Is OLED's response time fast enough to enable CRT electron beam emulation? - AVS Forum
Forum Jump: 
 
Thread Tools
Old 01-18-2012, 06:29 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I have two CRT's at home, a NEC XM29 and a SONY PVM50. Their purpose is to act as displays for older generation gaming hardware that simply just look fantastic on these display technologies because they were meant to be viewed on them with all of their quirks aiding their visuals as assets.

I am not talking about black levels or colour reproduction either, but things like the scanlines and other inherent quirks of the CRT display technology. Most 32 bit era and below gaming hardware output a 240p signal that produces visible black spaces between each scanline which really helps with the low resolution visuals.

Now, obviously such quirks of an electron beam meticulously tracing a frame line by line in real time on a phosphor surface are not inherent to newer display technologies thus older games just don't look right when viewed on a LCD/Plasma/OLED display.

There are niche hardware solutions however which come between your game unit and display and which emulate the black spacing between scanlines by inserting grey vertical lines into a frame before passing it to your TV.

Alas this is still a far cry from the genuine CRT look and only a half measure. So it is that whenever I play anything on my CRT's I cannot help but think that the authentic CRT visuals will die whenever my CRT's give up the ghost because they belong to obsolete technology no longer being produced.

But reading about the advantages of OLED technology got me thinking. If OLED has fast enough response time and obviously given that most OLED displays sport HD resolutions then perhaps a more accurate emulation of CRTs could be possible.

Imagine that a OLED is fed a 240p frame or even a 480i half frame field which it is supposed to display for 1/60th of a second.

Now instead of doing what all progressive displays do, which is just display that full frame for the full duration instead it could emulate an electron beam drawing those 240 lines one scanline at a time complete with dimming of the previously drawn lines as the scanline scanning is moving further ahead as well as the glow of a scanline making each visible line thicker than the spacing between them by aid of the higher display resolution.

I am not sure wither OLED technology truly has fast enough response time to be able to do this. But if it is then it should be possible to create a true alternative to CRT's using this technique for posterity and preservation of the genuine look of vintage games.

The faster the true response time of OLED elements are the more accurate the tracing of an electron beam can be emulated.

The ideal would be for a display to be able to split a 240p frame into 240 individual frames which are then drawn in rapid succession within 1/60th of a second.

But that is asking for a bit too much I think. Perhaps the CRT look can even be achieved by drawing 2 or 4 scanlines at a time to reduce the requirements. But is that even feasible?

Bear in mind this would have to be something achieved by the displays internal panel drivers as long as it is fed a 240p 60fps signal. (Or a 480i 60fps signal.)
-=Kamikaze=- is offline  
Sponsored Links
Advertisement
 
Old 01-18-2012, 08:22 AM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by -=Kamikaze=- View Post

Now instead of doing what all progressive displays do, which is just display that full frame for the full duration instead

Are you talking about duty cycle or activation sequence? LCD scans and displays an image row by row but with a 100% duty cycle. Some PDPs fire all pixels at once while some examples fire groups of pixel rows sequentially and the effective duty cycle is well below 100%.

SED/FED prototypes were passive matrix line by line scanning devices with short duty cycles IIRC.

Quote:
Originally Posted by -=Kamikaze=- View Post

I am not sure wither OLED technology truly has fast enough response time to be able to do this. But if it is then it should be possible to create a true alternative to CRT's using this technique for posterity and preservation of the genuine look of vintage games.

If you measure response time via the rise time and fall time of light then OLED by far has the fastest response time of any emissive display ever AFAIK. Much faster than CRT.

Quote:
Originally Posted by -=Kamikaze=- View Post

The faster the true response time of OLED elements are the more accurate the tracing of an electron beam can be emulated.

Given the intrinsic fast response of OLED materials the active scan line would require very high current to produce enough light in such a short period of time. This would burn out the OLED material very fast AFAIK.

I've seen a few papers on AMOLED driving schemes. They range from full 100% duty cycle (like LCD), to full frame flash like PDP, to line by line scanning and emitting like CRT. The circuits for some of these pixels contain 3+ TFTs. The main TFT is the master switch opened by the gate electrode and the other TFTs control the on and off current IIRC.

Over thinking, over analyzing separates the body from the mind
xrox is offline  
Old 01-18-2012, 12:44 PM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
The terminology that you use goes way above my head unfortunately so let me explain how CRT emulation would work in a different manner.

Say I have a frame consisting of 240 lines that I want to display it in 1/60th of a second in a way that would emulate the way a CRT would display it.

The algorithm I use can take that single 240p frame and put out 240 unique 1080p frames that simulate how a CRT would display it in that 1/60th of a second. Those 240 frames would be like as if you put a very high speed camera in front of a CRT and recorded footage of a CRT displaying a screen and you recorded it at such a speed that yields 240 unique recorded frames for every 1/60th of a second or 14400 frames per second.

Now if you had a display other than CRT upon which you wished to emulate the CRT look you would need to display these 240 frames in rapid succession in 1/60th of a second. If possible by just switching from one frame to the other instantaneously.

So for 1/60/240th of a second you would show frame 1, then the next 1/60/240th of a second you would show frame 2.

Obviously that is quite unfeasible as then forget about 120hz your panel would need to be 14400hz to be able to show all unique frames. So the real question becomes what is the highest hz that an OLED screen can reach?
-=Kamikaze=- is offline  
Old 01-18-2012, 01:31 PM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by -=Kamikaze=- View Post

Say I have a frame consisting of 240 lines that I want to display it in 1/60th of a second in a way that would emulate the way a CRT would display it.

This would already be possible using existing OLED driving technology IMO.

Quote:
Originally Posted by -=Kamikaze=- View Post

The algorithm I use can take that single 240p frame and put out 240 unique 1080p frames that simulate how a CRT would display it in that 1/60th of a second. Those 240 frames would be like as if you put a very high speed camera in front of a CRT and recorded footage of a CRT displaying a screen and you recorded it at such a speed that yields 240 unique recorded frames for every 1/60th of a second or 14400 frames per second.

Now if you had a display other than CRT upon which you wished to emulate the CRT look you would need to display these 240 frames in rapid succession in 1/60th of a second. If possible by just switching from one frame to the other instantaneously.

Ok, I understand.

Quote:
Originally Posted by -=Kamikaze=- View Post

Obviously that is quite unfeasible as then forget about 120hz your panel would need to be 14400hz to be able to show all unique frames. So the real question becomes what is the highest hz that an OLED screen can reach?

If you took the camera you suggest and filmed an actual CRT in the manner you suggest then each unique 1/240 frame would show a 69 microsecond snapshot of the CRT scan. This will show a group of pixel rows alight and the rest of the TV area black.

I don't currently see any advantage of this proposed technique compared to producing a driving circuit that directly scans the OLED similar to a CRT? What do you think the advantage would be?

Over thinking, over analyzing separates the body from the mind
xrox is offline  
Old 01-18-2012, 05:26 PM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I think if you look at the frames from the camera you would see that the phosphors of a lit scan line hold their glow for some time before gradually dimming as the scan lines furthest away are being scanned which gives CRT its smooth aesthetic. If you just scanned a line across OLED there would be no after glow dimming of scanlines which would prolly look strange.

The ultimate purpose here would be to reproduce that unique feel of a CRT display as opposed to newer display technologies. The ultimate goal would be to preserve an authentic experience of playing vintage games as they were supposed to be seen for posterity.

This is a enthusiast site where people fuss over the tiniest iota of visual detail, surely it should be easy to understand how some gamers are very partial to the look and feel of older games on CRT's, especially Arcade CRTs and would be interested in preserving that somehow in a CRT less future.

So far inserting black lines into a frame has been about as far as you could go for CRT emulation on LCDs. But with super fast response time displays the behaviour of phosphor glow could be reproduced and 2K-4K resolutions could allow accurate reproduction of other details such as colour element locations on CRT aperture grills, etc.

I wish there was a market for such a thing as then I wouldn't worry as much about my CRT's one day dying.
-=Kamikaze=- is offline  
Old 01-19-2012, 03:48 AM
ID4
Member
 
ID4's Avatar
 
Join Date: Feb 2011
Posts: 22
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Hi!

I understand what you want to say, You know about FED?

FED spindt is the only technology that actually replace the CRT, since it is really a new type of CRT.

The most interesting addition is that it can support different native resolutions as to the CRT, without scaling. This is possible thanks to the nanotubes, more information:

http://forums.anandtech.com/showthre...oto=nextnewest

Here is the patent for this technology that allows multisync CRT as well in the same way.

http://www.google.com/patents/about?...AJ&output=text

ID4 is offline  
Old 01-19-2012, 07:52 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
FED is old news, it is one of those technologies that has been sitting on the periphery of the ethereal for a great deal many years. Same with SED, as well as with OLED until just recently when suddenly from among many ever looming specters OLED leaped in to the realm of the tangible this CES with promises of 55" launches coming this very year, and not just by one manufacturer.

But leaving that be, even if FED is technically closer to CRT the problem of emulating the look of older console games is more of an issue of software, or rather the software/hardware that runs a panel.

When 55" OLEDs become available later this year and even if FED eventually makes it to the market some time in the future that still wont bring CRT emulation for vintage 480i/240p sources any closer to reality.

That would require specialized software/hardware support developed for this purpose by the manufacturers which will never happen. Hell, most newer panels dont even come with s-video or even composite inputs anymore, so even rudimentary support for vintage sources is a distant concern for panel manufacturers.

Anyhow, only way this could be possible is if there is an OLED TV which is fast enough to display 240 or more unique frames for each 1/60th of a second and enthusiasts somehow manage to hack the OS of the TV, reverse engineer the software that drives the panel and write custom software to enable a special vintage CRT mode.

Same for FED or any other technology really. Alas I doubt LCD and Plasma are up to snuff in terms of technology but it would be exciting if OLED is.

Oh, you would of course also need a very fast CPU inside the TV that would enable processing fast enough to generate 240 frames for each single frame from a vintage source.

But the algorithm I imagine for recreating the electron beam tracing is so simple that it could prolly be done with even modest CPU power but who knows.
-=Kamikaze=- is offline  
Old 01-19-2012, 08:04 AM
ID4
Member
 
ID4's Avatar
 
Join Date: Feb 2011
Posts: 22
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by -=Kamikaze=- View Post

FED is old news, it is one of those technologies that has been sitting on the periphery of the ethereal for a great deal many years. Same with SED, as well as with OLED until just recently when suddenly from among many ever looming specters OLED leaped in to the realm of the tangible this CES with promises of 55" launches coming this very year, and not just by one manufacturer.

But leaving that be, even if FED is technically closer to CRT the problem of emulating the look of older console games is more of an issue of software, or rather the software/hardware that runs a panel.

When 55" OLEDs become available later this year and even if FED eventually makes it to the market some time in the future that still wont bring CRT emulation for vintage 480i/240p sources any closer to reality.

That would require specialized software/hardware support developed for this purpose by the manufacturers which will never happen. Hell, most newer panels dont even come with s-video or even composite inputs anymore, so even rudimentary support for vintage sources is a distant concern for panel manufacturers.

Anyhow, only way this could be possible is if there is an OLED TV which is fast enough to display 240 or more unique frames for each 1/60th of a second and enthusiasts somehow manage to hack the OS of the TV, reverse engineer the software that drives the panel and write custom software to enable a special vintage CRT mode.

Same for FED or any other technology really. Alas I doubt LCD and Plasma are up to snuff in terms of technology but it would be exciting if OLED is.

Oh, you would of course also need a very fast CPU inside the TV that would enable processing fast enough to generate 240 frames for each single frame from a vintage source.

But the algorithm I imagine for recreating the electron beam tracing is so simple that it could prolly be done with even modest CPU power but who knows.

As I said, the spindt type of FED will alow multiple native resolutions just like the CRT does, so no need for emulation or tricks.
ID4 is offline  
Old 01-19-2012, 08:10 AM
AVS Special Member
 
specuvestor's Avatar
 
Join Date: Jun 2010
Posts: 2,399
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 13
1) OLED is not suddenly. You can read them up on the OLED thread.We talked about 55" for a few months

2) AUO owns FED and there are no plans for it.
specuvestor is offline  
Old 01-19-2012, 08:14 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
But all newer displays are driven by digital processing systems which receive the signal, decode it, process it and then draw it on the panel.

Most likely if you passed a 240p and 480i signal to any panel made in the modern TV era then you would get a deinterlaced 480p frame upscaled to the panels native resolution or some set resolution.

It would be amazing if a FED display came out that would sync to a analoug signal and display without any digital processing, but I doubt anyone would manufacture such a thing. I mean, why would they? Think about it, then people connecting old DVD players to such TVs would complain about flicker and whatnot.

Anywaste, for a demonstration of what a CRT looks like in slow motion I found the following youtube video:



Good old youtube, it has everything.

It think it is safe to say that no modern panel displays content in this manner, nor any will ever will again.

Quote:
Originally Posted by specuvestor View Post

1) OLED is not suddenly. You can read them up on the OLED thread.We talked about 55" for a few months

I had been casually following the OLED advancement thread on and off for years while anticipating OLED TVs. When SONY seemingly pulled the plug and OLED was a no show for, CES 2010 I think, I gave the technology up for lost. So seeing reports of 55" at this CES gave me a start. Happy times anyhow.
-=Kamikaze=- is offline  
Old 01-19-2012, 08:36 AM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by -=Kamikaze=- View Post

I think if you look at the frames from the camera you would see that the phosphors of a lit scan line hold their glow for some time before gradually dimming as the scan lines furthest away are being scanned which gives CRT its smooth aesthetic. If you just scanned a line across OLED there would be no after glow dimming of scanlines which would prolly look strange.

The afterglow is very minimal. 1ms or less. I don't see it making any difference in PQ compared to your proposed method.

Quote:
Originally Posted by -=Kamikaze=- View Post

This is a enthusiast site where people fuss over the tiniest iota of visual detail, surely it should be easy to understand how some gamers are very partial to the look and feel of older games on CRT's, especially Arcade CRTs and would be interested in preserving that somehow in a CRT less future.

Why CRT looks the way it does is an excellent topic. I'm not sure I agree that phosphor decay is the reason why it looks that way. If a CRT was built with 1 nanosecond phosphors it would still look just like a CRT

Quote:
Originally Posted by -=Kamikaze=- View Post

So far inserting black lines into a frame has been about as far as you could go for CRT emulation on LCDs. But with super fast response time displays the behaviour of phosphor glow could be reproduced and 2K-4K resolutions could allow accurate reproduction of other details such as colour element locations on CRT aperture grills, etc.

I don't think addressing could ever make an LCD look like a CRT. You seem to think it could?

Over thinking, over analyzing separates the body from the mind
xrox is offline  
Old 01-19-2012, 08:41 AM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by specuvestor View Post

1) OLED is not suddenly. You can read them up on the OLED thread.We talked about 55" for a few months

It is possible to make OLED an impulse. Some widely varying circuit designs in the literature.

Over thinking, over analyzing separates the body from the mind
xrox is offline  
Old 01-19-2012, 08:51 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by xrox View Post

The afterglow is very minimal. 1ms or less. I don't see it making any difference in PQ compared to your proposed method.

What is your proposed method again? just one lit line per each 240 frames? I guess maybe it wouldn't make a difference, I really couldn't say. But let me put it this way, I sure would like to know one day what either looks like on an OLED. Especially if I have a CRT right next to it displaying the same footage.

Quote:
Originally Posted by xrox View Post

Why CRT looks the way it does is an excellent topic. I'm not sure I agree that phosphor decay is the reason why it looks that way. If a CRT was built with 1 nanosecond phosphors it would still look just like a CRT

Well, it is true that there is nothing like the genuine article, but unfortunately every year there are fewer CRTs in the world and one day they will be damn near impossible to come by.

I bought both my sets used and while I love my XM29 I do get a feeling that if I had got one with fewer hours on it then the colours would pop more. Of course I was lucky to get one at all.

Quote:
Originally Posted by xrox View Post

I don't think addressing could ever make an LCD look like a CRT. You seem to think it could?

Heavens no! Forget about LCD, that is not at all suitable for that purpose. I was talking about OLED and future display technologies.
-=Kamikaze=- is offline  
Old 01-19-2012, 08:54 AM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,644
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 260
If you aren't too concerned about having the original hardware hooked up to the TV, there are a number of different projects focused on emulating old games with a CRT-like filter.

As far as gaming is concerned, I think the scanning display of a CRT and phosphor decay are the least important aspects when it comes to that "CRT look" it's more to do with the fact that there's no "true" native resolution with CRT (even though there is still a physical resolution from the aperture grille/shadowmask) and the fact that you have a variable spot size depending on how bright a pixel is supposed to be.

In my opinion, we just don't have enough resolution with 1080p displays to properly emulate this, but 4K seems like it should be enough. Fake scanlines are either too thick, or not dark enough at 1080p.

http://board.byuu.org/viewtopic.php?f=10&t=147

The other problem is that it seems a lot of people are trying to emulate a cheap portable CRT hooked up with a composite cable, rather than a high-end CRT hooked up via RGB. (every console since the NES has been capable of an RGB output) Phosphor3x is the first one that has even come close to a real CRT in my opinion.

EDIT: A comparison which shows why I think we don't yet have enough resolution. (not mine, from that forum)

Phosphor3x filtered, then scaled up:


Scaled up first, then the Phosphor3x filter applied on top:


While the former is probably more technically accurate (looks like a CRT viewed up close) the latter is more representative of the actual viewing experience with a high-end CRT in my opinion, where the shadow mask has far less impact on the final image.
Chronoptimist is offline  
Old 01-19-2012, 09:14 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I've tried a few filters here and there. So far I've seen nothing that came close to the real thing. The filter in the link you provided sure is bizarre, that is not what 240p content looks like on any of my CRT's.

Then again I have seen a lot of different CRT TVs and the look of each is unique depending on factors such as wither they use shadow masks or aperture grills, and how their electron gun behaves when fed a 240p signal.

Some CRT's I've seen don't even have dark lines between raster lines for 240p, but I feel the ones that do produce a more visually appealing image.

As for some people's pathetic attempt at not so much emulating older display techs as much as parodying them with excessive blurring and what not. I doubt these persons have even seen a CRT in years.

I grew up in Europe and would attempt to use SCART RGB to connect my consoles to a TV whenever possible, or at the very least use S-video. That is the look I want to preserve.

But I do faith that raster trace emulation could perhaps manage to reproduce a lot of the aesthetics of a CRT TV. At the very least you'd get that subtle pulsing flicker, which is much more pronounced on Arcade cabinets and which I love.
-=Kamikaze=- is offline  
Old 01-19-2012, 10:59 AM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,644
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 260
Well most of the examples given are from newer games which aren't 240p in resolution, that's mostly NES-era hardware.

The Phosphor3x filter literally emulates the aperture grille and spot size variation based on the brightness of a pixel. It's far from perfect (vertical gaps are far too prominent) but it's the closest thing to a CRT I've seen so far. You will never have a perfect CRT emulation, especially not when it comes to emulating the phosphor behaviour and scanning nature of the display. (decay, flicker etc.)

Most of the examples there were not including scanline emulation, as we don't have high enough resolution displays for that yet. (or this, eventhe aperture grille is too prominent)

What type of CRT are you trying to emulate though? I'm guessing not a high resolution aperture grille one:





And even when you're trying to emulate an aperture grille CRT, most did not have perfect alignment:


These are all problems that need more resolution to properly solve.

If you actually look at the full resolution of that "e" look at how much detail there actually is that has to be emulated. It's not just even RGB stripes, they're brighter in the centre than the edges. The full "pixel" isn't illuminated with darker shades. In many cases, it isn't even rectangular any more. Red/green/blue "subpixels" don't even have proper vertical alignment or geometry.

It's not an easy thing to do, and you're not going to have many people interested in such a thing. The work that the emulation community is doing, is likely going to be the best we'll see. (the only other thing I can think of would be scanning laser projectors, but I wouldn't hold out much hope)
Chronoptimist is offline  
Old 01-19-2012, 11:27 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Chronoptimist View Post

Well most of the examples given are from newer games which aren't 240p in resolution, that's mostly NES-era hardware.

That is incorrect. Even 32bit systems such as the PSX, N64 and Sega SATURN put out 240p signals. There was a single generation where 480i was the standard with the occasional game supporting 480p and that was the PS2/GameCube generation. That generation overwhelmingly consisted of games using 3D graphics and thus looks pretty decent upscaled on a modern flat panel display.

The screens you show me lack the distinctive gap between each raster trace line.

Of my two CRT's the NEC a multisync that can display anything from lowres 15khz AKA 240p up to 1024x768 resolutions. It looks nothing like those screens when displaying 15khz sources. I also have a early 90's SONY studio monitor that only does 480i and 240p and that one produces an image similar to the NEC. I forget wither they are Shadow Mask or Aperture Grill, they certainly do not have high res PC monitor mask as that sort of thing is only suited for displaying high rest graphics not lowres 15khz signals.

I do believe my raster emulation method combined with a 2-4k OLED display might stand a chance of emulating a TV CRT much more accurately though. I think even the flicker and the natural CRT softness could be reproduced with varying degrees of success. As for misalignment and such, yes, since the thing being displayed only has an actual resolution of 240p then with a 2-4k total resolution you should be able to emulate all the quirks of the mask in question. But I doubt I'll ever find out.
-=Kamikaze=- is offline  
Old 01-20-2012, 06:41 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Chronoptimist View Post

These are all problems that need more resolution to properly solve.

If you actually look at the full resolution of that "e" look at how much detail there actually is that has to be emulated. It's not just even RGB stripes, they're brighter in the centre than the edges. The full "pixel" isn't illuminated with darker shades. In many cases, it isn't even rectangular any more. Red/green/blue "subpixels" don't even have proper vertical alignment or geometry.

With a 4K resolution you have 24x18 pixels to work with to represent each colour element in a 240p image. Here are two quick examples of how one might go about representing a fully lit element on a aperture grill TV given a 24x18 block of pixels.

Here it is in actual size:


Here it is blown up:


As you see there is plenty of room to play around with and try different configurations to represent different intensities for each sub colour element. Even going as far as emulating misalignment.

Since this is just a simple filter algorithm anyone could write it for a emulator, no need to rely on any emulator. Some emulators today allow for use of custom filters, so as long as you have programming skills, a 4K display and some time you could do it yourself no problem.

I wonder how close that would get one to a CRT, maybe enough to be satisfactory even.
LL
LL
-=Kamikaze=- is offline  
Old 01-20-2012, 09:15 AM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,644
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 260
As I said, that's basically what they're doing, but newer iterations are also including things like blending pixels together to more accurately recreate how a CRT looked, depending on brightness and spot size:



Here's a magnified view from one of those images I posted earlier:


A couple of images from later in those topics (I have no idea what they're playing...)


MAME has a similar filter, but I believe it gives you the option of using a custom mask.
As I said before, those images I posted were without any kind of scanline emulation, just the phosphor mask.

I forget if this was also a Phosphor3x variant or not:


It isn't perfect, and it never will be, but they're definitely making good progress, and as I said before, I don't think we'll see anyone but people emulating games looking to recreate a CRT in this manner.
Chronoptimist is offline  
Old 01-21-2012, 03:47 AM
Newbie
 
ChuChu Flamingo's Avatar
 
Join Date: Jul 2007
Posts: 9
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by -=Kamikaze=- View Post

That is incorrect. Even 32bit systems such as the PSX, N64 and Sega SATURN put out 240p signals. There was a single generation where 480i was the standard with the occasional game supporting 480p and that was the PS2/GameCube generation. That generation overwhelmingly consisted of games using 3D graphics and thus looks pretty decent upscaled on a modern flat panel display.

There are a couple 480i games on 16 bit and 32 bit consoles. Sonic 2 Race mode, pokemon stadium/2 menus, tekken 3 all are 480i. There are more, but those are all the ones I can remember right now.

What we need in a display is something that is multisync (15Khz for 240p,31khz for 480p and more) and no set pixels. Some arcade pcb games even display in non-standard 240p so you have to accommodate for that

480i is gonna be a big mess unless it can display interlaced. Deinterlacing never works perfectly and creates input lag.

additionally we also need a way of having pixels get blended to create colors. Sega Genesis did this to make extra colors that otherwise would not be there.

We haven't even gotten to the input lag issue. Manufacturers could care less about input lag at native resolution. To make matters worse it isn't an advertised spec when it should be.

Very cool that you have a NEC and a Sony PVM. I wish I had one

Here is a great site from a member here named Fudoh. It explains in great detail all the problems of displaying retro hardware on modern displays.

http://retrogaming.hazard-city.de/index_.html
ChuChu Flamingo is offline  
Old 01-21-2012, 10:01 AM - Thread Starter
Senior Member
 
-=Kamikaze=-'s Avatar
 
Join Date: Sep 2002
Location: Canada, Denmark
Posts: 450
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Yeah, I know about Fudoh and the odd 480i game on older hardware. They were rare though. 240p was the overwhelming standard for over a decade so that is my focus.

Fudoh and his site introduced me to Micomsoft and their incredible suit of console upscalers the XRGB line. If you want pretty much zero latency 1080p upscaling of vintage signals XRGB3 or XRGB mini is the only way to go.

I shortly owned a XRGB myself but decided there was no substitude for the real thing and bought two CRT's instead as I am the sort of guy who prefers playing on actual hardware instead of via emulators.

For now I think perfect 240p emulation should be the goal as it is no doubt easiest to pull off.

If I had a 4K display at my disposal I would run some accurate CRT mask emulation tests myself, alas the highest I have at home is 1080p.

Too bad Micomsoft is a tiny niche operation otherwise we could hope to rely on them to perhaps launch a line of retro enthusiast line of OLED TVs with accurate 240p emulation built in.

Ah, if I had millions to burn...
-=Kamikaze=- is offline  
Old 01-21-2012, 02:13 PM
Member
 
andy2000's Avatar
 
Join Date: Oct 2004
Posts: 169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
It seems like a lot of the people trying to do CRT emulation have never seen a good CRT (or a CRT at all). A lot of them seem to think that the edges of the image should be curved to precisely fit the shape of a rounded corner CRT, but a real CRT will either cut off the corners, or would have a border around the active image. A good CRT should show no noticeable geometric distortion when viewed head on, and definitely not severe barrel distortion. In reality, slight pincushion distortion was far more common than barrel distortion. I also don't understand why they seem to be trying to simulate convergence errors.
andy2000 is offline  
Old 01-21-2012, 02:44 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,644
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 260
Quote:
Originally Posted by andy2000 View Post

It seems like a lot of the people trying to do CRT emulation have never seen a good CRT (or a CRT at all). A lot of them seem to think that the edges of the image should be curved to precisely fit the shape of a rounded corner CRT, but a real CRT will either cut off the corners, or would have a border around the active image. A good CRT should show no noticeable geometric distortion when viewed head on, and definitely not severe barrel distortion. In reality, slight pincushion distortion was far more common than barrel distortion. I also don't understand why they seem to be trying to simulate convergence errors.

It's not to simulate geometry problems, it's to simulate looking at the curved face of a CRT.

For example:

(a real CRT)

Personally I think it's going a step too far, same with simulating convergence errorsthough simulating a shadowmask/aperture grille/slot mask inherently adds some, and no direct-view CRT was without them, so it is inherent to true emulation of a CRT. (but not in such an exaggerated fashion as some people dothis is where having extra resolution helps)


There are many different projects, each with different goals. For some people, they want to emulate their childhood on a handed-down cheap portable CRT hooked up via an RF or composite cable.

Others want to emulate a high end arcade monitor or even a professional video monitor. (some people have even managed to get their hands on cheap broadcast monitors)

There are many different types of CRT as well, so it's not just trying to "emulate CRTs" it's trying to emulate a specific type of CRT, which may or may not be the kind you remember fondly.

My memories of gaming on CRTs are mostly using high resolution PC monitors which didn't have much in the way of visible scanlines. I don't think I've even seen a slot-mask CRT in person, for example. (at least not since I was aware of the different types)

Arcade monitors had a different look to more modern CRT TVs or computer monitors. Here's a comparison between a real arcade monitor and a PC monitor, for example:


As I keep saying, I Don't think it'll be perfect, but there's a lot of different projects out there if you search, and many of them are quite customisable, if they don't offer exactly what you want.
Chronoptimist is offline  
Old 01-21-2012, 03:32 PM
AVS Special Member
 
Nielo TM's Avatar
 
Join Date: Oct 2005
Location: London
Posts: 1,479
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
There's no way to simulate an analogue signal/format on a digital display. FED, SED, OLED, LCD etc. are all fixed pixel displays.

You may be able to simulate alternate scanning with OLED, but if you want to avoid scaling, you'll have to use Centre Timing. But that would mean the majority of the screen will be black.

Instead of trying to simulate CRT, why not focus on the deinterlacing and scaling to make sure the errors are kept to a minimum?
Nielo TM is offline  
Old 01-21-2012, 06:05 PM
AVS Special Member
 
Chronoptimist's Avatar
 
Join Date: Sep 2009
Posts: 2,644
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 260
Quote:
Originally Posted by Nielo TM View Post

Instead of trying to simulate CRT, why not focus on the deinterlacing and scaling to make sure the errors are kept to a minimum?

As far as 2D gaming is concerned, which is where most of this seems to be focused, that's already possible: nearest-neighbour scaling to the closest integer is perfect, and as has been mentioned here, most of those games ran at 240p, so deinterlacing is unnecessary. Via emulation, you can render the games at whichever resolution you choose, rather than relying on an external scaler. (most are built for video, rather than games) In my experience, even a box like the XRGB, doesn't do as good a job as emulating the games will. (and there's extra input lag from the external scaler in addition to what the display has) That said most emulators with the exception of bSNES are not perfect emulations, they use hacks to speed up the emulation process, and the biggest difference is usually that sound is not perfect when compared to the original hardware.

But "perfect" video doesn't suit the pixel art in most of these games very well at all, especially on the larger screens we have today.

Personally I don't like the filters which try to emulate a composite connection, a bad CRT, or distortion, but I wouldn't mind something that gets the "texture" of a CRT right. (the shadowmask, the scanlines etc.) The biggest problem is that most go too far, or we simply don't have enough resolution yet. Most filters which do a good job of emulating the scanlines or phosphor structure tend to dim the image a lot as a result.

Even then, they're not going to get the phosphor glow/afterglow right, but I think it'll be good enough at that point.

CRTs have more-or-less been overtaken now when it comes to video (at least they should be in a year or two with the first/second generation of OLED displays) but a CRT is absolutely the best way to play these games. The problem is that they're dying out and no-one makes them any more. I'm sure that televisions aren't quite so hard to get hold of, but arcade monitors are in short supply, and I finally had to move to an LCD after I simply couldn't get hold of a second-hand computer monitor of sufficient quality for widescreen gaming any more. (in my case, it was for modern gaming at 720/1080p as I was done with retro stuff at the time)

Unfortunately a lot of the filters these days seem to also have a "CRT gamma" component, which is probably fine if you're on an uncalibrated LCD, but if you have a CRT-like gamma on your display already (I use 2.40) then it just makes things look even dimmer. This is the case here, but it seemed like a fairly good comparison between "perfect" emulation/scaling and "degraded."

(just to be clear, anything I've posted in this topic has been sourced online, they're not my images)

And that's only scaled to 720p, when scaling to 1080p, the "perfect" image looks even more stark. The lines are too clean, a pixel becomes a giant square, and what would have blended on a CRT, just looks aliased on a modern display.

It's probably difficult to understand if you are more of a videophile and not really a retro gamer. As good as modern displays can be, and as good as OLED is going to be, I still miss having CRTs. I thought I was over it at this point (for a while, I was trying to source an affordable Sony PVM/BVM for retro stuff) but all this talk lately is giving me the itch again...
Chronoptimist is offline  
Old 01-21-2012, 06:27 PM
AVS Special Member
 
Nielo TM's Avatar
 
Join Date: Oct 2005
Location: London
Posts: 1,479
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
ah I see. I got the impression that the OP wants to configure the display to simulate CRT, which is not possible because the way FPSs are addressed different to CRT. So everything has to be done at the software level before sending the feed to the TV (e.g. like the filters you've mentioned or a hardware video filter to give the source retro look).

I'm not a retro gamer and my first Game Console was a Dreamcast . So I don't get the fascination with 2D games but that's just me ^_^
Nielo TM is offline  
Old 01-21-2012, 06:55 PM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by Nielo TM View Post

ah I see. I got the impression that the OP wants to configure the display to simulate CRT, which is not possible because the way FPSs are addressed different to CRT.

What about having a scan driver for selecting the pixel row and a scan driver for sending data (column)? Combined with an OLED pixel circuit that had no hold phenomenon then each pixel would flash for a few nanoseconds one after another line by line.

I think this is very possible but would require massive current to create enough light in such a short period of time.

Over thinking, over analyzing separates the body from the mind
xrox is offline  
Old 01-21-2012, 07:11 PM
Member
 
andy2000's Avatar
 
Join Date: Oct 2004
Posts: 169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by Chronoptimist View Post

It's not to simulate geometry problems, it's to simulate looking at the curved face of a CRT.

For example:

(a real CRT)
.

I understand that's what they are going for, but some of the examples I've seen stretch/distort the picture to match the rounded edges exactly, including the corners. I would say that your example CRT has some barrel distortion in addition to the curvature of the screen. When I set up CRTs I usually tried to make things look perfect when viewed head on, but there will always be distortion when viewing a curved screen off axis.

Ironically, a lot of what they're trying to simulate are the things people always tried to eliminate with a real CRT. If they're going to try to simulate the look of an NTSC CRT, at least try to simulate a really good one! It's almost as bad as when they add noise to a CD to make it sound like a dirty LP.
andy2000 is offline  
Old 01-21-2012, 07:13 PM
AVS Special Member
 
Nielo TM's Avatar
 
Join Date: Oct 2005
Location: London
Posts: 1,479
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Since current OLEDs are AM based, the entire row needs to be addressed with each pass. So we can't draw from left to right. But we can do alternate lines.

So line 1 (active), skip line 2 (black), line 3 (active) etc. But since OLED has negligible charge and discharge time, we may have to simulate CRT phosphor decay using multiple scans (e.g. driving the panel above 120Hz) so that the OLED isn't lit for prolonged period.

But I'm not sure how the end result would look. We may end up with flicker or combed/blurredimage. It may be better to add filter instead of trying to configure the panel no?
Nielo TM is offline  
Old 01-21-2012, 07:20 PM
AVS Special Member
 
xrox's Avatar
 
Join Date: Feb 2003
Posts: 3,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 50
Quote:
Originally Posted by Nielo TM View Post

Since current OLEDs are AM based, the entire row needs to be addressed with each pass. So we can't draw from left to right. But we can do alternate lines.

I'm no expert on addressing but if the pixel rows can be scanned one by one then obviously the pixel columns can be scanned one by one as well. Like I suggested you just need two scan drivers. One for row and one for column. Of course the pixel circuit must have a design such that the hold time equals the column scan time.

Over thinking, over analyzing separates the body from the mind
xrox is offline  
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off