Overdrive - 1 : ISO, grey

The magic word: "overdrive"

After neglecting it for years, manufacturers realized that monitor reactivity is the determining factor for quality. You must not think that crystal reactivity has always been the main objective for manufacturers, who have told us for many years that it was none of their clients’ concerns. Also there are still some (fewer and fewer every day, fortunately) who don’t want to consider the problem of dead pixels as being a real issue and who still don’t want to provide the least guarantee on this point to their clients. But that is not the point we wish to make today.

Basic reminder: ISO response time

Reactivity is supposed to be represented by the "response time" in the monitors’ characteristics. The response time (RT) has been developed by the worldwide normative organization, ISO. This explains that to measure RT you need to add up the rise time (time needed to change a pixel from white to black) and the fall time (i.e. the other way round from black to white).

Two liquid crystal cells.
On the left a white pixel, on the right a black pixel.

We quickly realized that (from the first LCD roundup in french) that this value had little to do with the actual monitor reactivity. For example in 2001, it was a 40 ms monitor that turned out to be the most reactive. It was however compared to 25ms monitors. The reason for that was that we don’t interact with only two tones: black and white, but with color. It is, however, important to emphasize that we see colors but the computer only work in shades of grey. Each panel cell emits a variable light level (256 shades in principle). In front of each cell is a green, blue or red filter. The juxtaposition of all three enables us to see a picture in color (see this other article to understand how the monitors work).

First measurements between greys

The ISO response time measurement follows this principle:
RT = 255 (white) – 0 (black) – 255 (white).
Two problems arise with this principle. If we measure the RT = 255 (white) – 128 (average grey) – 255 (white), the response time is almost always strikingly different. The gap might be from 25 ms for the ISO RT to 100 ms for the white – grey – white measurement. This is why we thought that it would be interesting to measure not only the ISO RT as well as all color changes from white to any other color. We obtained this type of graph:

Response time from white to grey (ms)
This is indeed rather more accurate. With this example, we see clearly that the white to grey response time might be a lot slower that the transition to black.
Unfortunately, this is still not enough because, in certain flattering situations (when the response time measured with this method remains low) it can be highly misleading. Response times from white to grey can also vary greatly from grey to grey.

Bars only represents rise or fall time

For example with this graphic provided by NEC:
RT ISO = (0 - 255) 18 + (255 - 0) 7 = 25 ms,
RT MAX for white to grey = 0 – 192 – 0 = (0 - 192) 38 + (192 - 0) 5 = 43 ms approximately (72 % > RT ISO),
RT MAX all grey levels = 255 – 160 – 255 = (255 - 160) 55 + (160 - 255) 36 = 91 ms approximately (264 % > RT ISO).

It was also our conclusion last September (see this article), without carrying out so many measurements. There is a small difference with the previous graphic: bars don’t only represent rise time but the full RT from one color to another (rise + fall).

Bars represents complete response time (rise + fall)
These graphs show one point: accelerating the ISO RT isn’t enough to improve reactivity. It is imperative to accelerate all color transitions.

Overdrive - 2 : FFD

Since 2001 the Feed Forward gives us hopes

End of 2001, the website ExtremeTech published (in this article) some studies carried out by NEC development team in charge of panel development for TVs. They published their results with a new idea: the Feed Forward or FFD.

The idea is as simple to understand as it seems powerful.
The widest color change is from white (initial position for liquid crystals) to black (final position perpendicular to the light). The voltage applied by the transistor responsible for the cell is, at maximum, let’s say 1 Volts for 20ms.

NEC’s idea is: why not apply 2 Volts during half the time: 2V during 10 ms. According to NEC’s study reported by ExtremeTech, it was a success: the response time from one color to another would be divided by 10.

On the left, the panel without FFD. On the right results with FFD.
Bars represents rise or fall time only

The monitor in the center of this study sees its response time changing from 55 ms approximately (left graph) to 6 ms approximately (right graph). The white - black - white response time is however unchanged. This is normal, since the response time applied is already at its maximum.
Expected in 2002 according to NEC, this technique has never been applied, as far as we know. But that is the same principle used today in the Overdrive.

Overdrive - 3 : overdrive

The FFD becomes the overdrive

Whether this idea comes from NEC or not, the fact is that we have found it again under the name of overdrive. This Overdrive also has different names from one manufacturer to another. ViewSonic, who doesn’t build any panels, named it ClearMotiv and explains it like this:

In blue on the above graph, the liquid crystals normal answer when the voltage is applied, is in black on the graph below. This response is slow and represents a slow moving change materialized on the monitor by trails of light (smudge). On the dotted line above, is the crystals’ targeted response time. If they reacted thus, there wouldn’t be any afterglow.

The overdrive, or ClearMotiv here, consists, as the " Input voltage " graphic shows, in applying an overvoltage during a limited amount of time, so as to motivate the crystals a little bit more, then return to normal voltage once the correct orientation is achieved.

If we consider this graph as realistic and accurate (it is extracted from a whitepaper exemplifying this technology), and if we apply the ISO rule of the +10ù and -10% of the signal, the response is 60% faster in this particular case! In practice, the initial 12 ms response time would be reduced to 5 ms.

New measure

The main interest of the Overdrive would be to greatly increase the color changes between greys. On the other hand, in principle, it doesn’t change anything from black to white and the other way around because the voltage applied to these crystals is already the maximum. The manufacturers face another problem. They have on one side panels that are estimated as clearly faster than in the past and an ISO norm that doesn’t allow them to show these increases.

This is why we have seen for more than a year, the release of panels with a response time including a G2G, or an equivalent (GtoG…). That means that the new measure isn’t "white-black-white" like the ISO norm says, but "grey-grey-grey" (light grey-dark grey-light grey). Manufacturers measure all the response times and communicate the fastest one. It means that a 16ms ISO monitor could be introduced a few weeks later as a 12 ms G2G.

Overdrive - 4 : our doubts

A couple of drawbacks

Even if it looks great on the paper, the overdrive still has a couple of problems. Some clues lead us to ask ourselves if they are not overselling the product.

1rst doubt: the real gain

For example if we only trust measures published by ViewSonic (probably obtained by the panel manufacturer, AU Optronics), the rise and fall times after the crystals’ overtension are:

Bars only represents rise or fall time

It seems very efficient. Some of the rise times have been decreased from 80 ms approximately to 20 ms. But, if you look carefully, it is almost a bit too efficient. Even the white-black-white time is decreased by a couple of tenth of seconds despite the fact that this voltage was already set to the maximum. Therefore, it shouldn’t be possible to amplify it and logically performances could not be improved. ViewSonic justifies this increase and explains that this ClearMotiv technology is an improved version of the Overdrive. We really want to believe them but it does seem quite surprising.

2nd doubt : AU confirms our feeling

During the CeBit 2005, we met one of the AU Optronics panel specialists. He confirmed to us what we have been telling you for years, TN 16 ms, 12 ms and 8 ms panels are absolutely identical. The only difference is in some of the electronics and the overdrive on top. He was not surprised to hear that we haven’t noticed any improvements in practice from one generation to the other. He explained to us that this overdrive was far from regarding all grey color changes. Only some of them, but he didn’t give a figure, are concerned. The final RT between greys would be the best one found among all of those set in place.

Compared to the previous ViewSonic, this graph, published this time by Eizo (who doesn’t manufacture any panel either) seems a little more realistic and appears to meet AU Optronics remarks:



The improvement brought about here by the overdrive is obvious but unequal from one color to the other. Here, not all color transitions are necessarily accelerated. The one from white to grey especially remains more or less steady.

3rd doubt: what is the real influence of response time on the monitor reactivity?

If we follow the theory, the monitor reactivity would be directly connected to the afterglow noticed.
It is interesting to notice that the manufacturers’ position has changed over time.
Initially, a few jokesters told us that it was related to the refreshing frequency, as with the CRT monitors. That explanation didn’t last long.
Then, the ISO response time was given as the explanation for the presence, or not, of blurred effect with motion.
Today, a few manufacturers say that the response time doesn’t mean anything anymore and that it is the grey level that is important. They base this argument on their own measurements, the same that we introduced to you under the shape of the 3D graphs. The problem is that the same manufacturers also admit the following table (also based on ViewSonic’s whitepaper):



According to these measurements (LG-Philips and Nec at least have published similar results) TN panels would be the least "gifted" for color changes between greys. According to this chart, the MVA would be the best, followed by IPS panels. Of course, when an IPS manufacturer speaks of these charts, the order of the two first is reversed.

In fact, that doesn’t matter: in practice our eyes led us to an entirely different conclusion. Whatever the values obtained are, up to today no IPS or VA panel has ever been as reactive as a TN panel announced as 20ms or below. Whether it is a monitor with overdrive or not.

So two possibilities have to be considered:
1 – All measurements are fakes. It would be unfortunate if engineers designing panels aren’t able to correctly measure this response time.
2 – Measures are accurate. In this particular case, it means that the crystals’ response time factor would only be one factor among others for monitor reactivity.
Unfortunately we are unable to give you a lead on what could influence the panel reactivity except for the crystal reactivity. This is an announcement to all electronics enthusiasts, liquid crystals specialist, and panel manufacturers: please share your knowledge!

Contrast ratios are increasing

Contrast ratios are increasing

The contrast ratio measure is fairly easy to establish. You need to measure, on one side the maximum brightness of a white monitor, and on the other, the minimum brightness with the monitor still switched on with only black displayed. These measures are both expressed in candelas and you need to divide one by the other. For example, a monitor with a white at 250 cd/m², and a black at 0.50 cd/m² has a contrast ratio of 250 / 0.50 = 500 : 1.

This example is a typical case and these values are rather widespread. They are perfectly representative of monitors currently on the market.


For the past year, we have seen the release of 1000 :1 monitors. To attain such figures, manufacturers have two possibilities: either they push the white brightness (and the risk is too much brightness for the user) or else they improve the black level (which is better). We have tested both types of improvements. When monitors were based on IPS technology the improvements were only noted on the white. It is the main downside of this technology: the black level isn’t really convincing.
TN monitors have progressed in both areas. Black has been improved and white is even brighter.

The main improvements have been made on VA, or to be accurate PVA panels. White still has a reasonable level close to 250 cd/m² and the black is twice as dark at 0.25 cd/m². That was the case for the Samsung SyncMaster 193P tested last September.

We have now reached 1500 :1

This year it is the same thing: manufacturers announced record contrast ratios of 1500:1. By manufacturers, in fact we mean Samsung and the SyncMaster 750P to come. Knowing that it still features a maximum brightness of 250 cd/m², it therefore means that it could display a black of 0.17 cd/m². Right now, we would be tempted to believe them for two reasons: the first one is that the announcement made for 1000 : 1 was true and the other is that we have already obtained 0.10 cd/m² black for a monitor after calibration. But in this situation, the white is much darker than the 250 cd/m² of the SM 750P.

What is the interest ?

A really dark black should first of all be of interest to graphic designers. I say should, because between theory and reality there is a huge gap. For example, in theory graphic designers are supposed to calibrate their monitors once a week, work in rooms with perfectly controllable lighting and add an eyeshade to the monitor to make sure that the picture display will never be altered. This situation, which should be a standard for all graphic designers, is in fact exceptional. A lot of monitors are not calibrated and have never been calibrated and the lights in the working conditions are the same as everyone else’s: light in the daytime and dark at night time. They work like any other users.

In practice, and this time it is for all users, a deeper black means a better quality for movies. Black stripes above and below the picture won’t have a grey or blue shade and images will stand out more. Also details will be easier to see.



The impact is less important for games. On the contrary, we usually increase the level of brightness to have better vision in dark areas to see if an enemy could be hidden somewhere…

It is however very interesting for picture editing. Colors are more accurate and the picture modifications are more specific.

Viewing angles are increasing

ISO point of view

Viewing angle is part of the area described by the norm ISO 13406-2. This norm has established the rules for LCDs (including dead pixels). The normative text mentions this aspect but it isn’t as clear as the rest of the components. It says that the "image quality must be superior to the value-limits for individual stimuli". The text provides instructions, but admits that for economical reasons, limiting the tests to only six directions wasn’t enough to describe all flat monitor performances. The directions mentioned here are: up, down, etc.
In other words, from our findings, the ISO norm recommends studying the viewing angles from six different point of views but doesn’t give a precise formula.

Manufacturers establish their own rules

As we know, viewing angles are basically given for 4 directions (left, right, above and below) and summed up on only two data: horizontal and vertical angles. The ISO norm has been completely set aside.

In the end, manufacturers use a standard formula that consists in working from the center pixel of the monitor, and seeing from which angle its contrast ratio falls under 10:1. It is the same contrast ratio as the one in the previous page. A 1000 or 1500:1 monitor will feature viewing angles based on 10:1 images. This figure means that we consider as valid an image with a contrast ratio of let’s say 1cd/m² (which is very bright) and the white at 10 cd/m² (it is actually closer to light grey). At 10:1 the image is awful. No one will ever agree to work under these conditions. This value in no way represents genuine usefulness.

We queried the accuracy of this measure and the manufacturers agree with us that it is stupid. But everyone uses it! It offers the opportunity to announce that TN monitors have 140° viewing angles, IPS have 160° and VA have 170°. That was the type of value usually found during the first half 2005. To be more realistic, it would be necessary to establish measurements with a contrast ratio of 100:1. Another problem with this measure is that it only relates to one point in the middle of the monitor. With one eye positioned in a zero angle, compared to the center of the monitor, there is already a difference of a couple of degrees with the edges of the monitor. With a TN monitor over 20° angle for vertical viewing from below, the center of the image will still be OK but the top of the image will start to darken.


What should we consider as good then? That the picture is still OK because the center point is good or that the image isn’t globally satisfying?

Progresses since last September

Since last September each technology seems to have improved. TN monitors have reached 160°, IPS and VA 178°. What kind of miracle happened? Have crystals changed? Is it due to better filters?

Not at all. They have just all reduced the 10 :1 value to 5 :1. We already found the 10:1 measure inappropriate, so you can imagine how we feel about 5:1. To give you an idea: if you put your monitor contrast and brightness ratio to 0 you certainly wouldn’t obtain an image with a white of 5 cd/m². Even with the worst parameters, the image will always be better than the one manufacturers consider as good for viewing angles measures…

Conclusion

Conclusion

Paying a lot of attention to characteristics and test monitors always allows us to spot any flaws. In the beginning, manufacturers told us that the technology was still very young and that it needed some time to have good norms and to correctly regulate this area.



Five years later, the speech is the same. We even see them changing the few allowed measures to adopt others, more to their advantage. Thus, characteristics are more impressive and that is a good thing, but at the same time consumers, including ourselves, have taken a closer look at the problem. This behaviour can only work against their pronouncements.And this is unfortunate because some of them are better than others. There is visible progress, we can’t deny it. Between our first tests and today, we must admit that the panel reactivity has been improved, viewing angles have increased slightly, and color quality is better for all three technologies…

Now would be a good time to take some more realistic measures. ISO says they have been working on a new norm for several years. We are still waiting…But maybe our savior will come from somewhere else. And he had better hurry himself because we seriously need him!

To those who only read the introductory and concluding pages, we are sorry but you will have to read all of the pages to understand what we are talking about. We feel that it is an understandable technical article. A little bit of intellectual work to understand our point of view or to make up your own mind, if it is different from ours, has never hurt anybody.