply here). Since the ZX spectrum's screen is roughly twice the dimensions of the 89's screen, 4 pixels had to be represented by one. There are also some applications for the PocketPC that use a very similar sounding method to bring full VGA resolution to half-VGA sized screens.

My question is, is this something new because its more clear? or because it's a hardware implementation?

Article missing critical technical information

[francisew]

Here is a link to the Samsung website about the technology: http://www.samsung.com/Products/TFTLCD/Technology/ 4colorrandering.htm

I wouldn't complain too hard about the confusion in the details. They couldn't even spell 'rendering' right on their own site (4 color randering???).

It also discusses 'physicail' pixels. I dunno about that.

They seem to have created smaller pixels, which are spatially located across a different area than normal.

They then need fewer wires to connect the given number of pixels. Meaning a higher resolution with fewer interconnects. Maybe I'm completely wrong in this 1 minutes evaluation.

The neat thing is the overlap of their 'logical' pixel arrangements. It would seem they are using traditional dithering with a complicated arrangement of pixels. This should do exactly what they state. Ther weird thing is that their sub-pixel seems to have the wrong number of color sub-elements.

One would expect a ratio of 2:1:1 for green:red:blue emitters. They have 4:2:1. Maybe their red emitters are much brighter than the blue, which would make sense.

They mention replacing some rows with white pixels, but their diagrams don't show anything. Maybe the media-relations people just don't know how the technology works, and are making stuff up until someone corrects them.

Tiny hints from the article...

[raygundan]

According to the article, they're generating a white signal from the RGB input and have four color elements for each pixel-- RGBW. I suspect they're arranged in a square, like:

RG
BW

or some such. This would let them apply a system like ClearType or OSX or the old Apple II subpixel rendering in two dimensions, rather than just one as with the typical horizontal RGB subpixel arrangement.

No substitute for physical resolution

[Anonymous Coward]

Samsung's press release about "sub-pixel unit driving methodology" is total hype and bull in my opinion. This technique provides better color and smoothing but no higher resolution by any means. They should be honest and call is what it is - color contrast and sharness enhancing technology - and not suggest that it provides a higher resolution for a given and fixed physical resolution.

Re:More Information Call me "AcronyWhore"

[davidsyes]

for Acronym Whore...

They better change the name of the gun/emitter from Sub-Pixel Unit to something else, or it will be "stuck" with the acronym...

S.P.U.G.E.

Maybe Micro-Pixel Unit.

Actually...

[WARM3CH]

... VGA has only 16 colors in 640x480. It could only show 256 colors in the 320x200. Comparing it to what most PDAs do now, it seems that getting 64K colors in 320x200 is already beyound what VGA did!

Re:Actually...

[spinlocked]

Programmed normally, they had 64k of memory mapped into a segment in high-mem. To get some of the more bizarre 'ModeX' modes you had to program the VGA registers directly to change the timing and remap segments of it's larger memory in and out of the normal 0xa000h (or something like that). I think VGA cards had at least 256K to play with. You could do some really cool stuff in ModeX, anyone else remember the smooth scrolling in Bananoids? IIRC MCGA was IBM's cheaper adapter. It only had enough memory to do 320x200x8.

God, I can still remember that stuff, but I can't remember my Mother's birthday.

Maybe

[Andy Dodd]

They seem to be indicating that the RGBW trick is a whole different thing used to increase brightness (similar to CMYK for printers to make dark black colors).

There is a chance the subpixel rendering trick might depend on the new RGBW setup though, but it seems like they're two seperate technologies.

Applicable to LCDs?

[Gentlewhisper]

I'm sure Samsung won't want to cannibalise their own panel business, but if they make some sort of inline attachment that we can connect to existing LCDs to boast their resolution, wow... its gonna be so cool!

I'm sure everyone will buy one!

Good writeup on subpixel rendering

[IGnatius T Foobar]

Here's a very good writeup on how subpixel rendering works:

http://grc.com/ctwhat.htm

It goes into detail with pictures and everything, demonstrating how the technology takes advantage of the separate red, green, and blue subpixels to achieve additional smoothing.

I'm not sure how Samsung intends to implement "white subpixels" though.

blast from the past

[Tumbleweed]

Ahh, the venerable Apple ][, inspiring people even today!

taters

[maxchaote]

Did anyone else notice that the acronym for this technology is "SPUD Methodology"?

Re:So does that mean...

[rpenguin]

Whoa, you're not getting it, rewt66. It sounds like you think what's going on is simply sampling of higher resolution data down to a lower resolution and claiming it's as good. (i.e. if a white pixel and a black pixel were next to each other, they would be replaced with a single 50% grey pixel.)

That's not quite the technology here. You see, a normal LCD has 'subpixels' which are really just pixels that can display one of the three additive primary colors (red, green and blue.) These pixels are necessarily not in the same exact space, and are usually arranged into rows This means that you can increase pixel resolution at the cost of color accuracy.

Today this technology is utilized by software to provide sharper text display, although if you squint you can sometimes see strange blue and red artifacts around the edge of fonts. Here's an example: close-up of black text on white background

As far as I can tell, the technology here differs in the arrangement of subpixels and the addition of a white/brightness subpixel.

Sounds bogus

[Theovon]

First, it sounds like they're simply scaling 640x480 down to 320x240 with antialiasing. Big whoop.

Second, if they only do a luma blend (ie, ignore the nonlinearity of human perception of light), then it really won't be quite the same thing. I just don't think they're doing it right, because a proper luminance blend is computationally expensive.

Technology from Clairvoyant

[happynut]

The sub-pixel technology was actually licensed from Clairvoyante, and is available to all comers. Clairvoyante calls it a PenTile Matrix.

I know they are working with other panel folks too, so you will probably see more of these type of sub-pixel displays soon.

I wrote a program to simulate it

[Dink Paisy]

I know this is so late that everyone has moved on to the next story, but curious about the idea, I wrote a program to simulate the idea using my shadow mask CRT monitor, and compare it to downsampling and true 640x480. It may work on aperture grill and LCD monitors as well, but it probably won't look as good. Download here. Sorry; it's a windows binary only, and it requires .NET.

For best results set your resolution low, otherwise it has very visible moire patterns. As a side effect of the conversion, the image gets darker. My program also has a colour cast, which the article claims is due to adding the white pixel. The article also says that Samsung has overcome this problem.

It works by setting up the subpixels as a 640x480 square grid, with each pixel consisting of a starting pixel, and the right, lower, and lower right subpixels. Subpixel values are calculated using the average intensity of the corresponding colour value in each of the four pixels the subpixel is a part of.

Visually, aside from the darkness and colour cast which are artifacts of the simulation and wouldn't appear in the real product, it looks decent. It's blurrier than a true 640x480 display, but retains more detail than the 320x240 downsampled version.