bsnes v0.039 released

[ShadowFX]

I have a quick question about the video/audio sync function. I recently invested in new hardware, the audio crackling I had with my previous system (which did meet recommended requirements), is now completely gone. Could it be that sync requires a bit more powerful system or simply better tweaking in the audio settings?

I'm sorry if this has been asked before.

[tetsuo55]

For some reason(i have no idea why) the first games decided to ignore interlaced and used a progressive hack. (instead of rendering to alternate scanlines render to the same scanline twice as fast)

The result was black empty scanlines between each colored scanline.
Due to all kinds of CRT wierdness these lines where more or less visable.

There was also the NTSC color artifacting to play with.

At some point(very early) programmers decided to use the artifacts (and to a lesser extent the scanlines) to render parts of the image.

Artist intent often requires ntsc artifacts/scanlines/kell-effect.

Several games require these to render things like backgrounds, transparency or colors beyond the limited palet.

IMHO emulators should have 3 options.
Pure direct, (current non-filtered style)
Pure direct with scaling filters, (current filtered style)
Display simulation(Simulating CRT, in the case of mame this could very precisely simulate the included arcade monitor)

[blargg]

The scanline effect encourages the brain to fill in the detail that doesn't exist, resulting in a picture that looks less jagged, even though it really isn't.

Yes, it probably does enhance image reproduction, since the black lines make it closer to a gaussian function. Displaying samples ("pixels") with a gaussian function yields better results. Even having a video game console output interlaced on a regular TV (basically eliminating the black scanlines) causes everything to look blocky.

disgruntleddesigner.com has comparison shots of several games on several systems.

Too bad his composite screenshots are all displayed on a TV with a comb filter, which causes excessive hanging dots on many systems. For example, see the Pilotwings shot near the top. If you have a Sony CRT TV with the service menu, try changing CTRP to 0. It makes a big difference on my 27-inch.

[creaothceann]

The scale2x filter makes this god awful watercolor effect that smudges everything haphazardly together and destroys detail.

IMO it creates detail by creating color transitions.

It depends. See pic1 and pic2 - I'd say the interpolation is very good for the grass and the trees, but not for the small text.

[Snark]

Anyone who seriously believe old-era NES-SNES etc (not talking about modern console here which are definitely meant to be played on HD displays) consoles actually look "better" on modern displays (unfiltered) have some serious vision and/or mental issues. /end extremely annoying and pretentious comment as if I possessed "The Truth"

[ZH/Franky]

Anyone who seriously believe old-era NES-SNES etc (not talking about modern console here which are definitely meant to be played on HD displays) consoles actually look "better" on modern displays (unfiltered) have some serious vision and/or mental issues. /end extremely annoying and pretentious comment as if I possessed "The Truth"

Could you recommend me a good doctor?

I like my pixels sharp.

[Snark]

Anyone who seriously believe old-era NES-SNES etc (not talking about modern console here which are definitely meant to be played on HD displays) consoles actually look "better" on modern displays (unfiltered) have some serious vision and/or mental issues. /end extremely annoying and pretentious comment as if I possessed "The Truth"

Could you recommend me a good doctor?

I like my pixels sharp.

Actually, I don't think this require any medical attention :P And I think whether someone prefer the image filtered or unfiltered is a matter of preference and is fine either way. I was mostly echoing Fitzroy comment.

[ZH/Franky]

Anyone who seriously believe old-era NES-SNES etc (not talking about modern console here which are definitely meant to be played on HD displays) consoles actually look "better" on modern displays (unfiltered) have some serious vision and/or mental issues. /end extremely annoying and pretentious comment as if I possessed "The Truth"

Could you recommend me a good doctor?

I like my pixels sharp.

Actually, I don't think this require any medical attention And I think whether someone prefer the image filtered or unfiltered is a matter of preference and is fine either way. I was mostly echoing Fitzroy comment.

Haha, I was only joking about asking for a doctor

[Snark]

Haha, I was only joking about asking for a doctor :P

Ya I know :P Just making sure no one takes my post literally.

Too bad his composite screenshots are all displayed on a TV with a comb filter, which causes excessive hanging dots

Googlebooks says "Restricted page"... (not that I would understand any of it anyway)

[FirebrandX]

I don't particularly like a completely unfiltered image. Just 25 or 50% scanlines on a a sharp, point-filtered image is all I need. It reminds me most of using my analog 25" RGB monitor from years back. I did a lot of wiring and soldering back then and had just about every system hooked up to it in RGB. I'll never forget the first time I hooked the Genesis/Megadrive up in RGB, the difference was phenominal.

At any rate, what got me ranting on this whole fullscreen debate was certain people like Fritzroy wanting to take away control from the user when it comes to being able to render a non-stretched image in fullscreen. Every emulator that does not have this ability is SORELY lacking in my book. I hope byuu never listens to Fritzroy on that one. It would be a bad mistake.

[_willow_]

But still we can share some points, yes?

1) Aspect Ratio. We need the width to be scaled by non-integer anyway.
2) Software filters only making sense on upscaling. There is no value for 1:1 scale filter.
3) Point filter is a no go solution for integer scale. Simply because we cant get the integer multiplier for width (see issue 1)
4) Any filters supposed to do the postprocessing job by altering the colors of the scaled image by random noise or some matrix rule.
5) I would like the filters to stack in two passes
- Scale (nearest,hybrid,linear)
- Noise (ntsc,scanlines,whatever else)

Self-scale filters like HQ2x and Scale2x should not taking place in precise emulation. It's an example of self-generation which should not taking place in software like bsnes. Scanlines are ok to me as they do not generate nothing new, just alters.

I do stress on replacement for point filter so that it would allow linear-filtered rows but point-filtered columns. It's very tube like. Nor linear nor point like. Let's name it hybrid scaling.

We can leave linear filter alone as it is cheap hardware filter with no processing penalties involved. Point filter is deprecated it can't work right in any possible way. It should be replaced with hybrid scale

[FitzRoy]

First of all, how did the argument mutate from scanlines to bilinear filtering? The elvis example is contrived. Why isn't the image something typical like a real photo, why is it only using a max contrast outline? Are they afraid a more subtle image would be indistinguishable on both? Even if we grant that in this manufactured case, the filter helped something become distinguishable, it can only become distinguishable through straining the eyes to make sense of a thick haze. It would be completely undesirable for any long-term viewing.

Scale2x and HQ2x try to accomplish what gaussian does without the haziness. The more it tries to overcome jaggedness on what should be rounded, the more edges that are supposed to be jagged suffer. It's just trading one unpleasantry for another, except this time your brain can't figure out why wavy stairsteps exist in what's purporting to be high resolution.



I want you guys to look at this image. Not only does scanline filtration do virtually nothing to defeat stair stepping for such a lo-res source, but it makes areas that are supposed to be contiguous swaths of color look like wire mesh: text, the sky, the cement walls to the left, the lifebar, the sunspots on people's clothes. They're all interrupted by pinstripes that Gil Hamilton says you're supposed to imagine as the colors they would have been had they not been inserted, and that creates detail. Can you follow that logic? I can't.

At any rate, what got me ranting on this whole fullscreen debate was certain people like Fritzroy wanting to take away control from the user when it comes to being able to render a non-stretched image in fullscreen. Every emulator that does not have this ability is SORELY lacking in my book. I hope byuu never listens to Fritzroy on that one. It would be a bad mistake.

Before you call the wahmbulance, note that (a) you still would have been able to do your precious integer-scaled filtration in windowed mode and (b) I was still willing to do separate settings if "a" unleashed the nostalgic hordes, which it did. I'm not on a crusade against filters and 192khz audio (although the ignorance they beget makes me want to). Let's review what I really thought would benefit usage:

1. no menu or statusbar toggles, no menu or statusbar in fullscreen mode
2. fullscreen mode to be configurable without being in it.
3. space to have multiple, descriptive aspect ratio correction options, which the menu is unable to provide
4. space to explain that filters are hardware vs software, which the menu is unable to provide.

byuu knows these are valid arguments, but he has gotten too used to the current setup, and the point/scanline people prevent us from making fullscreen fullscreen. So that's it, we look for ways to improve elsewhere and move on.

[henke37]

I like that super eagle filter, too bad that it will never find it's way into bsnes due to licensing issues.

[Gil_Hamilton]

... it makes areas that are supposed to be contiguous swaths of color ...

I'm just going to ignroe the personal attack and jump on that word "supposed".

It is SUPPOSED to have those dark lines. It was DESIGNED to output to a low-resolution CRT monitor with widely separated scanlines.


It's also supposed to be a 4:3 image, and something seems to have gone VERY wrongin the upper-right, as I'm pretty sure that sea of random black dots ISN'T supposed to be there, in code OR in the assumed final output device.

[FirebrandX]

What I am saying is a max-stretched image looks like crap on an LCD if the native res is not a multiple of the emulated image. This is why I prefer the current system of letting the user choose a scale multiple for fullscreen.

I took a Super Metroid image with lots of text and expanded it to 2x with scanlines. Then I expanded it to 4x with square pixels, and also 4x with stretched pixels for proper aspect ratio. Finally, I stretched it to 4.6875x with square and stretched pixels, to eliminate black borders on the top and bottom of an LCD that's 1050 pixels high. Not sure how you wanted scanlines, but ignoring possible issues with them, how do these look on your 1680x1050 LCD, particularly the stretched ones? It's not simple bilinear, so pixels have very little blur.

Terribly sorry I overlooked this post.

I had a look at all the images in a fullscreen preview environment to test them out on my display. Here are the results:

The 4x scaled + aspect stretching looked pretty decent. There were only a couple spots where I noticed the "fat-thin" pixel issue of software stretching. Overall sharpness looked acceptable, though not as good as an unstretched image of course. I might go with slightly less scanline intensity as well.

The fullscreen 1050 virtical res squared image revealed the issue I was discussing earlier of how the scanline effect just doesn't look right when it doesn't match the native res. In this case, there was noticable color banding issues, where various parts of the image appear lighter or darker. I noticed it right away and was disturbed by it when I previewed the image.

Finally, the fullscreen stretched image had the same issues and is a combo of the above two results.

Its pretty much like byuu said, where you need an insanely high resolution to mask the mismatch of the scanline effect when it doesn't match a multiple of the native LCD res. This was one area where CRTs were better for fullscreen displays.

[FirebrandX]

Oh just to make a few points clear about the FInal Fight pic Fritzy posted:

1. Something is terribly wrong with that pic. A scanline filter does NOT fuck a pic up like that, so there's gotta be something else going on there.

2. I don't use a scanline filter just to try and get rid of "stair-stepping". I use it to reproduce how the original hardware displayed game content. Its not about making it look worse (which some people are actually into doing with boob-tube filters). Scanlines were how game technology was displayed for decades. This goes into point 3....

3. Areas of solid color only look that way now on unfiltered images because we have the benefit of new display technology, so they were NOT intended to look that way in the first place. Each horizontal line was sent as a SCANLINE through the CRT. When they designed these games, they NEVER expected it to appear as a "solid swath of color" without any scanlines. Claiming otherwise is looking at it from a biased perspective.

4. When it comes to CPS games, again I prefer ZERO stretching ("squishing" actually in this case) and only scale in multiples on an LCD image. The problem is the native res of a CPS game is quite wide at I believe 384x224. Only when I had a CRT would I adjust the display (using my video card of course) to fit a 4:3 frame. Now that I am on LCD, There's no way to correct the apsect ratio without severely altering image quality. I've gotten used to playing CPS games unsquished, and now I rather like it. You basically get to play Street Fighter in a high-def format without having to alter the image, which has its own merit in that regard.

5. All that being said, here is a sample pic of how I like to play MAME games on my display:

http://www.firebrandx.com/downloads/finalfight.png

Basically just scaled up with a scanline filter and that's it.

[_willow_]

Cant draw the idea out of pic. Just serious pain to eyes

[_willow_]

Here is my Hybrid scale concept, very much scanlines folk friendly.

imagehosting.gr:


imageshack.us:

http://img145.imageshack.us/img145/2736 ... ridwc8.png

(no NTSC TV gamma ramp applied so it can be somewhat pale for you)

[kick]

@ _willow_ : Please use a better (browser-friendly) image hosting service such as Shup or ImageShack for uploading your screenshots.

@FirebrandX : That's not really a scanline filter,but a mesh filter. Scanlines are usually much thicker and blend with the pixels.

There's no way to correct the apsect ratio without severely altering image quality.

...unless you have an LCD display with a relatively small pixel pitch (< 0.25mm)

[FitzRoy]

... it makes areas that are supposed to be contiguous swaths of color ...

I'm just going to ignroe the personal attack and jump on that word "supposed".

It is SUPPOSED to have those dark lines. It was DESIGNED to output to a low-resolution CRT monitor with widely separated scanlines.

Why can't you see the fallacy in your shotgun approach? Not all CRT characteristics affected the way the art was designed. That's like saying PS3 games today are designed for motion blur just because it exists on today's televisions. In reality, it is an unavoidable deficiency and its successor (OLED) exists in part to overcome it. No game has ever used it to its advantage, it's just a blotch on the era.

While it's true that certain tube characteristics can affect art design (blending, overscan, painted pixels), few of those affected it positively and I certainly doubt scanlines affected it at all. No artist would want pinstripes on a patch of blue sky, it doesn't make any sense. And although we don't have any old game designers at the moment to ask directly, we can see how such effects have carried over into this era, and they haven't. The first LCD projectors were reviled for visible pixel separation, also known as the "screendoor" effect. And nothing but nostalgic mimicry seems to be using this effect in an attempt to "enhance" modern games of the same resolution.

It's also supposed to be a 4:3 image

According to the same logic you tried to use against me, not it isn't. Actual CRT televions deviated slightly from this and byuu and blargg believe that artists accounted for it based on the fact that that's what the technology did. They also believe that artists intend for the same art to be stretched differently on NTSC and PAL based on what the technology did. This was disproven by

(a) FirebrandX's triforce comparisons.
(b) My logical assertion that the degree of difference would have been too small for artists to recognize over standard at that low of a resolution.
(c) My logical assertion that art, being comprised of shapes with set proportions, cannot possibly be intended for two different corrections.

This was another perfect example of technological "accuracy" coming before the accuracy of intent. Is this some kind of revelation for you that in some regards emulation can achieve greater precision and consistency than what was possible in the past?

and something seems to have gone VERY wrongin the upper-right, as I'm pretty sure that sea of random black dots ISN'T supposed to be there, in code OR in the assumed final output device.

Just ignore that, it's an issue with the Intel IGP on the computer I was using.

[blargg]

I haven't been following this argument really closely, so I might be missing some of the points.

... it makes areas that are supposed to be contiguous swaths of color ...

It is SUPPOSED to have those dark lines. It was DESIGNED to output to a low-resolution CRT monitor with widely separated scanlines.

Why can't you see the fallacy in your shotgun approach? Not all CRT characteristics affected the way the art was designed.

It's hard to say what aspects, known and unknown, affected art design. If the artists were displaying things on the target system and display, any aspects could have influenced the design. If they designed it on a high-end workstation and never displayed it on the target system, it's less likely to have taken the target system into account, as the artists would have had to intentionally design for it and know how it affected the appearance.

That's like saying PS3 games today are designed for motion blur just because it exists on today's televisions.

If the games are tested on such televisions, they may very well be. By this I mean that the game might be showing things that wouldn't look as good on a TV without motion blur, but these don't get noticed because their TVs have motion blur.

While it's true that certain tube characteristics can affect art design (blending, overscan, painted pixels), few of those affected it positively and I certainly doubt scanlines affected it at all. No artist would want pinstripes on a patch of blue sky, it doesn't make any sense.

But eliminating the spaces between scanlines has more of an effect than simply eliminating the pinstripes. Removing them makes things more blocky, less rounded. If one has limited resolution, non-blocky pixels are more natural. Similarly, eliminating the blending on systems without lots of colors and high-resolution results in less-smooth objects and shading.

[...] we can see how such effects have carried over into this era, and they haven't.

Lots of good gameplay elements also got scrapped when 3D came around, so something not being carried over isn't necessarily a good thing.

I don't think people are arguing that one would intentionally choose low-resolution with scanlines over high-resolution without them. I think they are arguing that the graphics at that time had scanlines, and rendering it without them causes issues.

It's also supposed to be a 4:3 image

According to the same logic you tried to use against me, not it isn't. Actual CRT televions deviated slightly from this and byuu and blargg believe that artists accounted for it based on the fact that that's what the technology did. They also believe that artists intend for the same art to be stretched differently on NTSC and PAL based on what the technology did.

I don't think I ever said much about the artist's intentions, which aren't much relevant when the goal is reproduction of the gaming experience.

There is a fine distinction between an artist's intention (which probably can't be precisely actualized) and his choice among the actual options. If an artist is designing graphics for a PAL game and displaying it on the target system, he is making choices among actual options the system can do, like making an object 10 pixels wide versus 14. He might not even be looking at the pixel counts, and simply the proportions. By displaying on the target system, he will be taking all its aspects into account, since he's looking at the end result, not logically working things out based on a mathematical model of the system. He makes choices within the constraints of the system, but his intentions are still only approximated. Both things, his choices and intentions, are an expression of the artist. The first is represented exactly (unless someone else overrides his decisions), and what emulators are trying to reproduce.

For PAL and NTSC, I imagine in most cases the game is designed for NTSC, and later ported to PAL with little attention given to the graphical proportion changes that occur. If a game were designed from the start with NTSC and PAL in mind, the artist might choose one set of graphics for both.

This was another perfect example of technological "accuracy" coming before the accuracy of intent. Is this some kind of revelation for you that in some regards emulation can achieve greater precision and consistency than what was possible in the past?

I think Mega Man is a good place to start if we're trying to reproduce intentions (attempt at humor here).

[byuu]

We can argue about anything for pages here, hm? :/

Actual CRT televions deviated slightly from this and byuu and blargg believe that artists accounted for it based on the fact that that's what the technology did. They also believe that artists intend for the same art to be stretched differently on NTSC and PAL based on what the technology did.

I don't purport to know what artists' intent was -- only the original artists know that. I doubt it even matters, as it's going to vary on every TV set, size, color setting, etc. I do know what the average pixel aspect ratio NTSC and PAL televisions use, though. I also know that the number of RGB phosphors per scanline vary based on size, but the number of scanlines is constant.

This is why I have the scaling set up like I do now.

I also account for all tastes. Want it max scaled? Set it to scale max one time, done forever. It even keeps your setting across releases. Want an integer scale? Set it to that. Want fast access to change them constantly? You can do that, too. The menubar is hardly cluttered as it is now.

Want pixellated graphics? Use the point filter. Want bilinear? You have that, too. How about software? Want to make up detail? Use HQ2x. I always play Zelda 3 like that because it looks amazing. I'd never play Star Ocean with it. Want something a bit lighter? Use Scale2x. Want scanlines? You got it. Want to simulate TV shimmering, if only to see how your game will look on a CRT television? Use the NTSC filter.

Everyone gets ~90% of what they want (sorry, "I want HQ2x+scanlines" guy; and I still want that SoM Photoshop filter made into a software renderer), everyone's happy. Who cares?

We spend so much time fine tuning the UI, when I really should be working on the BS-X base unit Matthew gave me; using the serial cable blargg made me.

1. no menu or statusbar toggles, no menu or statusbar in fullscreen mode
2. fullscreen mode to be configurable without being in it.
3. space to have multiple, descriptive aspect ratio correction options, which the menu is unable to provide
4. space to explain that filters are hardware vs software, which the menu is unable to provide.

byuu knows these are valid arguments

1. I can do now, we'll make it default to hidden menu+status in fullscreen mode.
2. Since it's not true fullscreen, it's never going to bug out. And if it didn't work anyway, it wouldn't matter. Sure it'd be nice in the UI too, but I don't want to duplicate it and I prefer the menu-based approach.
3. The only time I've seen someone adjust this was because they were using 1280x1024 on 4:3 CRTs with its bastardized 5:4 pixel size, and people who want to stretch the image widescreen for some bizarre reason. Advanced is a good place for such ridiculously uncommon options. With the new UI and a QScrollArea, we can put the listbox on that page. Or on the video page. Not a big deal.
4. The menu can provide it with a QStatusTip, like it explains the point/linear now. We can tack on hardware/software filter: at the beginning. Or we can separate them into their own named menus. No, it's not as good, you're right. But it's faster.

Again, my usage pattern is to change these settings a lot. I don't want to go into the config panel to do it. Try using v017 again with its profiles. It was a pain in the ass compared to what we have now.

[Panzer88]

I'd have to agree with Gil here, arguing against artist intent is silly, it's like saying "we can render it in the systems hardcoded aspect ration instead of rendering it at 4:3 like they made it to be output on televisions years ago"

that doesn't make sense, I remember reading a magazine about developers using scanlines for transitional colors, it was a sega genesis game but I can't remember which, I'll try to look it up. Personally I don't use scanlines, but that doesn't mean I don't get the concept of recreating the format that developers made games for.

Personally I am more of the mind of enhancing games through emulation, like in polygonal games rendering in higher resolution etc. but there is definately something to be said for not losing things along the way. Hence simply wanting to run it at original resolution rather that blowing up and filtering the hell out of it.

If someone had a 25,600x16,000 19" monitor, you wouldn't really notice.

working on it

[tetsuo55]

I am glad that CRT simulation i being discussed again.

The examples posted miss some very important things, specifically kell-effect.

This image was made by blargg some time ago. It simulates all the important parts of CRT.

http://4.bp.blogspot.com/_ylyxHOid874/S ... 5jz0nt.jpg

You can read the blogpost here:
http://security-emulation.blogspot.com/ ... ation.html

[h4tred]

We can argue about anything for pages here, hm? :/

Mmmhm byuu, I guess that puts the nail in the coffin for my development ideas. Since people can rant on about how thier software filters are evil, I cannot begin to fathom the absolute outroar if I go ahead and instead rewrite the filter system to operate entirely on the GPU. (considering hardware graphics programming is my specialty )

Least that idea completely avoids licensing issues.[/code]