6

I'm using a 32" 'LED' TV as a monitor (1920x1080 over HDMI). I've been noticing some annoying horizontal stripes where the source is displaying grey pixels.

It seems that what's happening is the TV is using multiple dots/sub-pixels vertically for each source pixel. And instead of having all the sub-pixels for a given source pixel at the same intensity, to produce dimmer pixels it switches off some of the sub-pixels, which I'm sure is fine for video, but is a bit ugly for text.

EDIT: new photo added, previous ones are below. In this new one, I've stitched a digital zoom of a gradient I generated to a photo of how that gradient is rendered on the TV (ignore the slight incline, that's my fault).

This time the photo is from an old Sony DSC-W70 in macro mode. You can clearly see the different sub-pixels' intensities changing at a different rate, and it's clearly a pattern rather than bad pixels...

enter image description here

In the previous images below, you can see bright sub-pixels in the middle of the dark text strokes.

Is this a standard technique for current displays? Is it because I have a TV and should really be using a proper monitor?

(Apologies for the photo quality - taken on an iPhone through a magnifying glass).

source

pixels

zoomed

CupawnTae
  • 175
  • 7
  • I know subpixel rendering is used to increase the horizontal resolution when drawing text by drawing specific colors around the edges of the text to turn on specific pixel components. Sometimes this can get screwed up if the subpixel rendering code settings do not match the pixel configuration of the display. However, I don't think it is possible for that to fall like this. – alex.forencich Feb 21 '15 at 01:57
  • 1
    Interesting indeed. Intensity variations is often achieved on LCD by using fast switching of pixels (faster than the liquid crystal itself). I suppose that using several pixels on your TV allows a greater dynamic range. – TEMLIB Feb 21 '15 at 02:00
  • Many TVs have a PC input setting somewhere (often well hidden) in the menu that disables overscan and disables features that screw up text like that. Does yours have such a setting? – Grant Feb 21 '15 at 02:55
  • @alex.forencich yeah, I don't think there was any subpixel rendering on the initial text examples other than the pure grayscale anti-aliasing. I've added a new example which is a pure grayscale gradient, so there should be no subpixel stuff from the source at all. – CupawnTae Feb 21 '15 at 17:22
  • @Grant yeah, I had gone through every menu option one by one to make sure - aspect ratio is set to "Just Scan", and sharpness (which seems to add artifacts) is set to 0. The only other "PC setup" options apply to the VGA input only and are for size/position/phase, so no help there. I also tried all 3 HDMI inputs just in case one might be more PC-friendly (have seen that in the past) and no difference there either. – CupawnTae Feb 21 '15 at 17:25
  • @TEMLIB I'm pretty sure that's what's happening here unfortunately. I'd love to be able to say "give me fewer intensity levels but solid pixels", particularly when I'm reading text, but I don't seem to have that option, and looking at the gradient I added today, everything between very light and very dark gray has some amount of "dithering". So I'm probably stuck unless I completely disabled antialiasing somehow (if that's even possible), but that would probably be even uglier. – CupawnTae Feb 21 '15 at 17:28
  • Please see Chroma Subsampling, here is a wikipedia link: http://en.wikipedia.org/wiki/Chroma_subsampling#Sampling_systems_and_ratios – rioraxe Feb 21 '15 at 21:17
  • @rioraxe interesting reading, and indeed my TV fails that test too, however, it's not relevant to this particular issue. For one thing, my issue manifests even in the middle of a large block of a single grey colour, which chroma subsampling wouldn't affect. Thanks for making me even less happy with my picture though ;) – CupawnTae Feb 21 '15 at 21:46
  • I have noticed this on both LCD TVs in my house. Note that it's actually happening within individual subpixels; in your case the top half of a subpixel dims more before the bottom half. I haven't seen this technique on any LCD monitors or phone screens, probably because it's not great for text as you've shown. – binaryfunt Feb 14 '20 at 15:52

2 Answers2

1

Does it appears like vertical blending from incorrect row number sync? Like 1920x1200

Make sure the TV is pixel sync'd to the video signal using the onscreen menu (often advanced mode ?) to ensure no overscan on vertical, horizontal and no aliasng.

To verify this , use one of the first test pattern with alternating black white pixels then alternating 2 pixel pairs, then gradient colors to check gamma for each color. http://download.cnet.com/Dead-Pixel-Tester/3000-18512_4-82448.html

DPT.exe can also be used to toggle and small zone to restore dead pixels by discharging with AC video pattern.

Guest
  • 11
  • 1
  • Off the top of my head I would say this is not it. The mode you're talking about is called "Just scan" on this TV and is enabled. And I don't see evidence of blending. The pattern is regular, and seems to be exactly the same for every grey pixel - one source pixel corresponds to two rows of screen dots, and for mid-greys, one row is significantly darker than the other. But I'll check out DPT to see if it turns anything else up, thanks. – CupawnTae Feb 25 '15 at 11:55
0

There seems to be no issue with your top pair of pictures. You are displaying a vertical dark line, and the individual color dots are dimmed correspondingly. I'm not sure what you expect to happen other than what is happening.

The other cases are because you are looking at detail you are not intended to see. The system only works if you can only resolve the screen to the point that a triad of color dots appear as one. That is how the appearance of continuous color is achieved, even though the colors come at discrete locations for each of red, green, and blue.

Your monitor may be further using multiple clumps of RGB triads to display whole pixels. There is nothing wrong with that. Part of your confusion is apparent from you calling the individual color dots pixels. Neither the individual dots nor groups of them that form complete color triads necessarily map directly to pixels. Pixels is the data into the monitor, and color dots is what comes out. When viewed as directed, which means at least whole triads should appear as single items, this works out. Basically, you are poking around where you're not supposed to be.

If the monitor is using, for example, a 2x2 array of triads per input pixel, then there may be more artifacts, but only if you're looking at details you're not supposed to. Individual triads could have lower intensity resolution, but the overall clump of triads representing a single pixel still have the advertised resolution. 2x2 dithering, for example, give you roughly 2 extra bits of intensity resolution at the cost of some spacial resolution. If the result is still within the spacial resolution of the incoming data (the pixels), then this is a valid implementation method.

The above may be what's happening inside the top of the "a". There may also be various low level "sharpening" and other "enhancements" and compression artifacts. Again TV is meant to be viewed far enough back that you can't resolve spacial details below a pixel.

Olin Lathrop
  • 310,974
  • 36
  • 428
  • 915
  • Apologies if my terminology was confusing - in general I was calling the individual dots "sub-pixels". I understand what you're saying, but my problem is that it appears to be using vertically-stacked pair of triads per source pixel, but varying the intensity of each of the pair *independently*. In certain circumstances (mostly dark text on a white background, where antialiasing adds grey underneath), this results in very visible artifacts - almost like the letters are underlined. In the middle of the gradient photo, you can see this as very obvious striping. – CupawnTae Feb 21 '15 at 18:06
  • Having said all that, if (as I suggested in the question) you're basically saying, yes it's because it's a TV and not a computer monitor, then I may have no choice but to accept it. However, if I bought a computer monitor (which will typically be viewed at much closer-range than TV I admit) and it behaved like this, I would be returning it as unfit for purpose. – CupawnTae Feb 21 '15 at 18:15