I haven't looked at the code but the way features get into DOSBox is if there is an actual need for the feature. If there is no need then it does not get implemented.

It's hard to say exactly what the average monitor size that DOSBox users have but going by Valve's hardware survey most people have 1024x768 size monitors. This may be one among many of the reasons why 4x hasn't been implemented yet.

I'm sure the devs have a huge list of what needs to be in DOSBox and does not. The things most on top of the list are compatibility....extra features are towards the bottom. Extra features are usually implemented around the time when the official versions come out but only usually because a LOT of time has been spent on testing and finetuning those features.

DosFreak wrote:

It's hard to say exactly what the average monitor size that DOSBox users have but going by Valve's hardware survey most people have 1024x768 size monitors. This may be one among many of the reasons why 4x hasn't been implemented yet.

well that's just valve's denizens
it didn't stop zsnes from having hq4x

I would really like this to be added, would help with the crappy ATI drivers. 4x would be nice for full screen doom/build games

why not use fullscreen in the first place then?

and how long has zsnes been out? Compatibility for NES/SNES games ceased to be an issue a long time ago. It could be argued that we're pretty close to the point where that's the same for DOSBox.

Anyways we all know that when 4x is implemented people will be bitching about why 8x hasn't been implemented because anyone that is a gamer and uses DOSBox runs their desktop at 2560x1600. 🙄 (By the way I run mine at 1920x1200) (I'm all for 4x scaler but as I mentioned above other issues probably have taken priority).

I am reminded of the old saying "In for a penny, in for a pound."

Your view is a bit narrow when you say that it takes little time and effort to implement. While you may think it's acceptable to have a 4x version of the simple scaler without a 4x version of the other 2x/3x scalers, not everyone will think the same. Perhaps you noticed during your copypasting that those other types of scalers are not so simple.

ripsaw8080 wrote:

Your view is a bit narrow when you say that it takes little time and effort to implement. While you may think it's acceptable to have a 4x version of the simple scaler without a 4x version of the other 2x/3x scalers, not everyone will think the same. Perhaps you noticed during your copypasting that those other types of scalers are not so simple.

Indeed I did. I could not begin to implement them. Also, it's very likely that my simple patch is missing something, and possible that it breaks something somewhere in some situations. Fortunately, I have no illusions of it getting into the codebase as-is.

But your view strikes me as odd, too. Having Normal4x does not take anything away from anybody. The old 3x advanced scalers will continue to function as before, and they'll surely eventually reach 4x. Only a spoiled child or a similarly immature adult whinebag would complain about a situation where someone else's needs are met before his/hers. They can (and should) be ignored. It's an imperfect world, and some features are inevitably completed before some other features are.

(Just to be clear, I'm not implying that you are a spoiled child. No insult intended.)

If someone doesn't have the patience to wait for a feature, then it's in the open source spirit that the impatient person produce a patch himself. That's what I did, so I can afford to shoot my mouth off for a bit. Now that I have done as convention dictates, my part has been played, and the issue is in more capable hands. If anyone whines about the other missing scalers, then they're free to try and implement them themselves, too.

Of course, if it's developer policy to only release new scaler multipliers when ALL scaling methods have them, then so be it. I might put that policy to question, but it's not my project. The devs shall do as they please.

(Okay, I lied about not beginning to implement them. I'm actually looking at AdvMame4x right now. Scan4x looks trivial. We'll see if anything comes out of this idle dabbling. Diffs will be posted, if I don't lose interest.)

Elamaton wrote:

http://elamaton.no-ip.com:8080/dosbox/Normal3x.png
http://elamaton.no-ip.com:8080/dosbox/Normal4x.png

I must be getting old - I don't see a f....cking difference except the mouse-pointer jumping back and forth when I flip between the images using Firefox 2.x on a 1920 x 1200 LCD screen.

Ehm... Elamaton, are you aware of Gulizoka's DOSBox build? It has a special output method "direct3d", which supports shaders. In the "Shaders" directory, there are files named (for example) "GS4x.fx", "GS4xHqFilter.fx", "GS4xScale.fx" and "GS4xSoft.fx". These files are relatively simple text files, so i'd say it would me much easier to implement a shader, than change DOSBox code. I have no idea how they work, though. The only thing i can tell is that it's easily possible to change parameters by editing these files (Gulizoka gave me an example when i needed a special effect).

MiniMax: buy a new monitor, or get a set of glasses (; . The 4x shot is "clearer" and not as blurred as the 3x one. The pixels are "sharper". I've used IrfanView, and the difference was subtle, but easily visible. Firefox might scale the pictures, if they're larger than the size of the FF window.

DosFreak wrote:

It's hard to say exactly what the average monitor size that DOSBox users have but going by Valve's hardware survey most people have 1024x768 size monitors. This may be one among many of the reasons why 4x hasn't been implemented yet.

Nothing against the rest of your post, but I call bullshit on that Valve survey.

http://www.steampowered.com/status/survey.html

Now, the most common Primary Display Resolution listed is 1280x960, not 1024x768 (although it does come second). Some 20% are running resolutions higher than that. That's plenty enough reason for 4x scaling to exist. That's not the point, however. The point is, the survey contains so many discrepancies and oddities that it's nearly worthless.

The most common 4:3 LCD resolution (1280x1024) isn't mentioned at all, while x960 is. Many CRT owners naturally use that resolution, but no sensible LCD user will (unless their display can letterbox the image). There are tons of 17" LCDs out there, and they should show. The only imaginable scenario is that the x1024 LCD users are rolled into the same result as x960 CRT users. That's just stupid.

Then, looking at the 4:3 aspect ratios, we see the dominance of the venerable 16" display. *cough* What the hell were these people smoking? I dare you to go out and find a 16" display. Jesus. Viewable screen area of 16" on 17" CRTs is no excuse; no one uses that metric in the real world. Similarly, the equally ubiquitous 18" displays are well represented. All in all, another blow to the credibility of the survey.

I mean, for the sake of all things holy, they combine 1280x1024 and 1280x960 into the same group - although their difference in the implied display type is significant - but not 16"-17" and 18"-19"? Idiots.

Almost 65 percent of people have 60Hz as their refresh rate. That's the default rate of just about all LCD displays. 75 comes second, and is probably divided among LCD and CRT users, being a reasonable CRT rate and the common secondary LCD rate. 60Hz must be nearly all LCD, since CRT on 60Hz just means headaches. (Some very specific Vsync demands are the only reason 60Hz CRT is even imaginable nowadays.)

It can be extrapolated that the majority of people are running LCDs. (Obviously. This is 2008, ferchrissakes.) And the last desktop LCD that could only reach 1024x768 was a 15" one, and they've gone the way of the dodo a long ago. Laptops with that resolution are more common, but laptops that can realistically run Valve games will almost certainly have a larger, probably widescreen resolution.

There's more, but enough ranting, since it's pretty off-topic anyway.

MiniMax wrote:

I must be getting old - I don't see a f....cking difference except the mouse-pointer jumping back and forth when I flip between the images using Firefox 2.x on a 1920 x 1200 LCD screen.

Yes, you probably are. I ran those by some DOSBoxing friends of mine and they all confirmed the improved picture quality which I also perceived. (Yes, I did want to rule out the possibility that the difference I saw was just psychological.)

ADDiCT wrote:

Ehm... Elamaton, are you aware of Gulizoka's DOSBox build?

Yes. Windows-only (I'm primarily a Mac user, secondarily Linux, tertiarily Windows), and an unofficial experimental build, however good, is still an unofficial experimental build. I like to keep it stable, steady and official in the long run.

(To that end, official MT-32 support would be veeery nice, so people not running the worst of the major operating systems could enjoy it as well. So what that it needs the ROMs which are of "questionable" legality, that's no problem of the DOSBox project since it wouldn't be distributing them. ScummVM has support, and its faaaaabulous 😉 )

Elamaton wrote:

MiniMax wrote:

I must be getting old - I don't see a f....cking difference except the mouse-pointer jumping back and forth when I flip between the images using Firefox 2.x on a 1920 x 1200 LCD screen.

Yes, you probably are. I ran those by some DOSBoxing friends of mine and they all confirmed the improved picture quality which I also perceived. (Yes, I did want to rule out the possibility that the difference I saw was just psychological.)

Hmm. I saved the images to my hard disk, put them in a temp folder, and used the View-as-slide-show in Window flip-flop between the images. No difference what-so-ever (except the mouse pointer) to my old, weary eyes. Anyone else see a difference?

The difference is in the interpolation at the edges.
Either way if somebody wants to play at big resolutions then he shall
use fullscreen which is why it's there in the first place.

Elamaton wrote:

(To that end, official MT-32 support would be veeery nice, so people not running the worst of the major operating systems could enjoy it as well. So what that it needs the ROMs which are of "questionable" legality, that's no problem of the DOSBox project since it wouldn't be distributing them. ScummVM has support, and its faaaaabulous 😉 )

Quite of topic. And the answer is that I don't like distributing a halfworking program. (e.g. something that needs something else to operate properly, especially if that is possibly illegal.)

And just get a real MT32. it works perfectly with dosbox (cvs version of dosbox also with a mac) and the emulation isn't on par with it yet.

plain scaling for 4x ? why not use windowresolution with a hardware scaler like openglnb or ddraw/overlay
for fullscreen i see currently no use for 4x scaling at at all.

Who knows what happens If one of us ever gets a monitor that big that needs 4x scaling. I don't expect it to happen soon 😀

wd wrote:

The difference is in the interpolation at the edges.
Either way if somebody wants to play at big resolutions then he shall
use fullscreen which is why it's there in the first place.

I used windowed mode here because it was easier to take screenshots of the final scaled image, and because windowed mode on an LCD doesn't do the ugly strecth to the LCD's native resolution. But still, fullscreen looks better with a higher scaling multiplier just as windowed mode does.

I just tested it. Here's how:

Native resolution of LCD screen: 1680x1050 (16:10), although for DOSBox pillarboxed 1400x1050
fullscreen=true, fullresolution=original, aspect=true, scaler=[none, normal2x, normal3x]
Game used: Tyrian (sticking with it 😉 ) at 320x200

Subjective results for scalers:

none with 320x200 stretched to native res looked awful.
normal2x with 640x480 stretched to native res looked better.
normal3x with 960x720 in the top left of a 1024x768 full screen with thin black bars stretched to native res looked even better.
normal4x, if it existed, would almost certainly look better still.

So: the higher the multiplier of the scaler, the better the image looks even in full screen.

Your statement of "playing at big resolutions" makes no sense. If I had a screen of 4000x3000 pixels, I'd certainly want normal12x to exist, regardless of my choice of windowing vs. fullscreen.

And the answer is that I don't like distributing a halfworking program. (e.g. something that needs something else to operate properly, especially if that is possibly illegal.)

I'll accept that you feel that way. I disagree completely, since nothing illegal would be required for DOSBox to function, given that MT-32 emulation would be completely optional. And there are legal owners of those ROMs too. But it's your call, and I'll rest my case.

And just get a real MT32.

Umm, no. Buying decades-old hardware and the necessary adapters to make it work with modern machines just to complement an EMULATOR instead of voicing support for that hardware to get, you know, EMULATED (or rather, the existing emulation, even if imperfect, to be implemented) would be silly. I thought the whole point of emulation was to dispense with old hardware, so I could play old games without owning old computers or their old peripherals?

Well, hopefully something like Gulikoza's build eventually finds its way outside Windows.

plain scaling for 4x ? why not use windowresolution with a hardware scaler like openglnb or ddraw/overlay

Because it looks like shit in comparison. I tested it, naturally. It's possible that there is yet some combination of config options I haven't tested that would rival what my normal4x patched version looked like at 1280x960, but so far nothing has come close.

for fullscreen i see currently no use for 4x scaling at at all.

See my previous post.

none with 320x200 stretched to native res looked awful.

Then use a different output and/or fix your gfx drivers.

Because it looks like shit in comparison.

Be sure you're not doing something wrong before stating stuff like that.

And there are legal owners of those ROMs too.

obviously you aren't and I guess there are 99,99999999% illegal owners of the ROMs...

Umm, no. Buying decades-old hardware and the necessary adapters to make it work with modern machines just to complement an EMULATOR instead of voicing support for that hardware to get, you know, EMULATED (or rather, the existing emulation, even if imperfect, to be implemented) would be silly. I thought the whole point of emulation was to dispense with old hardware, so I could play old games without owning old computers or their old peripherals?

Quite frankly, I had hoped for good Mt32 emulation myself but after ages of no further development I gave up and bought a real one. Look at the Mt32 forum, the emulation is far from perfect, some may even say far from good and the code hasn't changed in almost 2 years. If you need a built of Dosbox with this emulation then look at how to integrate the Mt32 emu with Dosbox. There is a description of this somewhere and there might even be a patch on Dosbox' SF site (or ask the guys who offer the windows cvs versions with it compiled in)

As for getting the devs to add anything you want to main Dosbox sources you definitely are going the wrong way, you are only hardening their views on such things...