Please understand: I didn't say that you should not buy 970P, or that this monitor is bad in general. I have just revealed a widespread misbelief regarding a technical parameter of the monitor.
Despite the common belief, Samsung SyncMaster 970P is not a true 3x8=24 bit monitor, but a 3x6=18 bit monitor that uses "virtual 8 bits" with 2x2 (or similar) spatial (i.e. standing, not temporal, not FRC) dithering. At least the series that is shipping here in Hungary nowadays (2006, January) does, but it's not entirely impossible that there are 8 bit ones as well elsewhere, but then nothing indicates this difference for the customer. I didn't disassembled the monitor to check this 6 bit thing, but, well, it looks very very much like that, and that's what maters after all. Hence, Samsung's claim that it's a 16.7M monitor hardly stands, or maybe it stands because of some mathematical trick, like, they are possibly using bigger than 2x2 dithering (2x2 gives 16.2M mathematically, 2x4 gives even more colors mathematically, etc), but so what, bigger just looks even more "dithered". The point is, when (expert-) people see 16.7M, they expect 8 bits per subpixel, and not 6 bits with whatever spatial dithering. I'm sure Samsung knew that very well. Note however, that Samsung didn't say that it's a 8 bit monitor... it's just that people have deduced this and then spread this misinformation.
What is 6 bits with 2x2 spatial dithering? 6 bits means that a pixel can show only 262.1K colors (64 grades per subpixel, and 64 * 64 * 64 = 262.1K), not 16.7M (256 grades per subpixel, 256 * 256 * 256 = 16.7M). More colors can be imitated with tricks like temporal dithering (also often referred as FRC, it basically means quickly altering the pixel color back and forth between the two nearest colors that the pixel can really display; causes some flickering) or with plain spatial dithering (standing chessboard like pattern at the native resolution of the TFT). With these tricks you can achieve 16.2M colors (253*253*253), almost 16.7M (256*256*256). This montor seems to use 2x2 spatial dithering (or maybe not exactly that, but whatever, it looks like that). As a consequence, you can display only all 4th gray grades clearly, the others will be done with checkboard-like patterns as shown below. At its setting that is optimal for the uniformity of gradients, I can concretelly see the exact same patterns on this monitor at the darker part of gray gradients as shown below.
If we are talking about the full set of 16.2M colors, you get a worst sounding result than with the grays: only each 64th color can be displayed without chessboard patterns! However, dithering is mostly only noticeable at the darker colors, so don't panic if your monitor is already shipping...
I have made some photos. Note that the photos are close-ups so you can see the dithering patterns. The image doesn't look this horrible with naked eyes at all. It's a bit like looking at a newspaper photo with microscope. From bigger distances things melt together, so you the image looks much more uniform. So, the photos:
Note: Watch the photos without any resizing! Some image viewers and browsers like to automatically resize images to save users from using scrollbars, so be careful.
Who should care?
Those who see it in everyday work, which is a minority of the humanity, I guess... However, even if somebody doesn't see dithering consciously, maybe he still would find the image more pleasing with a true 8 bit monitor; I don't know for sure, since I couldn't do experiments. But anyway, I still think that what Samsung (and who knows how many other manufacturers) did is a marketing trick. Like, if all the cheaper models with 6 bit + FRC are 16.2M, then how is this one 16.7M? Well, sure, marketing data are not facts, we have known that already...
I don't see any dithering on my 970P! Why?
Now this is actually not a frequently asked question at all... but I find it important. So, the answer is that it's possible that:
A) Samsung has changed the panel used in the this monitor since then. It's not a rare thing among TFT manufacturers to change the panel of a model during time, and it would be good news in this case. In 2006 October I got an e-mail from a French guy with photos where there was no dithering. So maybe there are different variations. But as usually, you can't know what you get... (Note that a switching from simple spatial dithering to the usual temporal dithering (FRC) while not increasing the number of bits conceals the chessboard patterns. This is an acceptable solution of the problem as far as it doesn't make the display too restless (flickering), however it in itself doesn't increase the number of colors.)
B) Samsung sells different variations of the product in different regions.
C) You are one of the fews who are unable to see the dithering even if (s)he looks for it.
I'm seeing someting strange on my non-970P/770P monitor too... is that the dithering?
It is most probably not that... Banding (unevent brightness steps in gradients) is NOT dithering. Color seepage (lighlty discoloured bands in gradients, like some grays have a touch of blue or yellow) is NOT dithering. (See some more about these here.) Random noise that is typical in the case of D-SUB (analog) connection in the darker colors is NOT dithering. Also, all TFT panel has a bit grided look, because of the black gap between the pixels; that's NOT dithering either. So please don't call them dithering on forums or anywhere. Thank you! By the way, if you do want to see dithering, then look at the gray-gradients.tif (don't zoom in or out the image, and use 24 bit or 32 bit colors in Windows). Examine the block that have 0-25 numbering over it at its left side, from very close. Of course, you will have to do it on a 970P or other TFT that uses dithering.
Are these dithering patterns moving?
No, they are totally standing. It's not what most people know as FRC or temporal dithering. They exactly look like if you were made them with a painting program, which is annoying for some works, because you can't tell if the image is really checked, or just the monitor shows it like that.
What is if you use a lower resolution, like 800x600?
The dithering will still use 1280x1024 resolution. It's possible because it's the monitor that does it, not the video card.
What is if I use 16 bit color?
Since the colors from the video card will not exactly match the "native colors" of the TFT, it still has to use dithering to display them.
Isn't the monitor or video card you have seen defective?
I have seen 2 monitors with my own eyes, also a few photos from other people. Also I have tried my own 970P on 3 different computers (one had a Radeon 9600, another had an old S3 Virge, the third I don't know). Obvious 2x2 dithering in all cases. Also note that I have seen a few photos from a Samsung 193P+ and a Samsung 173P, which also shown 2x2-like spatial dithering (not temporal), and both were PVA and were sometimes marketed as 16.7M color model. Of course, it doesn't mean that all 16.7M color Samsung models do this. It just shows that this trick is not new, so better you are careful when buying monitors.
Did you configured your monitor and video card properly?
Yes, I did. I tried all kind of settings, even different refresh rates. No difference. (And of course I have used 32 bit color mode, etc... no dumb mistakes)
But it is PVA, and PVA is 8 bit! Or not?
This monitor uses PVA (or something like that), definitely not TN. But, what can I say, it just does what I said. And certainly a PVA monitor doesn't have to be 8 bit. From what technical reason should it be that?
Other than with visual experimentation, can your claims be proven?
They can be, but I have no money or qualification to do that. One could open up the monitor (hence losing warranty and maybe even wounding the monitor to death), and see what IC-s are used inside, and then get the data sheets of the relevant IC-s (source drivers, timing controller, etc).
How is it that no big reviewer site has noticed this?
Now that's The question. I have only theories, but I'm afraid at least one of them has to be true. One theory is that Samsung sent special (more expensive to manufacture) monitors for reviewing, and then sold cheaper ones for the customers (or at least to some customers; see the first FAQ). Another theory is that all reviewers are visually disabled. You see, it's really not the kind of phenomenon that's hard to spot. Anybody who is a bit competent and not visually weak should spot it immediately when he looks at the darker part of a gradient, and reviewers regularly do test the monitors with gradients. So the third theory is that some reviewers are paid to see what they have to see... but this must not be all reviewers, not even the majority. But still, I would like to see some of the well know reviewer sites to give some explanation. Oh, as an addition to the story, some say that lowering the color precision to 6 bits is a technical compulsion for achieving better response time, but since MVA/PVA panels have the marketing image of being 8 bit, manufacturers try to hide this sacrifice they made for the better speed. OK, but then the monitors that were sent for testing had either show the dithering effect, or had to be too slow. But they was said to be fast enough and having great colors (I'm not talknig about the deltas... that test doesn't point out dithering). So how was that? Something is not right here.
What about 971P? Does it use spatial (i.e., standing, not temporal, not FRC) dithering too?
Certainly not. I got photos from two 971P-s that were bought at the end of 2006 in
Hungary, and neither did this kind of dithering. But of course, you never know when do they change it, or if they sell this model with different parameters somewhere...
Back to the main 970P page...
Feedback: ddekanyREMOVETHIS@freemail.hu; please delete the REMOVETHIS from the address (that's against spam)!
