1 (editado por Eboshidori 18-08-2010 22:38:20)

Tema: Looking for the ultimate gaming screen

[ split from the thread " El PC y la baja resolución." [>] ]


Pegote escribió:

Hey Eboshidori. I'm afraid I didn't make that post :vP and I don't have any of those beautiful monitors. I only linked to that post earlier (actually, to one of its photos) for reference. I even got in contact with that dude to ask him some questions about the monitors. (He bought a few for a video exhibition. 26 of them!) Me, I'm only gathering some info before I buy a monitor similar to the one mentioned before (a couple of composite inputs, and one 9-pin RGB input - no SCART.)

Still, your post is full stuff I should look into, so thanks for commenting!

OK, so, the thing you must remember when considering buying a tri-sync monitor: a broadcast/video wall/graphist monitor don't need video amplification to connect any home device that run at 15 kHz (because the signal is 1 volt for the consoles and the monitor input).
If you plan to connect a few arcade PCBs, it's very easy to add potentiometers (~ 220 ohms) to reduce the voltage amplitude.

If you choose to buy an arcade monitor, you have almost the same sort of picture tube, but you will need to buy a video amplifier (around 100 € for RGB stuffs), or you will need to modify a MGCD (a chinese PS2 to JAMMA adaptator with embedded amplifier, for about 50€).

Other way, if you are lucky, you can find one of those rare TVs that have a VGA input (15-31 kHz, no 24 kHz support of course). For now, I only saw Grundig ones. Other brands (Thomson, Philips...) did some TVs with VGA input, but they were "100 hz" (meaning every 15 Khz signals are modified to be "double interlaced", and when you have the possibility to skip this treatment, you get your 240p source transformed to a standard 480i, which is pretty bad).


Personnaly, I would buy an arcade monitor if I can find a 29" picture tube with a 0.68 pitch, instead of the larger 0.79 common one, which is really fine for 15/24 kHz but a little bit too large to display precise dots beyond 500 pixels per line. When you tend to match the number of pixels with the number of phosphore triads, you loose picture quality, and begin to see moire errors. For every CRT, you need to work bellow the limit of the tube. The pitch is one of those limits.

You can find smaller pitch for large tubes (like some Mistubishi "Megaview" monitor, some early HD CRT with 4/3 format), but in many case, they don't go bellow 31 Khz, and when they take 15 kHz signal, it's upscaled to 31 kHz. Because when you want to increase resolution in CRT technology, you need to have a electron gun which must have thinner cathodes to display a smaller spot... meaning it's no more capable to display a huge 15 khz spot.

For now, 0.68 is the lowest pitch I found for big screens that handle 15 kHz properly.

2

Re: Looking for the ultimate gaming screen

Recap escribió:

What do you use for 31-kHz gaming these days? A tiny 20'' VGA PC monitor? I hope you don't !

I use a Diamondtron (Trinitron licenced to Mistubishi) 22" PC monitor. 20.5" of visible picture (52 cm). An UNMATCHED quality, no other screen can display this resolution with such a precise and pleasant to the eye pixel, no other screen can produce this "scanline feeling", so "video games" like.
There's no way I would play 480p on fixed res flat screen (no matter LCD, plasma, DLP...). Even If I find one of those 4/3 40" 640x480 plasma (to avoid any loss of picture quality and lag due to scaling), I would not trade my little PC monitor. Even when you consider CRTs (the best technology ever for gaming :P ), I wouldn't buy a large 29" tri-sync monitor (from any type), because I know I wouldn't get such an amazing picture (because of the large pitch, and quite unprecise cathodes, focus system and yoke winding). Sure, I would still get a very nice picture (much better than any crap flat screen), but "plain" (= no "scanlines", no little black space between lines) and rather soft, with some moiré and convergence issues (which can be slightly reduced for the few tubes with a slightly smaller picth, and with a precise yoke that you will need to adjust anyway for convergence issues, because they become very annoying when you go to the limits of the tube).

So, for me, the ultimate gaming screen (only one, for every use, meaning every resolutions) doesn't exist.
You need CRT technology of course :D , but you definitely need at least two kinds of CRT:

- one for 15 Khz (and the few 24 Khz stuffs).
Easy, it exists since forever, it's the "standard resolution" tube, that can display the best 240p output in large format, direct view (29" to 40"). No other tubes can display such a large beam, with all the quality we like from the behavior of the beam (visible variations of scanlines size).
The curved Trinitron is your best friend, especially in tate, because it gives you not only the sharpest beam and precise scanlines (with nice variations of size according to colors, and still precise for the smallest spot size), but the cylindrical shape gives you an "Axelay effect" to all your vertical shooters, much more than a spherical tube (and of course a flat CRT ^^').
Playing some masterpiece like Layer Section (with lots of "3D" effects due to paralaxes and multiples sprites with their own speed for big enemies) on a finely tweaked curved Trinitron is really amazing, a quality you had never find in any arcade cab'. It's like you rediscover the game ! ^^

The old analog chassis (before 1994) are the best to provide you accurate and well balanced colors, with the least modifications of video signal, and easy geometric corrections. Digital chassis tend to have too much "enhancement" features (denaturation of the signal), and on flat CRT, it's a pain in the ass to get a good linearity (when you want perfect scrollings, something you especially want in shoot'em ups), and even difficult to adjust colors (because of design choice, every manufacturers wanted to show a model with the blues more blue than blue and reds more red than red, because the digital ships allow it easily).
But you can only find "raw deinterlacing" (as a feature in the TDA-8366 for example)  in digital chassis, and until I managed to achieve a separate circuit to do it properly (or someone else do it ^^), it's the only way to do it with genuine hardware and games.

- then, if you want the best 31 kHz picture, you need CRTs that can go higher than 31 kHz. But because they can go higher, it means they can't go lower than 31 Khz (when they accept 15 kHz input, they upscal it to 31 Khz).

The easiest and cheapest way is to buy a high-end PC monitor, you can buy some for a few peanuts (<- we use this expression a lot in french, for saying "ridiculously inexpensive" ^^' ).
I have 5 big 21-22" monitors, from 1995 (curved Trinitron) to 2002 (flat Diamondtron). They all produce nice pictures, but the last one delivers an amazing display. Each lines have the same characteristics as the ones you see from a regular 29" Trinitron at 240p. It's your 31 kHz stuff with a 240p look, Trinitron quality. Something that I personnaly highly appreciate. :P But OK, it's small...

To find the same quality on a bigger display, you don't have to look over regular tri-sync monitors who are basically "simple TV tubes" (large pitch) with electronics that allow 31 Khz (and think that many large "XGA" CRTs (that are supposed to handle 1024x768) have a large picth too, meaning you can't achieve the same quality than on a smaller PC monitor (fine pitch). No, you need to find one of those early HD TV in USA (around 1999, fine pitch and 4/3 format) or expensive (even in the second-hand market) broadcast/graphist monitors. Almost none of them support 15 kHz. Because they are basically PC monitors of bigger size (and they were damn expensive when they where out, and non available to home market).

An other way is to use a tri-tube projector. You need a dedicate room to operate it if you want a decent picture, and of course you need to know how to set-up the projector. You can also make a rear-projection system with tri-tubes projector (the basic 7" ones can produce a nice 800x600 easily and are cheap nowaday), so you can use it in a daily room, but you have the typical issues of rear-projection: narrow view angles, low contrast. Never as good as a direct view tube.


So, for me, in tri-sync monitors, only the chassis are interesting. Just buy the electronics and adapt it to a good TV tube (you find it for free on the streets, or buy it in repair shops for 10-15 €). But, in arcades, there aren't any Trinitron tubes, so all the chassis are designed to run on the classic shadow mask tubes, either spherical or flat. And the difference of quality between a 240p picture on a Trinitron and on a shadow mask tube will be even stronger for 480p stuff, because a Trinitron is better to display thin spot (but you always have the large pitch issue, a 29" large pitch Trinitron can't produce the same précision as a 22" low pitch). But, hey, it's a decent quality anyway, but from everything I saw until now (in arcades and home cab' of friends), the 480p result is far from my little 22" monitor. Even worse, a 480i non filtred picture on a classic Trinitron is more precise (because the shadow mask dont' have enough space to let the beam strike the phosphore in every situations). That's why I don't want to spend several hundred € (monitor and shipping) to buy a thing that my little 30€ PC monitor will put to shame (and even the old 1995 one, found for free in the street, will produce a better 480p picture :P).


Yeah, it's not easy, you need to know exactly what you want, what is available, and your degree of exigency.

3

Re: Looking for the ultimate gaming screen

So, back to PC at low resolution, with an issue I'm sure absolutely nobody talk about:

Voltage variations of the video signal at the VGA output.

The norm of the VGA signal is fortunatly the same of the home devices: 1 Volt.
The video signal is included between 0.3 V (black level, the screen is supposed to emit no light) and 1 V (peak white). So, you have a "0.7 V" amplitude, from which you will get 256 values to get a 16.8 millions color picture.

0.7/256= 0.002734375. Basically, you have 0.003 V (or 3 mV) for each of the 256 RGB values. Any variations of this range and you loose color accuracy. I won't bother you with shit load of measure tables, but I got lots of different measures not only for different resolutions between 640x480 and 1600x1200, but the same for anything under 640x480, and different level between R,G, and B signal (but when red is higher than the other, it is always higher in every resolutions, for example), and different measure according to genuine drivers, newest ones, and in the Window's  "safe mode" (absolutly no driver and no service or softwares to apply any correction of the display).

On my Ati Radeon 1250, with pure white full screen, no software calibration (raw ouput), it goes from 0.930 V (red signal, 1280x1024, safe mode) to 1.114 V (green signal, 640x480, safe mode). I never got the same results between resolutions, using the same mode (either safe mode or ATI drivers). I tried another card (Radeon 2400 HD, PCI express, which by the way can't go under 7.40 mHz pixel clock, meaning I cant' display anything under 352x256 :[ ) , and I got different results, with almost the same (wide) range around 1 V. Very few combinations of resolutions and driver gave me something close to 1V (what is normallly expected...).
You will tell me that you can modify the graphic card settings via software (in order to calibrate your screen), but it's very limited (it can be OK for a slight correction, say a 0.996 V signal, but when you deal with 1.114 V, it's too much way of). Now, I understand better why it has always been so difficult to maintain a good greyscal and to display proper tones under 5% and above 95% of intensity...
Damn, the graphic cards should send a signal in the 0.7 V range, and no more than 1 V, for every resolutions ! And most of all, the same level for every lines (R,G and B). The only time you would have to tune the levels  should be when you can't tweak anymore your monitor set-up (because of aging cathodes, at different level).
But when you think about the fact that almost nobody takes time to properly calibrate his PC screen (even most of graphist users), nor tweak the levels via software, plus the fact that back then almost everybody used the "brightness" setting in a wrong way (to get greyish black, not an actual brighter picture...), plus the too much high color temperatures... it's hopeless. I never read any advertising nor review of graphic cards saying : " it's magic ! Perfect video signal amplitude at any resolution !"

I wanted to use the PC to display a precise calibration mire (thanx to soft15kHz) to finely tune my TV screen (because the connection between G2, cut-off and gains is crucial to display the best scanlines, the sharpest spot), but as always, it's not "plug and play", you have to burden yourself with problems to solve. Damn, whoever create the PC is a devil ! ^^'

4 (editado por Eboshidori 18-08-2010 22:40:11)

Re: Looking for the ultimate gaming screen

This is the last reply of Pegote, but when I moved the messages here, I forgot that he doesn't  have access to this place. If he shows his email adress in his profil (after reading my last reply on the other topic), I will send him the explanation :


Pegote escribió:
Eboshidori escribió:

on flat CRT, it's a pain in the ass to get a good linearity

Maybe I'm only taking this particular line completely out of context, but: do you mean, for example, that vertical lines on a flat (flattened?) CRT can never be parallel?Like, is this a known issue that is unavoidable with this type of screen?

No, on direct view tubes, you always have parallel lines. The only tubes that can display a trapezoid image (with on side bigger than the other) are those in tri-tubes projector /rear projection sets (because it is requiered to converge the red and blue picture to the green one. But after a good set-up, from your point of view, you should see horizontal, parallel lines on the final picture).

The geometric corrections on direct view tube allow you to change the shape of the picture, but you always have parallel lines (you can have a trapezoïd picture, but with the top lines having a different size of the bottom ones, and still always parallel).

The linearity (in this case: the horizontal one) concern the variation of the pixels size from left to center, and center to right. To display a regular picture on any tube, you need to add some variations to the signal (wave forms), because the course of the beam, which should describe a perfect circle, never matches the shape of the tube. When you have a spherical (or cylindrical) tube, it's a different radius, but it's relatively easy to adapt the radius of the beam closely to the radius of the tube. When you have a flat tube, you need to change a radius into a straight line. And it's never perfect. It involves not only wave forms but also a really precise driving of the high voltages in the THT, something you can't find in large TV set for home market (because accurate electronic for such high voltages cost lots of money, and home TVs are not supposed to display graphics with straight stuffs, it's movies, broadcast shows etc.).
The linearity issue is really annoying for 2D stuffs, especially when you are a shmup addict like me, because the slow,regular, continuous scrolling always shows the defect. That's why I prefer spherical tubes over flat ones.
On spherical tubes it's not perfect too, but is much less visible, and easy to correct to a decent level.



I have to ask, because a while ago I spent some time with my 29" flat CRT TV trying to calibrate it to have a "straight" image and the best I could get (after a lot of swearing) was an image that kinda looked like this: ))|((

(Mind you, my TV is some locally assembled, chinese manufactured mammoth. As much as I'm pretty ignorant about all of this, maybe it's just the TV that's shit.)

A picture tube displays whatever you send to it. But if you want a good picture, with good geometry, you need electronic circuits that allows you to finely tune the most parameters possible, you need something well designed. And on cheap chinese TV sets, you can barely find something good enough to correct your signal. But on those TVs, you can find good picture tubes, because only big and longtime manufacturers make tubes (RCA, Thomson, Philips, Sony, Panasonic, Samsung, Nec...in fact, they are very few). Most time, a small compagny buy tubes to big manufacturers, then they design their own eletcronics to drive the tube or simply buy
cheap chassis made by brands that don't make their own CRT too, they put all together in a plastic case, they put a sticker in the front, and here you go !

The quality of a monitor relies heavily on the chassis (the electronics) that drive the tube, and on the proper set-up that needs to be done by trained technicians (or crazy old-school video game addicts :P ), because it's very difficult and involves lots of constraints and compromises. And much more, it requiered time to do it. Time is money, so if you want to sell cheap monitors/TVs, you choose an easy tube to drive (a spherical one), you put a cheap chassis, and you don't spend any time to properly tune the picture at the end of the assembly line.


For you your picture problem (with a  ))|((  shape), it's a typical pin-cushion issue (it's different from linearity). Normaly, even on flat CRT, you should be able to correct it easily, via the "service menu", in which you can enter by taping a special code on your remote (you need to know the reference of the chassis and find it on the web, on http://www.eserviceinfo.com/ for example).

5 (editado por Calamity 24-08-2010 19:51:34)

Re: Looking for the ultimate gaming screen

Hi Eboshidori,

I highly appreciate your technical posts on CRTs. I am currently working on a software to get the best possible results from ATI videocards on CRTs by extending the regular driver capabilities, by means of a simple hack that allows more resolutions to be used at the same time (regular driver only has space for 60 modes). However, the main part of this soft is the little program that calculates the required modelines depending on some initialization values. You may have seen the project in this thread:

http://postback.geedorah.com/foros/view … 23&p=3

While the modelines produced by the program have been succesfully tested on my arcade monitor (Hantarex 9110) and Recap's Trinitron set, I still have to find a proper explanation for the behavior observed by Daicon-X on his tv, a Bluesky 21". In order to avoid you the translation of the whole thread, I would say that this chassis has a "non-linear" behavoir, with seemingly random results from subtle changes in the input modeline, specially around 57 Hz. The most obvious problem appears as a block of lines displaced to the left on the bottom of the picture.

I lack the electronics knowledge to give an explanation. I've read that modern CRTs use chips for signal processing. How could these chips affect the results? Is it possible that these chips are hardcoded to filter PAL/NTSC signals and do not know exactly what to do with non-standard timings? Or should all chassis accept via SCART any proper RGB signal provided that the timings are good, whatever the refresh is? (and obviously, using an horizontal frequency within the monitor range).

6

Re: Looking for the ultimate gaming screen

Calamity, that's good you are here, I just would like to ask you few questions about drivers.

I am currently working on a software to get the best possible results from ATI videocards on CRTs by extending the regular driver capabilities

So, you made a hack to allow more different resolutions at a time. But do you know why all different resolutions can produce different voltage levels at the VGA output ? It's not a hardware question (few resistor or amps) , because values change according to different drivers and safe mode... For now, I'm quite clueless because I just noticed it recently, in the past few years I would never had the idea of mesuring the voltage out of the VGA port.
Because you know, wide range of resolutions is good, but signal accuracy is really crucial to get the best result possible on CRT. You will say that most video games have a very limited on-screen color palette (and that's true), but to properly calibrate a CRT, you need to have a signal between 0.3 and 1.0 V .
When you display 15 kHz picture at 321x240 for example, you can't change the values of gamma/brightness/contrast for each channel (red green and blue). You can do it at 480i, but anyway, you can get slighty variations between 480i and 240p ...

The ultimate driver would be the one that allows many different resolutions and a constant voltage signal (video part always between 0.3 and 1.0 V), or at least, a possibility to quickly adjust parameters in any resolution (like the gui of Zsnes, which suits every resolution, unlike many other emulators designed for 640x480).

Since PC users had never ever complain about voltage levels (who the fuck cares ? Even those who wanted accurate colors just tweaked the drivers options, whitout mesuring the voltages), I think this issue had been there for almost the beginning. I need to check other graphic cards, and different motherboards to confirm if there are some set-up that delivers more variations or more regular level among resolutions. (if you guys could do it with your graphic card, even with a basic 10€ multimeters, that would help !^^' )




I still have to find a proper explanation for the behavior observed by Daicon-X on his tv, a Bluesky 21". In order to avoid you the translation of the whole thread, I would say that this chassis has a "non-linear" behavoir, with seemingly random results from subtle changes in the input modeline, specially around 57 Hz. The most obvious problem appears as a block of lines displaced to the left on the bottom of the picture.

Bluesky is a brand of cheap TVs. While they have some good Philips/Thomson tubes in it, they got some crappy Vestel chassis ("11AKxxx") that break easily and are not well designed. And of course they come with a poor set-up.

In general way, digital electronics tend to be much simpler to design, they have a more regular behavior than analog one. So, on older analog chassis, you often find very wide tolerances on timings, porshs, frequencies. In digital ones, especially because they are designed according to very few broadcast signals (and not video games :P ), you tend to have very few room for your non-standard signals.

If you have troubles, it's better to display a non standard picture in a standard frame ( for example, a 304x224 picture in a 320x240 frame), and choose refresh rates close to 50hz and 60hz (meaning that you won't be able to have perfect emulation for 53hz and 57hz games, because they are too much out of range, generaly a 2 Hz only one).

In other way, you need to try several pixel clocks and porchs to avoid picture deformations. Some chassis won't show you any picture at all if the signal is too much "way off" (but the same signal can be send into a much "friendly" analog chassis...).

So, as usual, older means better, you will enjoy the wide variety of the 15 kHz picture on spherical analog TV sets. If you choose a TV from 1994 to 1996, you can still find "hybrid" chassis that have both digital (geometry correction) and analog parts (especially for the video amplifier on the neck-board).

Recap's Trinitron may be the ultimate TV, because it's a cylindrical tube with an early digital chassis that allows you to easily deinterlace 480i and still has lots of analog components, and it takes lot of resolutions around 15 kHz. Ok, it's a pain in the ass to properly converge it ^^, but I got many skills by working on it (in fact, this TV set had been my "conejillo de india" :P to test everything about CRT convergence and geometry).

7

Re: Looking for the ultimate gaming screen

Eboshidori escribió:

So, you made a hack to allow more different resolutions at a time. But do you know why all different resolutions can produce different voltage levels at the VGA output ?

I don't know if this has to do with voltages at all, but I do have noticed a difference in intensity of colours depending on the resolution, specifically on the horizontal frequency of the video mode. I had associated it to the fact that, as you increase the horizontal frequency, the width of the scan lines gets reduced (a single line at 15.625 KHz is 64 µs long, the same line at 16.200 KHz becomes 62 µs), so the video gets more concentrated in the middle of the screen, burning the colors. Even if the constrast had been adjusted for video modes around 15 KHz, you could see how the colors faded to the right with a blue shadow as you change to video modes with higher horizontal frequency.

As I said, I can't figure if this could have any relationship with voltages, but reading your post I remembered this curious effect. It's also true that horizontal frequency for a given resolution may vary from a driver to another, as the timings will probably be different. It would be good to check if voltage differences have to do with horizontal frequency variations or any other feature of the modeline.

Eboshidori escribió:

In general way, digital electronics tend to be much simpler to design, they have a more regular behavior than analog one. So, on older analog chassis, you often find very wide tolerances on timings, porshs, frequencies. In digital ones, especially because they are designed according to very few broadcast signals (and not video games :P ), you tend to have very few room for your non-standard signals.

So, you are confirming my fear that these modern chassis may not be flexible at all for proper emulation. However, I will try to find a workaround for these users, problably including a manual set up of modelines.

8

Re: Looking for the ultimate gaming screen

Eboshidori escribió:

Recap's Trinitron may be the ultimate TV, because it's a cylindrical tube with an early digital chassis that allows you to easily deinterlace 480i and still has lots of analog components, and it takes lot of resolutions around 15 kHz.

And you forgot the best feature of they all -- it's incredibly easy to rotate for a vertical-orientation display. Not only the sides are flat and speaker-free, the corners are perfectly rounded to serve as the pivot, hence saving you from low back pain and whatnot.

9

Re: Looking for the ultimate gaming screen

the width of the scan lines gets reduced (a single line at 15.625 KHz is 64 µs long, the same line at 16.200 KHz becomes 62 µs)

It's not a question of width (space) but time. The old name for chassis is "time base", because all you do is drive the electron gun in time. There is no space data, unlike flat fixed res' screens. That's one of the major difference between CRT and every other display technologies. That's why we talk about frequencies for CRT, because "resolution" isn't enough to clearly describe the capacity of a tube.

So, as the scan frequency increase, the line duration decrease. You also have a slight difference between PAL (64µs) and NTSC (63,55 µs). Every dynamique geometric distortion is a matter of delay added to the signal.

so the video gets more concentrated in the middle of the screen, burning the colors.Even if the constrast had been adjusted for video modes around 15 KHz, you could see how the colors faded to the right with a blue shadow as you change to video modes with higher horizontal frequency.

When I first tried to create my own modelines (especially to display very low resolution at the minimal pixel clock allowed by my card, for example 256x224 at 6.5 mHz ) , I made lot of tests. I noticed a fading for very "out-standard" resolutions, but from top to bottom, not from left to right, and no color tinted. Some resolutions were stable and readable, but only the first few lines were bright, all the other were darker and darker.


It would be good to check if voltage differences have to do with horizontal frequency variations or any other feature of the modeline

From what I saw, you have different results with the same frequency (same resolution), depending on the driver (or lack of it). But yeah, maybe different drivers use different pixel clocks and porshs to display the same thing, so at the end of the digital>analog conversion on the graphic card, you can have various voltages. Remember that this conversion is always a weak point in many electronics (except high-end products). For now I don't really know much about the subject.


To Recap: yes, the screen is already good, and so the plastic case ! ^^

10 (editado por Calamity 30-08-2010 13:02:22)

Re: Looking for the ultimate gaming screen

Eboshidori escribió:

It's not a question of width (space) but time. The old name for chassis is "time base", because all you do is drive the electron gun in time. There is no space data, unlike flat fixed res' screens. That's one of the major difference between CRT and every other display technologies. That's why we talk about frequencies for CRT, because "resolution" isn't enough to clearly describe the capacity of a tube.

So, as the scan frequency increase, the line duration decrease. You also have a slight difference between PAL (64µs) and NTSC (63,55 µs). Every dynamique geometric distortion is a matter of delay added to the signal.

Although, as you state, all the information to build a modeline is of time nature (that's what we talk about "timings" of a video mode), being the speed of the electron beam a constant value, the shorter (in time) the scanline is, the narrower (in space), the resulting picture will be (in absence of a compensating mechanism). I am positive of this fact as I observe it all the time, and has lead me to look for a way to compensate it.

To explain it with an example: if you adjust an arcade monitor so that a resolution of 320x240@60Hz (15.625 kHz) fits perfectly within the screen frame, then if you keep the adjustment and show a resolution of 320x256@60Hz (16.700 kHz) you will have big borders on the sides of the picture (in fact, as we also keep the vertical refresh rate, the time for the extra lines of vertical resolution have to come from somewhere -- from those borders!).

However, I'm not saying that this has to do with voltage differences, it was just a suggestion for searching.

Eboshidori escribió:

I noticed a fading for very "out-standard" resolutions, but from top to bottom, not from left to right, and no color tinted. Some resolutions were stable and readable, but only the first few lines were bright, all the other were darker and darker.

I think I have seen that effect also, but the one I mean is different, constant on the whole picture, as when te contrast potenciometer is too high.

Eboshidori escribió:

From what I saw, you have different results with the same frequency (same resolution), depending on the driver (or lack of it). But yeah, maybe different drivers use different pixel clocks and porshs to display the same thing, so at the end of the digital>analog conversion on the graphic card, you can have various voltages. Remember that this conversion is always a weak point in many electronics (except high-end products). For now I don't really know much about the subject.

When you tested the same resolution in different drivers I assume your samples also had the same vertical refresh. If finally it turns out to be a 'feature' of the digital-analog conversion, then there is not much we can do...