How is it measured? Are their any programs to test this? What does www.tomshardware.com use to test the response time over a broad range?
Thanks!
Thanks!
How is LCD response time measured?
- Thread starter ArchAngel777
- Start date
Ok, that is one part of my question answered. How is Toms Hardware able to to test this? Secondly, any programs that exist to test this? Or would it be a device? If so, what device?
Ronin
Diamond Member
Yes but you know TN panels are ISO 8ms. PVA panels w/ overdrive like VP191b are different tested. Gray to gray and black to white
Ronin
Diamond Member
Perhaps you missed the point. The point was, there is little consistency between companies when measuring response time, and how it's measured. What it does is create confusion for the consumers, who think they might be getting, say, a 4ms LCD, when it's really a 16ms, based on the accepted standard.
Not sure if this was directed at me, but I didn't miss the point one bit. I know that a manufactur has a different way to test theirs and therefore "cheat" the system. I have always known this and that is *NOT* my question. It never was... Toms hardware does an actual test on these LCD's to find the actual latency over the entire range. This test is therefore "accurate" and compares apples with apples when dealing with multiple LCD's. That is all I am asking for... Toms Hardware is able to test the latency and I want to know how it is done. How is he testing these? He isn't using the MFG claim, he is testing them.
Now, do you undertsand my question?
Isn't it surposed to be the time taken from a pixels crystals to go from completely off to completely on and off again. Hence white to black to white.
If so, why not state the time as frames per second, as 25ms panel should be able to go from white to black and back again 40 times a second. CRT's do have lag in terms of the time is takes for the phosphor to go completely black. So a 50hz crt would be slightley inferior than a 20ms panel in terms of update speed.
If so, why not state the time as frames per second, as 25ms panel should be able to go from white to black and back again 40 times a second. CRT's do have lag in terms of the time is takes for the phosphor to go completely black. So a 50hz crt would be slightley inferior than a 20ms panel in terms of update speed.
If I recall correctly, it actually takes longer to go from certain grays to other grays the longest. I was just skimming, so maybe I understood it incorrectly.
Exactly, which makes the whole way they are tested questionable. Some companies test from 75% grey and others at different percentages.
The official (ISO 13406) definition of response time is the time it takes to go from black (brightness = 0) to white (brightness = 255) plus the time it takes to go from white to black. Because of engineering constraints, you actually measure the 10% to 90% amounts of the transition (to remove things like noise).
The problem is that pixel response times tend to be worse for gray transitions than for white/black ones (before things like overdrive is implemented). By gray I simply mean shades of color somewhere in the middle, i.e. light blue or dark red, not necessarily gray per se. The reason is that although the liquid crystal has less distance to travel in gray transitions, they also move more slowly for smaller transitions, so much so that gray transitions can take several times longer than white/black transitions. So a 25 ms rated (black to white to black) monitor can actually be like 50 ms or more in terms of actual use, because when we use our monitors, we use the gray portion mostly, rather than strictly white/black.
However, there is no standard for gray transitions, so manufacturers make up their own ways of measuring them. One thing that manufacturers immediately dispensed with was the notion of going one direction and then the other, but to go one direction only. This effectively cut the reported time in half, as opposed to the previous method (ISO used one direction and then back again because one direction takes much, much longer...the ratio is something like 3 ms vs 22 ms or something like that for 25 ms monitors). Also, because there's effectively 256 * 255 = 65280 transitions to choose from, manufacturers naturally go with the faster ones and report those as their measurement technique. Because of this, response times don't really mean that much anymore; while they've improved the past several years, the improvement isn't anywhere near what manufacturers would want you to believe. Ever notice how we used to have 25 ms monitors, then 21 ms, then now 16, 12, 8, 6, and 4? As opposed to seeing 7 ms or 9 ms or 13 ms response times? Because manufacturers can choose their own measuring methods (under gray to gray), they pick whatever sounds better from a marketing standpoint (yes, even numbers tend to sound better). It's not as if response times only improve in 2 ms or 4 ms increments.
You can see how Tom's Hardware does their testing here:
http://graphics.tomshardware.com/display/20040923/index.html
Note that they modified things slightly. Previously, the measurement uses 10% to 90%. However, now it's 10% to *staying within 10% of the final value*. This is because the VX924 goes over the requested amount significantly, then settles down to it over the next several frames, and is a better engineering definition (just not used previously because for transitions, you don't usually expect it to go over significantly, so 10% to 90% was good enough). For example, the example Tom's uses is 0 to 175; the VX924 goes from 0 to 220 or so (forgot, not gonna check it out right now) then settles down gradually to 175 over the next two frames. However, Viewsonic only counts the time it takes to go from the initial 0 to 175 as it zooms past 175 to 220 as the response time...neglecting that it takes two frames to *stay* within 10% of what was desired. Under this definition, manufacturers can really pump their response times by just doing a bunch of black to white transitions when gray to gray transitions are requested. After all, it's quicker to overshoot than to actually stay in a target area; drag races would take much longer if those cars were required to *stop* at the quarter mile mark rather than being allowed to overshoot them and slowing down after reaching the quarter mile.
The problem is that pixel response times tend to be worse for gray transitions than for white/black ones (before things like overdrive is implemented). By gray I simply mean shades of color somewhere in the middle, i.e. light blue or dark red, not necessarily gray per se. The reason is that although the liquid crystal has less distance to travel in gray transitions, they also move more slowly for smaller transitions, so much so that gray transitions can take several times longer than white/black transitions. So a 25 ms rated (black to white to black) monitor can actually be like 50 ms or more in terms of actual use, because when we use our monitors, we use the gray portion mostly, rather than strictly white/black.
However, there is no standard for gray transitions, so manufacturers make up their own ways of measuring them. One thing that manufacturers immediately dispensed with was the notion of going one direction and then the other, but to go one direction only. This effectively cut the reported time in half, as opposed to the previous method (ISO used one direction and then back again because one direction takes much, much longer...the ratio is something like 3 ms vs 22 ms or something like that for 25 ms monitors). Also, because there's effectively 256 * 255 = 65280 transitions to choose from, manufacturers naturally go with the faster ones and report those as their measurement technique. Because of this, response times don't really mean that much anymore; while they've improved the past several years, the improvement isn't anywhere near what manufacturers would want you to believe. Ever notice how we used to have 25 ms monitors, then 21 ms, then now 16, 12, 8, 6, and 4? As opposed to seeing 7 ms or 9 ms or 13 ms response times? Because manufacturers can choose their own measuring methods (under gray to gray), they pick whatever sounds better from a marketing standpoint (yes, even numbers tend to sound better). It's not as if response times only improve in 2 ms or 4 ms increments.
You can see how Tom's Hardware does their testing here:
http://graphics.tomshardware.com/display/20040923/index.html
Note that they modified things slightly. Previously, the measurement uses 10% to 90%. However, now it's 10% to *staying within 10% of the final value*. This is because the VX924 goes over the requested amount significantly, then settles down to it over the next several frames, and is a better engineering definition (just not used previously because for transitions, you don't usually expect it to go over significantly, so 10% to 90% was good enough). For example, the example Tom's uses is 0 to 175; the VX924 goes from 0 to 220 or so (forgot, not gonna check it out right now) then settles down gradually to 175 over the next two frames. However, Viewsonic only counts the time it takes to go from the initial 0 to 175 as it zooms past 175 to 220 as the response time...neglecting that it takes two frames to *stay* within 10% of what was desired. Under this definition, manufacturers can really pump their response times by just doing a bunch of black to white transitions when gray to gray transitions are requested. After all, it's quicker to overshoot than to actually stay in a target area; drag races would take much longer if those cars were required to *stop* at the quarter mile mark rather than being allowed to overshoot them and slowing down after reaching the quarter mile.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter