1rst LCD at 100 Hz: the end of afterglow - BeHardware
>> Miscellaneous >> Monitors

Written by Vincent Alzieu

Published on October 17, 2006

URL: http://www.behardware.com/art/lire/641/

Page 1

Test of the 1rst LCD at 100 Hz

Gamers, the future of LCDs is at 100 Hz !
Partially unveiled at CeBIT 2006, 100 Hz technology is coming for LCDs. The idea is that monitors will no longer display 50 or 60 images per second but 100. How it does this is something we will go into later. What is important is that Samsung is headed in the right direction. Finally, LCD technology is considerably improving much more than going from 16 to 2ms. Under certain circumstances, the first 100 Hz LCD is as good as a CRT.
Testing the Samsung LE4073BD

The only problem is that once again the technology doesn’t come on a computer monitor but on 40" TV. Not testing this, however, seemed foolish to us and we were right. This TV really is different. For 100% computer use, similar technology will be only available in 2007. Here is a preview of monitors that will be released next year…

100 Hz technology is present in many projects by many manufacturers and they all say that the future of display technology is headed this way. To succeed, there are several possibilities. We are going to start with the ones that Samsung hasn´t chosen before ending with the better one, or their solution.

One method described in Samsung´s website consists in adding 10 images per second. Now you start thinking about this and you realize that there is something wrong. How does 60 Hz + 10 images = 100? Obviously there is another explanation and it has nothing to do with the technology that Samsung recently introduced in its TV. It’s actually a trick used by some plasmas to improve image flow. For others like LG and Philips, it seems that the targeted 100 Hz would consist in introducing a black image between two colored images. This action doesn´t have an impact on the flow of images or on real afterglow but rather on retinal persistence. The idea is to recreate an artificial screening like we used to have on CRT monitors. The black image "cleans" the eye from the previous image. To understand retinal persistence, look at a lamp 2 seconds and look elsewhere immediately after. You will see a very bright spot. If you look at the light bulb only one second, you will see that the retinal persistence is much lower.

According to Samsung, which also tested this method, this technology does noticeably reduce afterglow, but also produces a twinkling on the panel similar to tubes clocked at 60 Hz.

Either way, it’s interesting to note that alternating images with black ones at 100 Hz means a change approximately every 8 ms. Even with 4ms monitors and lower, with a couple of exceptions, this frequency is too quick for liquid crystals to turn black. Chances are that the intercalated image might not be black but identical to the previous one and be less intense.

BenQ explores a similar path with an alternation not of black but gray images. We will have the opportunity to test their system soon and we are eager to see it, because they decided to include it in a 24" monitor.
And now Samsung’s 100 Hz
Samsung´s solution is different from the previous ones. We will start with the theory first of all. The monitor tested is a European model, which means that the initial frequency isn´t 60 Hz like US TVs (for example), but 50 Hz. By increasing to 100 Hz, Samsung doubles the flow. The "easy" solution would have been to double the number of images but this isn´t what Samsung does. According to the manufacturer, a processor calculates one intermediate image between the two sent by the DVD player or computer. It’s an intelligent real time calculation.

Here is the marketing claim of this technology. Don’t take it too literally, but nevertheless it does illustrate the process quite well:

On the left, a 50 Hz sequence and on the right, the result at 100 Hz

In use, the result is very appealing, whether it’s with a DVD player or computer. With the 100 Hz deactivated, there is some afterglow. With 100 Hz activated, there is a new image flow, the afterglow disappears. Amazing!

Page 2
100 Hz in games / movies

100 Hz in movies, in games
We connected the TV to three sources: a rather high end DVD player (Philips DVD9000S), a recent game console (an Xbox 360) and a computer (home assembled).

First off, it’s possible to activate the 100 Hz in all three situations and the difference is obvious with or without. After that, results and improvements vary depending on the utilization. It’s important to single out movies from the rest.

100 Hz in moviesthe fluidity of movies is considerably improved. This is surprising but is even disturbing in some cases. Cartoons and computer animated movies look better overall and we appreciate that afterglow has been significantly reduced. The result is close to a CRT monitor for the blur effect seen in movement, moreover, we discovered a new fluidity, which isn´t obtainable via standard CRT monitors. Characters seem to float a bit, but all in all this is rather nice. We add, however, that cartoons were already very well rendered on normal LCD and Plasmas.

The theoretical explanations for the calculation in movies.:

Pro and personal movies are shot in 24, 25 or 30 images per second.
Let´s take the case of a frequency of 25 fps. With a standard monitor at 50 Hz, the framerate is twice as high. Each image is displayed every 40ms. Our standard 100 Hz monitor will calculate one image and in principle display two different ones in the same period of time. This will considerably modify the fluidity of the sequence.

There is a second possibility presented by Samsung, which is even more unfavourable. The initial movie framerate is 24 images per second or 1 new image every 41.7 ms. At t = 0, the monitor displays the first image. The monitor then asks for a new image at t1=1/50 Hz or 20 ms, where the first image is doubled with a standard monitor. At T2 = 40 ms, a new image is requested. As we are still with the movie framerate (1 every 41.7 ms); the image is tripled. The next image will only be displayed at the next request. In the meantime, according to Samsung´s presentation:

When the 50 Hz monitor displays three times the same movie image, the Samsung TV is supposed to display 3 different ones.

Now back to reality. If it was good with computer animated sequences, the images with real people are in fact less enjoyable. This floating of characters and new fluidity are quite disturbing, because they aren´t natural. Also, this is no longer a movie type rendering and it’s too real. We realise here that we are actually used to seeing a slightly blurred effect in fast camera movements. We all agreed that we wouldn’t activate the 100 Hz in movies.
However, maybe this is something we will get used to, because we only had the monitor for a week in our office.

100 Hz in games

Here, we had the exact opposite situation. Once we tried the 100 Hz, whether with the computer or console, it is impossible to go back to the previous configuration. And once again this was unanimous. What was amusing was that some people in the office say that they weren´t affected by LCD afterglow…. until they tried the 100 Hz. Samsung is right not to talk about response time anymore as this is no longer relevant in games. Liquid crystals should always have a response time of 8ms but afterglow – in fact the retinal persistence – is no longer comparable to TVs and monitors equipped with similar crystals.
The improvements are so nice that none of the 2ms monitors are able to compete with this TV. The end result is very nice and even soothing for the eyes.

So this is the reason why we wanted to take an even deeper look…

Page 3
100 Hz : closer look - 1

A closer look at 100 Hz
A car moves from left to right at high speed.

Movement isn’t perfectly fluid. Depending on its speed, the car is shown in several successive positions. If the car goes very fast, the positions are very close and the eye perceives a flowing movement.
Perfect monitor
monitor with 3 ghost images

A monitor without ghosting effects would have previous images completely fading away when a new one appears. This is the theory and in practice, it´s often not the case as images fade progressively. Sometimes up to 5 afterglow images remain on the monitor and represent the visible white trail behind objects. Some monitors have strong overdrives in addition to image anticipation algorithms. In this case, an image can appear in front of the main object, creating a white halo ahead of objects in motion.

With CRTs we captured afterglow with a camera at a shutter speed of 1/60 seconds as compared to 1/1000 s for an LCD. We take 50 pictures per test. We then can see a monitor’s ghosting effects, or all the car’s position in the entire process. The most important image is the one on the left, the better one. It will be the most displayed on the monitor, while the one on the right is in transition.

Here are the two extreme states with each monitor as afterglow oscillates.

The result deserves a closer look. It is easier to do so with the "worst" image:

We see that there are twice as many images at 100 Hz, as the smaller gap between two consecutive images shows. However, even if our eyes perceive a lower afterglow, or none at all, there are still ghost images behind the main one, even in the "best" picture at 100 Hz. Even more surprising, the brightness of persistent images at 100 Hz is a bit stronger than at 50 Hz. The above shows this at 100 Hz, and Samsung doesn´t only send twice the amount, but it could also add an inversed overdrive to accelerate the erasing of previous ones.
This system seems to be efficient. Look at the worst cases:

At 100 Hz, the monitor displays one image – calculated and not sent by the graphic card – every 10 ms. To erase it, Samsung sends an inversed tension to liquid crystal cells. The crystals answer quicker, but obviously the end positioning is still a bit inaccurate. The acceleration makes them slightly exceed the targeted position and the result is a lighter tracing of the background color. You can see it in the lighter outline behind objects in motion. This is probably why we seem to see objects singled out so much from their background.

So, with this analysis we better understand why the length of the afterglow shadow, especially due to colors, is shorter. The above image shows that it is half as long.
Another phenomenon adds to this. Each image will be displayed 10 ms at most instead of 20 ms. The intensity of retinal persistence depends amongst other things on the level of brightness and how long it is seen. The lower display time at 100 Hz would reinforce this sensation of lesser afterglow.

Page 4
100 Hz : closer look - 2

100 Hz : an even closer look!
We see that in the car test, the processor adds images between the two sent by the source. This is rather easy in the common case of a fixed object in lateral motion. In a game or movie, most images are minor modifications of the previous one. The test in the previous page represents what happens in most transitions. But what would happen if the T+1 image is radically different. What would the processor do? Would it really calculate, as Samsung says, an intermediate image?

To see this, we created a few scenes in which objects change shape or position every 20 ms. At 50 Hz, each new image is radically different from the previous one. For example, one of our sequences represents a simple triangle that spins, moves from left to right and changes color. In one step it turns from green to blue. Most of the time, here is the result:

Green and then blue, the triangles are well distinguished, and then the processor fails in the image calculation. Sometimes, it doesn´t add anything and our sequences indicate the same thing as soon as two consecutive scenes show strong disparities. At other times it does, however, and the image look like this:

The "intermediate" image is very approximate, and we could even say that it is a complete miss.

We had the same observations for another sequence, this time after changing a square into a triangle:

There isn´t necessarily an image added between the two received, but this isn´t systematic. If the processor isn´t powerful enough to calculate real complex intermediate images, it doesn´t try if it is too complicated. These are obviously the first steps in this domain and this is the first time that we could test such a system. Either way, the first results are very impressive and encouraging.

After verification, our impressions were justified. The calculation of the intermediate image is based on a vectorization of the area of the image, cut out into 8 x 8 pixel squares:

It is normal that for now we saw real changes only on objects that only go though minor changes without strong deformations.

Page 5
Normal LCDs: 60 or 75 Hz

Normal LCDs: 60 or 75 Hz ?
Our conclusion from tests, most of all our impression from games, is unanimous : 100 Hz rendering is much better than 50 Hz. Response time doesn´t change, but afterglow is much different. Increasing frequency considerably improves the fluidity of sequences and resolves the problem of persistent images due to the insufficient reaction time of crystals.

We logically came to ask ourselves about the optimum frequency of normal LCDs. Most support 60 or 75 Hz. We did a blind test of 6 monitors at the two frequencies. In the first case, there is theoretically one image every 16.7 ms and for the other this figure is 13.3 ms. Compared to our example of increasing from 50 to 100 Hz, we should logically get closer to the reaction time of 100 Hz at 75 Hz with our LCD monitors….but this isn´t what we found.

We found the following:
  • 2 didn´t support 75 Hz and we would have a black or unstable image.
  • 2 said that they supported 75 Hz, but when we measured the time between images we realised that they were in fact at 60 Hz.
  • Finally, the last two really ran at 75 Hz…partially. In fact the monitors really displayed four images and then skipped the fifth. The sixth one was displayed normally. When we looked at the results, we realised that this skipped 5th image was to resynchronise the monitor at 60Hz. In fact, it really displayed 4 images in 67 ms whether it was at 60 or 75 Hz.

    75 Hz with the 10 20 35 W, everything is fine during the four first images
    The fifth one jumps, back to normal at the sixth

    Starting a game confirms this jump and we see small cuts that were hard to ignore.

    We should test this function with more monitors as some may support 75 Hz. For now, however, the ViewSonic VX922, or the Belinea 10 20 35W can´t (they were included in the panel).

    But to be honest manufacturers told us that none of the current LCDs would be able to support 75 Hz. We don’t always believe them, but out of 6 monitors this was actually the case. We hope that they aren´t right and we invite you to test this at home. Logically, if your monitor is able to support this frequency, the sharpness in games will be better.

  • Page 6

    Finally, a real progress in games. Three cheers for 100Hz! We just have to point out that the miracle isn´t yet complete as movie rendering didn´t quite convince us. Fluidity and sharpness were ironically a little too much and it didn’t seem like a natural rendering. Too bad this system was implemented on a TV. For movies none of us liked it, however, 100 Hz rendering in games, sporting events and even TV shows was good.

    Now back to games. Whether it’s with the computer or game console, 100 Hz brings a considerable improvement. It is so good that for the first time it looks very much like CRT rendering.
    Now, this improvement is so significant that one silly question came to our mind: why work on interpolating an image to double the frequency and not impose real support of 100 or 120 Hz for graphic cards or game consoles? We agree on the fact that none of the current consoles are currently able to sustain this frequency, but our computers do. At least those who have a very good GC and /or who play games that aren´t too performance hungry. (With vertical synchronisation if your card couldn´t hold the 100 fps, you could reduce it to 50). For now there are no answers to this question and we should directly ask panel manufacturers.

    The improvement to a real 100 Hz would be all the more useful because the current solution has at least one inconvenience. Let´s say that the cards send at a frequency of 50 Hz images N and then N+2 and that the monitor calculates the intermediate N+1. To calculate this image, the TV would have to store in memory the previous image (N) and N+2. This mean that it will have to know the N+2 and consequently postpone the images displayed.

    We recently realised that LCDs have a natural tendency to be late compared to CRT monitors. Here we would voluntarily increase calculation time to display a new image. With a stopwatch in hand and compared to a CRT in clone (with VGA input and numeric for the TV), we measured a delay in 20 consecutive images:

    The minimum delay was 10 ms. The calculation can be very fast especially when the processor fails to calculate the intermediate image. We saw that sometimes changes were too complex for the TV to perform the calculation. In our case, with the stopwatch displayed on the monitor, variations were too significant for the TV to create intermediate images.

    However, it is possible to have a much higher delay as is shown in the graph (up to 60ms). This represents 3 full images at 50 Hz. This is higher than the average of monitors previously tested. We remind you that a monitor like the ViewSonic VX922 only has delay of 2ms compared to CRT monitors…

    Copyright © 1997-2015 BeHardware. All rights reserved.