XRGB-3

[Endymion]

Maybe you could take some shots in those two games, or something that I can throw onto my setup as well so I can see what you are talking about. Or maybe just tell me in words what the general problem with PS3 upscaling is? Is it that the end result is blurry?

The general prob with PS3 upscaling is that it does this by using a 480i signal which it then deinterlaces. The way to do it with best results visually would involve altering the field output (interlaced) into a frame output (progressive) first and then upscaling that. However, the PS2 has seen a lot of different development in its time and much of it was not progressive-friendly. As we have seen with Xploder and homebrew made to force progressive scan, if the game wasn't made to support it you can have other undesirable results by forcing this to happen. Sometimes you lose all control over the ratio, so a game you might like to have 16.9 can only work at 4.3, other times the engine just doesn't like being handled that way and refuses to run at all. This is why Sony went the route they did, by assuming all PS2 games at 480i and then deinterlacing before upscaling they maintain almost 100% compatibility.

The downside is as you said, the image is naturally blurred. Because it requires at least two fields to combine into one frame there is also lag inherent in the process, couple this with the lag of your HDTV and you have less window for reaction which can be important for certain games.

Speaking of which, did your PF20U ever arrive? Did you ever pick up a XRGB-3?

The 20U has been here a while now, I got it set up a couple of weeks ago. The upscaler that I had planned to use did not like the sync that my component switch was sending, which is slightly understandable as it seemed to have been damaged by lightning some years ago. I just replaced that last week, nothing older than a Dreamcast has been set up with it yet but it does look superb. It's a great deal sharper than the 9UK with a lot more control over colour, much more immediately striking than that set was. I do not have an XRGB-3 at this time, still debating it, not sure that I need it over my XRGB-2, which I'll be breaking out later this week with the Saturn.

[kamiboy]

The general prob with PS3 upscaling is that it does this by using a 480i signal which it then deinterlaces. The way to do it with best results visually would involve altering the field output (interlaced) into a frame output (progressive) first and then upscaling that. However, the PS2 has seen a lot of different development in its time and much of it was not progressive-friendly. As we have seen with Xploder and homebrew made to force progressive scan, if the game wasn't made to support it you can have other undesirable results by forcing this to happen. Sometimes you lose all control over the ratio, so a game you might like to have 16.9 can only work at 4.3, other times the engine just doesn't like being handled that way and refuses to run at all. This is why Sony went the route they did, by assuming all PS2 games at 480i and then deinterlacing before upscaling they maintain almost 100% compatibility.

The downside is as you said, the image is naturally blurred. Because it requires at least two fields to combine into one frame there is also lag inherent in the process, couple this with the lag of your HDTV and you have less window for reaction which can be important for certain games.

I see. Well, I don't think the way that the PS3 deinterlaces PS2 games is the wrong way per say. I believe it to be more correct than scaling each 240i field individually to construct each frame.

But this is largely depending on one thins, how PS2 games handle rendering of their graphical content. Let us assume that the games are rendered with a vertical resolution of 480. but for each 1/60th of a frame the game engine has to put out 240 lines out of its full vertical resolution onto the screen.

There is two ways to go about this, either render the game world 60 times a second, but only render the 240i lines that are displayable at that exact atom of time in the game engines progression. This would mean that each of these 240i fields are unrelated to the previous and blending them together would cause feathering for fast moving object, but would make static portions appear to be in higher res, and reduce jaggies in those areas.

The other way that the game engine could handle rendering of graphics is by running the game in 30fps and while rendering a full 480p frame render it in two passes one 480i field at a time. In this instance the PS3's recombining of each 480i field back into a full 480p frame is the correct way to go about it as it just puts back two pieces that belong together.

Personally I am not so knowledgeable about the inner workings of the PS2, Gamecube and other post 32bit consoles to know how they go about things. But assuming that developers can make the choice themselves I have to believe that since a game that runs in 60fps is as rare today as it was 10 years ago then most games display their internal rendered full 480p frame one field at a time, and thus they would benefit from the PS3's de-interlacing.

The way that the XRGB does it then, in this case, would be the method that would cause blurriness because instead of a 480p frame being displayed it is displaying 240 lines upscaled to 480 for each frame. You would get jaggies and blur in this instance.

If I had a XRGB3, a PS3, and a PS3 and two copies of the same game I could do a side by side and be sure. Alas I only have the XRGB-2 Plus my other PS2 is PAL, and the only PS2 games I have two copies of are not from the same region and would not be comparable due to difference in resolution.

[Endymion]

I see. Well, I don't think the way that the PS3 deinterlaces PS2 games is the wrong way per say. I believe it to be more correct than scaling each 240i field individually to construct each frame.

What? Any deinterlacing is the wrong way to go about it, at least, it is not the most desirable way to do it.

But this is largely depending on one thins, how PS2 games handle rendering of their graphical content. Let us assume that the games are rendered with a vertical resolution of 480. but for each 1/60th of a frame the game engine has to put out 240 lines out of its full vertical resolution onto the screen.

There is two ways to go about this, either render the game world 60 times a second, but only render the 240i lines that are displayable at that exact atom of time in the game engines progression.

This is very uncommon for the simple reason that it is a very inefficient way to do it, even on the PS2. In fact, I can't think of any games that do this right now.

This would mean that each of these 240i fields are unrelated to the previous and blending them together would cause feathering for fast moving object, but would make static portions appear to be in higher res, and reduce jaggies in those areas.

The other way that the game engine could handle rendering of graphics is by running the game in 30fps and while rendering a full 480p frame render it in two passes one 480i field at a time. In this instance the PS3's recombining of each 480i field back into a full 480p frame is the correct way to go about it as it just puts back two pieces that belong together.

The way it's done is this. The screen has to be drawn, 640x480. Or perhaps lesser resolutions (and on the PS2, this is often). And that entire screen resolution is what is rendered. What the video chip passes off (not the GS, the one that actually sends the video to your screen) is then either a field or a frame. If either the field or the frame is less than 640x480, the GS gives the video chip instructions on how to stretch it, accounting for typical overscans, so that what arrives at your screen appears fullscreen, and for all you know, 640x480 (unless you start counting stairsteps). If the output is to be progressive, then it sends the full frame and not an alternating field. It really is that simple.

The reason homebrews and Xploder glitch is varied. 1. They can mess with that stretching that is involved because they operate by flushing the entire framebuffer to the video chip; by doing this they may miss the information it needs to stretch it properly, leaving you adjusting your screen controls. 2. They have to "live" somewhere in the memory of the PS2 (or the PS3) and so they can actually occupy space that the game program needs in order to function, causing a freeze, a crash, a weird bug, at the start or later in the game, you never have any way of knowing because this is the development stuff that is not public knowledge and the program that forces the framebuffer has to go somewhere. And 3. ultimately it is messing with a program, making it do something it was never tested for. Some games, for instance, rely upon the framerate of the pass to the video chip as the tick for the game timing, and if you drop the contents of a full frame at once (two fields rendered in the same pass) you get a game that runs at twice the speed! TimeSplitters does this with XPloder and GSM, for instance. It makes the thing run so fast it's unplayable. More typically game programs that encounter this stuff unexpectedly and may crash.

But dropping the entire contents of the framebuffer is absolutely the "proper" way to achieve 480p from 480i with discrete computer graphics, and this is something that the PS3 never, ever, ever does, at least not in any way that the PS2 does not do it already. It deinterlaces two fields when this is not necessary to get a full 480 line frame, i.e. there are actually four fields alive and well in a single two-field pass, it's just that two of them are completely unused. That is why deinterlacing will never be the proper way to achieve 480p, at least not with any kind of computer game graphics. What the PS3 does is a compromise because so many devs did 480i games only, and did them with various rendering routines, stretching, and so on that you never know what you are going to wind up with when you flip the switch to activate progressive scan. Ideally it really should be just so simple as flipping that switch, and it is this simple on other systems, XBox for example, but there was a lot more going on with PS2 development, a lot more kittens to herd.

Personally I am not so knowledgeable about the inner workings of the PS2, Gamecube and other post 32bit consoles to know how they go about things. But assuming that developers can make the choice themselves I have to believe that since a game that runs in 60fps is as rare today as it was 10 years ago then most games display their internal rendered full 480p frame one field at a time, and thus they would benefit from the PS3's de-interlacing.

If I understand this, you seem to be under the impression that moving to 480p from 480i should yield 60fps. Believe me, I wish that were the case, but it totally isn't true. At any rate I'm not sure what you think you can gather from the framerate of a game with respect to its method of broadcast. It is never easier to render interlaced, and that's why nobody does it. They render progressive, at whatever framerate they can manage. The video broadcast sync is completely independent of the framerate that the game is rendered at, if its user has requested interlaced then half of the frame is discarded--all of those odd lines from that frame go out the window never to be seen, and the next field is taken from the odd lines of the next frame. If the user has requested a progressive scan, none of this trashing is ever necessary and the entire image that is rendered is what is broadcast. Case in point here, you're not going to have a lower rendered framerate between 480i and 480p, the same resolution is rendered, half of the work done is just chucked. That's why it's always desirable to just get 480p the right way. You have a smoother animation, fewer jaggies, none of the scissored edges inherent with interlacing, and yet no slowdown. You've just allowed yourself to see the animation the way the computer rendered it.

Deinterlacing is not the same as bumping the full frame out the door. Deinterlacing only ever works with the fields of adjacent draws, never with the same two fields created in a single rendered frame. This is why it is only ever undesirable. Of course depending on circumstances, it might be the best thing you can get.... but I'd rather just drop to 480i than deal with any deinterlacing, honestly.

[Endymion]

Missed this part:

The way that the XRGB does it then, in this case, would be the method that would cause blurriness because instead of a 480p frame being displayed it is displaying 240 lines upscaled to 480 for each frame. You would get jaggies and blur in this instance.

But you do get a sub-optimal image from an XRGB, at least if you consider that the source can do 480p! That's not really the purpose of the XRGB for the most part anyway, we use XRGBs for those sources that only work with interlaced modes (Saturn, PS1, SNES, etc.), and they upscale without deinterlacing. That's why there is a world of difference between a PS2 game (say one that cannot work with 480p, or cannot force to that with homebrew/Xploder) when played over XRGB and the same game played through the PS3. The PS3 deinterlaces it, the XRGB does not, it fakes scanlines or line-doubles. One makes it look more like a CRT, the other just makes it appear blockier vertically. The PS3 just blurs it by combining temporally adjacent fields, never the same fields from the same renders.

[kamiboy]

If I understand this, you seem to be under the impression that moving to 480p from 480i should yield 60fps. Believe me, I wish that were the case, but it totally isn't true. At any rate I'm not sure what you think you can gather from the framerate of a game with respect to its method of broadcast. It is never easier to render interlaced, and that's why nobody does it. They render progressive, at whatever framerate they can manage. The video broadcast sync is completely independent of the framerate that the game is rendered at, if its user has requested interlaced then half of the frame is discarded--all of those odd lines from that frame go out the window never to be seen, and the next field is taken from the odd lines of the next frame. If the user has requested a progressive scan, none of this trashing is ever necessary and the entire image that is rendered is what is broadcast. Case in point here, you're not going to have a lower rendered framerate between 480i and 480p, the same resolution is rendered, half of the work done is just chucked. That's why it's always desirable to just get 480p the right way. You have a smoother animation, fewer jaggies, none of the scissored edges inherent with interlacing, and yet no slowdown. You've just allowed yourself to see the animation the way the computer rendered it.

Deinterlacing is not the same as bumping the full frame out the door. Deinterlacing only ever works with the fields of adjacent draws, never with the same two fields created in a single rendered frame. This is why it is only ever undesirable. Of course depending on circumstances, it might be the best thing you can get.... but I'd rather just drop to 480i than deal with any deinterlacing, honestly.

I see, well, it seems I overestimated the quality of old hardware. I never envisioned a scenario where half of a game's resolution would be thrown away. I thought that either games would render a full 640x480 frame and once the frame buffer was switched to begin drawing the next frame the previous frame was rasterized onto the TV via two fields each containing alternate scanlines of the full image.

Either that or I thought that the PS2's graphics chip was made so that it could render the game graphics one field at a time. There are 3D hardware that can do this, and it would have made sense for the PS2 to have hardware that was capable of this, then it needn't throw anything away unused.

[Konsolkongen]

Just to rule out the possibility of the power supply being the problem with my XRGB-3 i want to try with another one. But 11V 2A PSU's are quite difficult to find. However i can find several 12V 2A PSU's on eBay, will those work?

[Endymion]

Either that or I thought that the PS2's graphics chip was made so that it could render the game graphics one field at a time. There are 3D hardware that can do this, and it would have made sense for the PS2 to have hardware that was capable of this, then it needn't throw anything away unused.

Doing it this way is incredibly inefficient however, if you are ever going to have anything progressive at all. This is excepting old, 2D hardware (Genesis, SNES) that would run a noninterlaced image for lower resolutions, but where modern, 3D graphics are concerned, if you set up a system that discretely renders interlaced in this way, you have a problem on your hands the moment you want to make it render progressively... you're immediately placing twice the workload on it! The benefit of culling a field from a completed frame should then become obvious.

[Fudoh]

That is why deinterlacing will never be the proper way to achieve 480p, at least not with any kind of computer game graphics

I have to disagree with that. Strongly. Have you ever used a good deinterlacer for 480i videogame material ? Motion and pixel-adaptive deinterlacing for video material can enhance 480i material in so many ways. Field scaling (no deinterlacing, just the processing of the every single field) is an option for games with lots of action on-screen and only few static areas. Nevertheless field-scaling always comes with heavy artefacts (line flicker, feathering, blurriness, interpolation). The beauty of a 480p picture derived from a 480i video stream using a good deinterlacer can get quite close to a natively rendered 480p output (depends on the game and the action on-screen).

we use XRGBs for those sources that only work with interlaced modes (Saturn, PS1, SNES, etc.)

vs.

This is excepting old, 2D hardware (Genesis, SNES) that would run a noninterlaced image for lower resolutions

you confuse people by such statements. While the classic systems' output is 15khz, it's not interlaced. You mean the right thing of course, but a statement like the one above doesn't improve people's knowledge...

...when played over XRGB and the same game played through the PS3. The PS3 deinterlaces it, the XRGB does not

it doesn't deinterlace in B1 mode, but it does in B0 - and badly, very badly.

The PS3 just blurs it by combining temporally adjacent fields, never the same fields from the same renders.

that's why the PS3 isn't a good choice for PS2 480i titles, but don't jump to wrong conclusions because of some dickheads at Sony. As said before, a good video deinterlacer makes a BIG difference in 480i titles.

[kamiboy]

Doing it this way is incredibly inefficient however, if you are ever going to have anything progressive at all. This is excepting old, 2D hardware (Genesis, SNES) that would run a noninterlaced image for lower resolutions, but where modern, 3D graphics are concerned, if you set up a system that discretely renders interlaced in this way, you have a problem on your hands the moment you want to make it render progressively... you're immediately placing twice the workload on it! The benefit of culling a field from a completed frame should then become obvious.

Yes, rendering a full scene in two interlaced passes will take longer, but I think you miss my point of rendering each interlaced half will each be faster than doing a full progressive pass. 3D hardware can be designed in a way that it can render a scene one line at a time but in a non-progressive fashion. Some older graphics cards for an example, had the ability to do this, and people could make use of it by buying two copies of the same card and link them together via a cable and each would render one part of a scene for close to twice the performance.

I thought since the PS2 and other previous gen proper 3D consoles were designed to be displayed in interlaced displays they would have specially designed hardware capable of interlaced rendering.

To make my point clear, here is how it would have been done. Once the game engine has finished doing all calculations for the scene this frame it would signal the graphics hardware to begin rendering the scene. The 3D hardware starts rendering first all the odd lines of the scene, and as soon at all the 240 odd lines have completed rendering that portion would begin being beamed onto onto the screen via the televisions electron beam doing its pass of the field of odd lines while the 3D hardware renders the even lines of the scene.

As soon as this process is finished the game engine begins calculations for he next frame. There are caviats in place for this process to work smoothly, such as the need for the game engine to be able to put out at least 30fps.

Anyway, today all of these issues have been partially solved by us moving over to consoles designed to be connected to progressive displays, alas too many games have vsync related tearing issues.

But regardless, I still think PS2 games look amazing, despite my running them via my PS3. The PS2 was the golden age of 3D gaming after all.

[RGB32E]

Doing it this way is incredibly inefficient however, if you are ever going to have anything progressive at all. This is excepting old, 2D hardware (Genesis, SNES) that would run a noninterlaced image for lower resolutions, but where modern, 3D graphics are concerned, if you set up a system that discretely renders interlaced in this way, you have a problem on your hands the moment you want to make it render progressively... you're immediately placing twice the workload on it! The benefit of culling a field from a completed frame should then become obvious.

Yes, rendering a full scene in two interlaced passes will take longer, but I think you miss my point of rendering each interlaced half will each be faster than doing a full progressive pass. 3D hardware can be designed in a way that it can render a scene one line at a time but in a non-progressive fashion. Some older graphics cards for an example, had the ability to do this, and people could make use of it by buying two copies of the same card and link them together via a cable and each would render one part of a scene for close to twice the performance.

I thought since the PS2 and other previous gen proper 3D consoles were designed to be displayed in interlaced displays they would have specially designed hardware capable of interlaced rendering.

To make my point clear, here is how it would have been done. Once the game engine has finished doing all calculations for the scene this frame it would signal the graphics hardware to begin rendering the scene. The 3D hardware starts rendering first all the odd lines of the scene, and as soon at all the 240 odd lines have completed rendering that portion would begin being beamed onto onto the screen via the televisions electron beam doing its pass of the field of odd lines while the 3D hardware renders the even lines of the scene.

As soon as this process is finished the game engine begins calculations for he next frame. There are caviats in place for this process to work smoothly, such as the need for the game engine to be able to put out at least 30fps.

Anyway, today all of these issues have been partially solved by us moving over to consoles designed to be connected to progressive displays, alas too many games have vsync related tearing issues.

But regardless, I still think PS2 games look amazing, despite my running them via my PS3. The PS2 was the golden age of 3D gaming after all.

Yeah, many PS2 games at 480i render at half vertical resolution - field rendering.... There's no point in rendering the full 6/7xx by 4xx frame when you're not going to support 480p. For example, a PS2 title that only supports 480i output (e.g. 640x448) will render 640x224 each 60Hz period. Rasterization takes into account the odd/even fields for rendering. Processing time is wasted when rendering a field that isn't sent to the interlaced display. Field 1 - 640x224, Field 2 - 640x224, effective resolution on the display - 640x448.

[RGB32E]

Speaking of which, did your PF20U ever arrive? Did you ever pick up a XRGB-3?

The 20U has been here a while now, I got it set up a couple of weeks ago. The upscaler that I had planned to use did not like the sync that my component switch was sending, which is slightly understandable as it seemed to have been damaged by lightning some years ago. I just replaced that last week, nothing older than a Dreamcast has been set up with it yet but it does look superb. It's a great deal sharper than the 9UK with a lot more control over colour, much more immediately striking than that set was. I do not have an XRGB-3 at this time, still debating it, not sure that I need it over my XRGB-2, which I'll be breaking out later this week with the Saturn.

Thats cool! The DC using a VGA box? How's the 15kHz RGB support via the PC input? Is Panasonic still treating it as though it's 480i?

Yeah, you need to get a XRGB-3... if nothing else to see how B0 @ 1080p looks! In any case, take some pix wo/flash and on a tripod of the XRGB-2/SS action!

[kamiboy]

Yeah, many PS2 games at 480i render at half vertical resolution - field rendering.... There's no point in rendering the full 6/7xx by 4xx frame when you're not going to support 480p. For example, a PS2 title that only supports 480i output (e.g. 640x448) will render 640x224 each 60Hz period. Rasterization takes into account the odd/even fields for rendering. Processing time is wasted when rendering a field that isn't sent to the interlaced display. Field 1 - 640x224, Field 2 - 640x224, effective resolution on the display - 640x448.

Actually the technique you describe there would yield a game with only half the vertical resolution. This because even though only 224 lines are rendered each 1/60th of a frame, they are the same 224 lines, thus you get a line doubling effect as the same 224 lines are being beamed for both even and odd lines. In this scenario static images would appear to have half the vertical resolution as would a scene rendered via specialized hardware that can render the even and odd lines of one full 480 frame buffer separately. Either that or just render the full 480 lines progressivally and then wait a frame for each interlaced portion of it to be displayed on the television before moving to rendering the next frame.

This was often done on PC games with a technique called frame swapping or some such. The graphics card would have two full screen sized framebuffers and would draw onto one as the previously drawn one was being displayed, I am surprised that consoles didn't even go that far.

[RGB32E]

Yeah, many PS2 games at 480i render at half vertical resolution - field rendering.... There's no point in rendering the full 6/7xx by 4xx frame when you're not going to support 480p. For example, a PS2 title that only supports 480i output (e.g. 640x448) will render 640x224 each 60Hz period. Rasterization takes into account the odd/even fields for rendering. Processing time is wasted when rendering a field that isn't sent to the interlaced display. Field 1 - 640x224, Field 2 - 640x224, effective resolution on the display - 640x448.

Actually the technique you describe there would yield a game with only half the vertical resolution. This because even though only 224 lines are rendered each 1/60th of a frame, they are the same 224 lines, thus you get a line doubling effect as the same 224 lines are being beamed for both even and odd lines. In this scenario static images would appear to have half the vertical resolution as would a scene rendered via specialized hardware that can render the even and odd lines of one full 480 frame buffer separately. Either that or just render the full 480 lines progressivally and then wait a frame for each interlaced portion of it to be displayed on the television before moving to rendering the next frame.

This was often done on PC games with a technique called frame swapping or some such. The graphics card would have two full screen sized framebuffers and would draw onto one as the previously drawn one was being displayed, I am surprised that consoles didn't even go that far.

Double buffering, or in some cases "tripple buffering"... having multiple frame buffers to render to....

Why render a full frame if only half of the rows of pixels are all that ever get displayed on the TV? Not to say all PS2 games are made this way...

[kamiboy]

Double buffering, or in some cases "tripple buffering"... having multiple frame buffers to render to....

Why render a full frame if only half of the rows of pixels are all that ever get displayed on the TV? Not to say all PS2 games are made this way...

Well, for one thing, you would be displaying the full frame of pixels, only in two successive passes each lasting 1/60th of a second. This, I believe, is fast enough for the human visual system to blend them into one complete high rest image, though with some perception of flickering.

[SGGG2]

Metal Gear @ 480p transcoded by the X-Select, scaled by the XRGB-3 @ 1080p via DVI. The ratio's incorrect forced into 480p, setting the XRGB-3 to 16:9 (as below) gets a closer picture. http://imgur.com/9E3YI.jpg http://imgur.com/DEovf.jpg http://imgur.com/kXXZj.jpg

[kamiboy]

Well, here is a shot of the same spot of one of your shots. For my money this game looks absolutely amazing, as beautiful today as it did 10 years ago when it first dazzled me.

My camera is not very good in low light conditions, as such things look a little more blurry than they really are, but all things being equal I think in terms of quality what I have here on my screen is not lacking in any area. At first I thought the image I was getting was a bit too soft, but after cranking up sharpness the details came alive and the game looked even more amazing, and there were no edge enhancement or artifacts to be seen anywhere.

I certainly have no grievances regarding what I see here in front of me, the image is colourful, sharp and as a result of eliminating all analogue steps in the path from chip to LCD cell is also devoid of any noise or the horrid shaking I see from the XRGB on my LCD. I only wish Nintendo could have at least been bothered to put out a backwards compatible console with an all digital video interface. But hey, its Nintendo, we should count our blessing that the Wii even does component out. I am sure there were a few bean counters in the bowels of that scrooge of a company that were rubbing topence together over how many of those denominations that could be saved per unit if only they'd remove the simple hardware that enabled RGB to component conversion from the Wii's motherboard.

[SGGG2]

Here's the same shot through the PS3, no smoothing:

I can appreciate soft scaling, and see why some would prefer it (less ailising, dithering, etc). Some games might plain look better, but Gradius V, for example, looks irrefutably terrible on the PS3. What I really want is a region free PS2 that renders internally @ 720p. I wonder if the next Nintendo console will render Wii and Gamecube titles at higher resolutions. From what I've seen, a 1080i HD CRT can't be beat in terms of image quality (for PS2, Xbox, GC) plus, there's no discernable difference between interlaced and progressive component sources. The downside's that the large ones (32+) can weight 200 pounds or more.

The XRGB-3 image is noise free thanks to the X-Select. The A/D conversion on the XRGB is really noisy with the PS2.

[kamiboy]

Well I say, that is a very soft image you got there. I'd say what I have on my TV is closer to your XRGB image than the PS3 image. I think you have your TV's sharpness setting set too low. On my LCD once sharpness is cranked up I get a very clear mage with no evidence of soft scaling, it looks great.

[SGGG2]

Is that so? Sharpness regulation on my set doesn't make any difference. I want a 3D plasma, dammit.

EDIT: It's just as soft on my ASUS VH236H

[ShutokouBattle]

I just had a crazy idea. They make some pretty decent, free video processing software for PC's, right? Why not build a small computer that would do the same thing as an XRGB-3? You might have to do a bit of programming to make it convenient and fast, but ultimately wouldn't that be cheaper than an XRGB-3? You would also have a range of options about image size, type of processing, and output - you could probably use an HDMI output at 1080p! It would also deinterlace 480i images well (probably), so you could play PS2 and Gamecube games on it. The only problem I can see with this idea is that I don't know how expensive or hard to find RGB capture cards are. Is this a completely idiotic idea?

[Fudoh]

I just had a crazy idea.

several problems with that... usually too many interfaces to keep the lag low (capture card to OS, processing in the OS, OS to graphics card). Hardware based deinterlacing solutions (Holo3DGraph) came to ~3 frames of delay with a fast PCIe graphics card. Another problem is that good 480i deinterlacing takes quite some hardware power. Just have a look at current video playback sofware. Video deinterlacers with 60fps output (Yadif 2x) have only just recently been implemented. Also newer capture hardware (which supports 480p) does not support 240p video timings, so you currently just don't have the hardware available to do everything with one capture board.

Even with the lowest-end hardware, you'll always end up with the price of a XRGB-3. There are relatively good and cheap 480i deinterlacers available out there. Just pair up one of those with a XRGB-3 and you've got a very nice all-purpose processing setup for less than - let's say - 500 EUR.

[RGB32E]

I really don't think B0 @ 1080p with the PS2 is all that bad... perhaps since I'm using a shortened official stock cable with premium capacitors?

Also, I use AFC Level of 1... this helps the shakes!

[kamiboy]

Is that so? Sharpness regulation on my set doesn't make any difference. I want a 3D plasma, dammit.

EDIT: It's just as soft on my ASUS VH236H

Hmmm... strange, but how would a 3D plasma help in this instance? My previous display was PDP, and they have a softer, but more natural looking image. Strange that both your TV and LCD have a soft image, so I doubt it is related to the processing of either.

Have you your PS3 set to upscale to 1080p? What about the scaling settings, have you set it to Normal or Full?

I have 1080p and Normal. I have super white on and RGB set to limited. Other than that you might want to try a few different preset modes on your TV. The ones named movie/cinema or some such default on a very soft picture. Vivid/colourful have terribly overblown colours but very good sharpness. I use the one called normal on mine for everything since it gives the best results overall.

[Endymion]

At first I thought the image I was getting was a bit too soft, but after cranking up sharpness the details came alive and the game looked even more amazing, and there were no edge enhancement or artifacts to be seen anywhere.

But that's what jacking up the sharpness does, edge ehancement. Don't forget the inherent delay.

Well I say, that is a very soft image you got there. I'd say what I have on my TV is closer to your XRGB image than the PS3 image. I think you have your TV's sharpness setting set too low. On my LCD once sharpness is cranked up I get a very clear mage with no evidence of soft scaling, it looks great.

Ideally you'd want sharpness to zero on an HDTV, which should turn it off completely.

I have 1080p and Normal. I have super white on and RGB set to limited.

Superwhite settings on the PS3 only affect movies (DVD, BluRay, etc.).

[RGB32E]

But that's what jacking up the sharpness does, edge ehancement. Don't forget the inherent delay.

Sharpness != Edge Enhancement... as they are different adjustments! Do you know of which manufacturers actually cause input delay when using EE and/or Sharpness settings (hard data)?

Ideally you'd want sharpness to zero on an HDTV, which should turn it off completely.

Just remember that sharness set to zero varies from manufacturer implementation... For my Sony TV, sharpness of zero is literally "50"! Lowering from 50 causes the image to blur!!! So, zero isn't necessarily zero, as it depends upon the manufacturer!

[Endymion]

But that's what jacking up the sharpness does, edge ehancement. Don't forget the inherent delay.

Sharpness != Edge Enhancement... as they are different adjustments!

When he's jacking up his sharpness, he's undoubtedly jacking up EE as well, pretty sure his TV doesn't have an EE setting you think? There is some "detail" in his image but Snake almost looks cel-shaded!

Do you know of which manufacturers actually cause input delay when using EE and/or Sharpness settings (hard data)?

I was referring to his use of the PS3. Could be some delay with processing that kind of variance as well but I don't know about comparatively. Every lag test I ever see attempts it with that stuff off so I don't know of any specific test against it.

Just remember that sharness set to zero varies from manufacturer implementation... For my Sony TV, sharpness of zero is literally "50"! Lowering from 50 causes the image to blur!!! So, zero isn't necessarily zero, as it depends upon the manufacturer!

Well whatever "zero" level of EE happens to be for your particular controls, you know what I mean.

[kamiboy]

Just remember that sharness set to zero varies from manufacturer implementation... For my Sony TV, sharpness of zero is literally "50"! Lowering from 50 causes the image to blur!!! So, zero isn't necessarily zero, as it depends upon the manufacturer!

Precisely, and different modes, as I said earlier, have different sharpness scales. Like on my TV the cinema setting gives a very soft picture, but if you turn up sharpness in it then you get a similar image as on 0 on another setting. On my television I usually have sharpness on its default, because cranking it up will not look good for video content as it brings out all the warts. But for PS2 games on the PS3 its seems, because it is a purely digital image I can turn it all the way up without any detrimental effects. In fact it makes things look a lot better.

[kamiboy]

When he's jacking up his sharpness, he's undoubtedly jacking up EE as well, pretty sure his TV doesn't have an EE setting you think? There is some "detail" in his image but Snake almost looks cel-shaded!

Not the one in my game for certain. That is just the bad camera and unsteady hand. look rather at the jaggies and the sharp lines of the soliton radar window. I posted the image because I thought that by looking at those two things you could see that what I had was a very clear picture. Seeing as how I spent mad amount of money on a device just so I could hook legacy consoles to my flat panel should speak of my being a videophile with a discerning eye. I know a bad picture when I see it, what I get from my PS3 looks stellar, at least I am very happy with it, which is not so much I can say about the XRGB.

[SGGG2]

I really don't think B0 @ 1080p with the PS2 is all that bad... perhaps since I'm using a shortened official stock cable with premium capacitors?

Also, I use AFC Level of 1... this helps the shakes!

I'm amazed you're getting such a clean image direct from the XRGB-3. I guess those caps really do the trick, my monster cables are noisy as hell unless they're sent through the X-Select first. 1080p B0 ranges from good to fantastic for 480p signals, except for my crappy Gamecube, ugh.

The moire pattern is from not scaling the image down evenly, right? Is the source interlaced or forced progressive? What kind of set is that? I can clearly see a black pixel grid.

Hmmm... strange, but how would a 3D plasma help in this instance? My previous display was PDP, and they have a softer, but more natural looking image. Strange that both your TV and LCD have a soft image, so I doubt it is related to the processing of either.

Have you your PS3 set to upscale to 1080p? What about the scaling settings, have you set it to Normal or Full?

Just bitchin, I hate the mediocre black levels on my LCD and 3D would just be cool. The PS3 is outputting the native 1080p resolution of my monitors. Smoothing off. Normal aspect. None of the other settings seem to do anything.

[trunk]

All these screenshots are good reference material.

I ended up buying a dvdo edge and I am happy to have even scanlines. Here is the before and after shots:

Without Edge:


With Edge:


The picture itself was just taken real quick with no lights or flash, the image quality looks better in person than the picture. It is also more obvious moire while the game is in motion.

Thanks Fudoh as well as everyone here for the great advice.