How to Pick a Good LCD

Picking an LCD goes well beyond brand recognition. Below, we have a small introduction to a few different facets of shopping for an LCD - in our order of importance. Generally, we find a price point and then choose an LCD based on the properties detailed below. For example, if we only have $500 to spend, we consider all of the monitors for $500 or less and go through the following checklist.

Input Type: DVI, DVI, DVI. We insist that when you buy a new LCD monitor, you buy a model with DVI capability. Even if you don't want to buy a DVI-capable video card right now, it is still a wise decision to get a DVI-compatible LCD. When DVI first appeared in the industry, there were a few issues with the quality of the connectors and thus, sometimes viewing a signal over a DVI connector would give you a worse signal than over a 15-pin D-Sub connector. However, in the last 3 or 4 years, most of those problems have been fixed, and virtually every new video card is more than capable of producing a clean digital signal. None of the LCDs that we review today (except the Samsung 910V) are limited to only analog connectors, but be aware that they exist.

Resolution, Aspect Ratio: If you buy a 4:3 LCD, the resolution had better be 4:3 as well. That is, a 19" LCD should have an aspect ratio of 1600x1200, 1280x960, or some derivative thereof. Most 19" and 17" LCDs have an aspect ratio of 5:4 (1280x1024). This is OK, but you're looking at a 5:4 signal crammed in a 4:3 box. Our Dell 2001FP, on the other hand, measures exactly 16" by 12" and runs at a resolution of 1600x1200. Generally, a skew aspect ratio is not enough to notice, but if you do any sort of graphic work, all of your circles will look like ovals. This goes the same for widescreen LCDs - buy a widescreen LCD with a widescreen resolution; a 30" wide format LCD with a resolution of 1024x768 will not look correct no matter what you do to the signal.

Response Time: Response time is an unusual preference and always a trade off. Typical response time (TrTf - Time rising, Time falling) refers to the time that it takes the LCD subpixel to twist from the fully "on" position to the fully "off" position and then back again. Response time has absolutely nothing to do with framerate. Pixel response times are independent of each other, and it does not take the entire screen 25ms to refresh if a monitor is labeled as a 25ms response time LCD. The time that it takes the LCD to go from black to white may be 15ms while the time that it takes the LCD to go from black back to white may be 10ms. Furthermore, your monitor is generally rendering a color that is not on end of the color spectrum. The time that it takes your LCD subpixel to twist from one half of a tone to another may be more or less than 15ms. The TrTf response time is normally a pretty useless measurement - but it makes for an easy specification in which to market LCDs.

The second method in measuring response time is "gray-to-gray" (GTG) response time. The measurement of GTG response time is actually more useful to LCD buyers, but it is harder to convey and is usually just conveyed as one number (which is incorrect). Gray to Gray response time refers to the time that it takes for a pixel to twist from some arbitrary position to another. On a 6-bit LCD, that's the time it takes the subpixel to twist from 1 of 64 different positions to one of the other 63 positions. GTG response times are useful if the manufacturer expresses the average of all the GTG response times, but that is rarely the case.

Everyone's preferences on response time are different. If you play a lot of games and feel that the few ms difference between a 6-bit LCD and an 8-bit LCD are worthwhile, then it's a worthy investment. Most people can't tell the difference - and that's not just most people who aren't gamers, but most people in general have to be shown the differences between two displays that differ by single digit transient response times.

Index How to Pick a Good LCD (continued)


View All Comments

  • benk - Tuesday, November 30, 2004 - link

    I just (like two hours ago) got my Dell 2005FPW...played an hour of CS:S. Didn't notice any ghosting, blacks were all black, etc. I have my desktop set up stretching across this LCD and my old Sony Trinitron 17" and the color and sharpness on the LCD is markedly better. It actually surprised me; I thought I was giving up color in trade for a wider monitor that's a little easier on the eyes. Nope. It looks great, plays great, and, according to my girlfriend, is lots more stylish. Reply
  • IceWindius - Tuesday, November 30, 2004 - link

    Screw LCD, SED is the wave of the future.

    Until then, i'll stick with my Viewsonic CRT monitor, thank you.
  • archcommus87 - Tuesday, November 30, 2004 - link

    Can anyone go back to the issue that someone asked prior about running non-native resolutions? My biggest deterrent about LCDs was always the fact that if I run my desktop at 1280x1024, I have to run all of my games at that, too. Sorry, but unless I'm buying two video cards a year, that's sometimes hard to do.

    Can you use other resolutions without getting crappy images?

    Yes, at times I have considered selling my 19" and 17" CRT dual monitor setup for one, single 19" LCD. But then I think, nah I love my Philips, and two monitors is cool. Plus I'd hate to have to run all my games at such a high res.
  • nullpointerus - Tuesday, November 30, 2004 - link

    Cat: The lower the refresh rate on the video card, the more sluggish the mouse feels to me. Anything below 75 Hz feels terrible. Setting it up to 100 Hz (assuming your card and display support it) feels extremely fluid. I'm just suggesting possibilities, so YMMV.

    TCfromNL: From what I can tell, the article doesn't make any such claims about whether you have problems if the GIF appears dithered on your display; it's just presenting a visual aid.
  • coldpower27 - Tuesday, November 30, 2004 - link

    It depends on each peoples eyes, I guess it's not quite as noticable to some. Also not all the 25ms screens are created equal.

    I have a 25ms LCD, and I don't really notice ghosting, I got it over 2 years ago, though when it cost alot. Though 25ms for an LCD to do continous motion as that enough to generate 40FPS, I also don't really play that many FPS. The new LCD's that are capable of 12ms are amazing that like double the FPS at maximum.

    Yeh it would have been nice to test some of the newer LCD panels as well, but to me I don't know why people want it so bright, my LC is around 350:1 range and I already fine that awfully bright, 800:1 just seems so much :S
  • drinkmorejava - Tuesday, November 30, 2004 - link

    I'm confused, how can all those 25ms monitors have no noticeable ghosting. I've always known that a black-white measurement does not truly show how much ghosting there will be, but a 5? Reply
  • TCfromNL - Tuesday, November 30, 2004 - link

    Okay. Small problem with the 256-shades-of-blue thing, referred to as "the image below".

    Since it's a GIF, it only contains 256 shades of color, tops. Including all the greys.

    I imagine you have a losslessy (or un-)compressed 24-bit copy somewhere. Still, it's not nice to scare your readers by displaying some 20-shades-of-blue thing while saying that if it doesn't display smooth as a baby's skin, which it doesn't, the viewer's monitor is at fault.

    Further, nice article. But since these monitors are all 1280x1024 (except for the 20" Dell), I can't help but leave disappointed. I don't like squarish monitors. There's a reason why TV evolved from 4:3 to 16:9. I agree with the cry above: IT DOES NOT MAKE ANY SENSE.
  • gwynethgh - Tuesday, November 30, 2004 - link

    Now to find a good but reasonably priced DVI KVM switch. Reply
  • KristopherKubicki - Tuesday, November 30, 2004 - link

    Klah: have they benchmarked any units using that methodology except the example? I checked around and couldnt find any.

  • klah - Tuesday, November 30, 2004 - link

    "klah: I was only aware of Xbitlabs doing so. We feel that the methods for measuring reponse time thus far are OK, but not represent gray to gray response time measurements well. Its something we are working on and we will probably have a better methodology before the next roundup.


    Here is Tom's methodology:

Log in

Don't have an account? Sign up now