/vr/ - Retro Games

I like HLSL in MAME, but I can never get it to look right. Copying and pasting other people's configs never looks like their screenshots, either. Sometimes I'm convinced it's broken somehow.

Output to a CRT monitor, of course.

>>1594175

>>1594192
What the heck?

I like the Halation CRT filter in Retroarch.

My favorite filter is the 4chan filter. Goodbye horrible filter threads!

bump

>>1594409

Indeed. These emulation shill threads are getting out of control. Within just a few posts it's clearly someone spamming RetroArch again.

>>1594192
Why can't you people resize your images to the correct proportions? It's not difficult and it takes maybe five seconds.

>>1594171
None. They either make the image blurry or eye-strain inducing.

>>1596657
Well, it's what's being output to the screen.

>>1594403
480p with every other line blanked, to look like a (very sharp) television displaying 240p.

Large number of pixels/line with a high dotclock for emulating NTSC artifacts, I'd assume.

>>1596757
Nah, the high horizontal width is solely to hide scaling issues that would normally arise from correcting 8:7 to 4:3.

>>1596762
Get away from me, your autism might be contagious.
>implying the snes even outputs correct 4:3 anyway
>implying most games didn't have crazy inconsistent overscan

>>1596762
>hide scaling issues that would normally arise from correcting 8:7 to 4:3.
This doesn't make sense. If you send the 8:7 image to a 4:3 CRT monitor it will be displayed correctly.
I think the real reason is that the video adapter has some minimum fillrate that would not be achieved with real 240p.

>>1596780
Maybe he wants to play games of various widths without changing video modes.

>>1596780
When I said "correcting 8:7 to 4:3", I meant doing so within the emulator. You can, of course, adjust the image through the monitor controls to make it 4:3, but then you'd have to adjust it again for games that output 320x224 or whatever. With a superwide resolution, you don't need to do that. It works for every game and every horizontal resolution, without needing to adjust the monitor or make a bunch of separate modelines.

>>1596780
>I think the real reason is that the video adapter has some minimum fillrate that would not be achieved with real 240p.

"Real 240p" is achievable as long as you use a refresh rate of 120Hz. However, this method is flawed because you're then limited to only 240p, and if a game switches to 480i mid-game, it gets downscaled and you lose half the picture. Not to mention all the issues with frame-doubling, needing to use black frame insertion to compensate, etc.

This is why 480p plus a scanline/interlacing shader is superior. It can handle all these cases.

The ones that don't make my games look like fucking shit. In other words, none.

Fuck the police, this makes games look better

>>1596964

So what do you do about games that were meant for 4:3 but had non square pixels due to being rendered at 8:7 or 10:7 or 12:7 ratios?

>>1597010
That ain't doing much for the HUD, but the sprites themselves look like cartoons, which is neat.

>>1596526

>Within just a few posts it's clearly someone spamming RetroArch again.

Where did this happen?

>>1597047
Well somone started posting shots from retroarch. I wouldn't really call it shilling though, unless you think that anyone mentioning what emulator they use is "shilling"

What the hell is frame doubling, anyone care to explain?

>>1597094
Repeating frames to display a 60hz game at a 120hz video mode. Ruins the low-persistence motion quality of CRTs.

What are some of the best emus for nes, snes, and genesis? Also, what's the best way to emulate those 3 on a dreamcast? I've seen some vids of other emus and the sound is off, and a few the colors are off.

>>1597010
That looks absolutely awful, do people actually use vilters like this seriously?

>>1597010

>>1594175
>>1594401
>>1598142
Damn you, now I want to play Another World again.

>>1597010

>>1594171

No one has really made the perfect one yet. There just is not enough resolutions in our displays. Once we get high enough resolutions we can make shadowmask like shaders, which should be pretty awesome. Until then, we just have these sort of hacky half measure CRT shaders. But they really are taking so many shortcuts. They are not that good. And people are put off by them since they don't resemble the CRTs that they grew up with (shadowmasks) but instead apertrure grille displays.

So how much resolution is needed for shadowmask CRT shaders? Who knows. Maybe 4K isn't enough.

>>1598117
>What are some of the best emus for nes, snes, and genesis?

See:

http://emulation.gametechwiki.com/index.php/Main_Page

>3 on a dreamcast

I don't recommend that. The Wii is the go-to system for emulation of 8-16 bit consoles. It can do it very, very well and connect to a CRT display at 240p. It's really the best.

>>1598129
>do people actually use vilters like this seriously?

There is in fact a small subset who uses smoothing shaders. Most people dislike them though.

>>1594407

what game, what settings? that doesn't look like default.

>>1597116
>Ruins the low-persistence motion quality of CRTs.
You're supposed to use black frame insertion to preserve motion quality if you're frame doubling. The real drawback is the reduced brightness.

>>1596964
>The ones that don't make my games look like fucking shit. In other words, none.

Nearest neighbor is a kind of filtering. It does alter how the game looks a lot.

Games on original hardware + CRT

Many, many, many people have grown up with emulation and think that nearest neighbor x4 is how old games "are supposed to look". I heard of someone's kid referring to SMB1 as "Lego mario" because Mario looks like he's made of legos. On a CRT and an NES that would look like a smooth sprite. Well, less blocky at least. But scaling with nearest neighbor on an LCD means old low resolution games look super blocky.

>inb4 people come in and say the shot on the right looks awful

It's blurry, but this is if you were to have your nose to the screen. The blur is only noticeable there. If you sit at a regular viewing distance it becomes sharp. So you get rounding of the sprites.

>>1598205
Or you can just use 480p with interlacing.cg. No need for 120Hz or black frame insertion then.

As for reduced brightness, find a monitor with a SuperBright mode.

>>1594171
filter my ass

>>1598209
You could not be more delusional if you tried. This air of "I know what the developers intended for the game to look like because I had a shitty CRT when I was little" has really gotten stale. Nearest neighbor is not a filter you lunatic.

I don't use filters because I like my games to look how they were drawn.

>>1598218
See, the thing is, what many people refer to as "filters" are actually better known as "scaling algorithms" or "scalers". To give a few examples, nearest neighbor, bilinear, bicubic, lanczos, and yes, even "smoothing filters" like 2xSaI, HQ, xBR, etc. are all scaling algorithms of some sort. Only certain algorithms are solely intended to warp and modify the image to achieve some kind of visual effect. Most of these are pixel shaders.

>>1598225
>I don't use filters
>proceeds to post a screenshot using a bilinear/blur filter

Then again, bilinear filtering is so ubiquitous that rarely do you see any shitstorm arise over its use, to the point that I dare say most people do not even consider it a filter. I've seen countless people post screenshots using bilinear filtering, and no one ever goes apeshit over it despite it blurring the shit out of everything, especially at high resolution. But god help you if you post a screenshot using xBR.

>>1598236
That image has no filters you idiot. I don't even understand how one can make that mistake.

>>1598247
What is you definition of an "unfiltered" image? Because I see obvious blur around the pixels, which indicates to me that it is using at least bilinear interpolation. Or does bilinear/blurring not count as filtering?

In any case, it's a moot point. A truly unfiltered/unscaled image is as follows.

>LQ2x
>HQ2x
>2xSAI
ideal filtering

>bilinear OFF
bilinear filtering = shit

Any upscaled image is filtered, period.

>>1598256
It's not unscaled, because you've changed the aspect ratio. Some filtering is unavoidable for NES/SNES games if you want to display them correctly on a modern LCD monitor.

>>1598675
Integer ratio nearest neighbor scaling is not filtering, it's simply changing the pixel size. There is no official standard for number of subpixels per pixel. On CRTs you even get different numbers of subpixels at different parts of the screen.

>>1598209
Then the solution is to use Gaussian blur and not sit close to the screen.

>>1598689
It is filtering as much as bilinear is filtering because you're increasing the number of pixels that make up the image, simple as that. And it does have an obvious effect on CRTs where upscaling 2x or more with nearest looks blockier than it does unscaled at native resolution

>>1598689
> it's simply changing the pixel size

By adding more pixels in a simplistic fashion, i.e being filtered.

Filters forever. Eat it bishes.

Use the RetroArch
xbr-dtt-dilation-soft2-aa-gamma.cgp

>>1598459
bilinear is good for post-filtering an already integer scaled image to a non-integer scale to reduce aliasing artifacts.

>>1598707
No, the pixels are literally bigger. Double the size and the pixel is made of 12 subpixels instead of 3.

After being a maximilist when it came to this kind of thing, I realize now the best way to filter things is to make them look like they would/could have at the time. The biggest piece of advice I can give is STEP BACK FROM THE GODDAMN MONITOR.

NES/Genesis/80's: composite+billinear
game boy: subtle grid
gbc: LCD filter
N64: render at 640x240, scale to 320x240 w/s-video
SNES/PS1/90's: simple2x+billinear (native res)

Scanlines optional, but should be subtle if present.

>>1598729
That's just the end effect, it is still a filter on a factual level

>>1598763
By your logic changing the resolution displayed on a CRT is applying a filter.

>>1598715
It looks like your scaling is broken anyway. It's hardly a fair comparison when it looks like you shit over the source.

>>1598256
He's just being an asshole troll. You are correct and that image is filtered.

>>1598813
It is if you're upscaling

>>1598813
You add more pixels to an image using a filter to upscale the image to fit a larger resolution, how hard is this to get?
Changing resolutions on a CRT has nothing to do with this

These sort of shitstorms are always stupid. Yes, playing the games with a filter looks better than pixelated upscaling, but no, filtering doesn't look even close to the real thing unless you go to extreme lengths like that ultrawide thing someone posted the other day.

>>1598840
And, I should add, if I read that thread right, they were using high resolution CRT monitors, not LCD screens. Which is definitely going to great lengths.

>>1598831
There is no rule saying a pixel has to be made of 3 subpixels. My phone has only 2 subpixels per pixel. If 12 subpixels are updated as one logical unit then they form a single pixel.

>>1594407
I use a CRT filter, but I don't remember which one is it. I use one for NES, then a different one for SNES. I haven't tried a Genesis game yet.

>>1598850
They're not "one single pixel", it's 12 pixels that are the same color reproduced from a single pixel using nearest neighbor interpolation. Stop talking about "sub-pixels", there's no such thing in a raw digital or analog images, that's an element of physical displays.

If it's being upscaled, some kind of interpolation filter is applied, and that's a simple fact.

>>1598898
Why do 2 subpixels make a single pixel, 3 subpixels make a single pixel, approximately 5 or whatever you get on a CRT make a pixel, but not 12?

And if I can't talk about subpixels then your argument is non-existent.

>>1598885
fuck CRT filters that doesnt look like a CRT at all. Those fake ass scanlines only darken the image and on an already darker LCD screen it looks like fucking shit.

>>1598905
Physical displays have different ways of representing a single pixel, doesn't mean nearest neighbor interpolation isn't a filter like you are claiming. I don't know why you keep insisting that, you must be associating "filter" with "horrible smudging of pixels" and hate the idea that "glorious pixel perfect nearest neighbor" is also in fact a "filter".

>>1598926
There's better ones that compensate by brightening and saturating colors and making brighter pixels bloom resulting in variable width scanlines

>>1598963
My physical display is representing a single pixel with 12 subpixels. There might have been nearest neighbor scaling involved, or maybe I hacked the display hardware to switch more subpixels from the same input. There's no way you can tell because they look literally identical.

integer scaling is not merging or altering pixels, only doubling / tripling / etc them

it is not filtering

>nearest neighbor: add pixels to image to make it bigger
>other filters: add pixels to image to make it bigger

What's the difference?

>>1599038
Nearest neighbor produces a result visually identical to replacing the display hardware with something with a lower resolution.

>>1599038
Some "filters" aren't there to scale the image, though. For instance, there's various shaders whose whole purpose is to achieve some kind of post-processing effect, such as color or gamma correction, adding bloom, anti-aliasing, etc. It's why the word "filter" is kind of useless and undescriptive.

Nearest neighbor is a scaling algorithm, just like bilinear. Meanwhile, FXAA is a post-processing effect. You don't use the latter to scale an image, but rather you use it AFTER scaling to achieve some kind of result to be output to the display. The distinction has to be made, and the usage of "filter" as a catch-all term needs to stop, as it just confuses people.

>>1599046
Until non-square pixels are involved then it becomes an aliased uneven mess. Best upscaling is one that does minimal blurring to hide unevenness due to non-integer scale factors

what looks better, your black and white flat gameboy emulator, or this shader thats a hell of a lot closer to a real gameboy?

>>1600190
The green looks like complete shit, give me monochrome.

>>1600190
Super Game Boy looks better.

>>1600190

>>1600190
C.) Game Boy Pocket shader

>>1598217
>filter my ass
wish I could

zing!

>>1600307
>playing it without filters

Glorious filter race incoming!

>>1600368
Your aspect radio is all wrong.

jesus, I've been working my ass off the past few days to improve the clarity of captured SNES video and you guys are actively trying to make it blurrier

I posted this on Emulation General earlier, I'm going to go ahead and repost this here.

I made some cg presets for RetroArch with the NTSC shader combined with a CRT shader:

http://www.mediafire.com/download/jo62id6p2hrm2pr/ntsc+crt.7z

And one for real CRT monitors:

http://www.mediafire.com/download/z4jeto62tn57cly/ntsc%2Binterlacing.7z

>>1600376
I forgot about the radio, I'm sorry.

>>1600392

Well emulators already have perfect clarity by default so it's kind of boring really, you actually have to work in reverse to get convincing composite video artifacts.

>>1598192

>what game, what settings? that doesn't look like default.

Game is Strider 2. The settings are default, except I removed the curvature. I'd remove the black border too if I could figure out how.

>>1600190
>>1600307

Mine looks better.

>>1600190
>what looks better, your black and white flat gameboy emulator, or this shader thats a hell of a lot closer to a real gameboy?
the emulator, since the real gameboy screen was ugly as sin

>>1598970
Like?

Shaders are controversial becuase there is no perfect crt shader. If they made one that was just like an actual CRT (shadowmask) then everyone would use it and no one would complain.

>>1600804
lol, no

I happen to like using LCDs. It's really nice not having to deal with the fuzzy garbage of a CRT anymore.

>>1600609
Yours looks more like GBC, his looks more like GB.

>>1598459
>LQ2x
>HQ2x
>2xSAI
>ideal

>>1600406
>have perfect clarity by default so it's kind of boring really,
You mean it's kind of displayed incorrectly.

This is kind of like trying to replicate an artists work. Anti-filter fags are bitching to print it poster paper it's clear and it let's you see all the edges clearly which is how it should be. Filter fags are suggesting we should put it on canvas using actual paint because that's what the original was.

It's not an issue of boring it's an issue of not replicating what the image was and making it look worse for it.

>>1601403
>It's really nice not having to deal with the fuzzy garbage
>uses fuzzy garbage of an LCD

>>1602139
No, filter fags still print it on paper, they just shit all over the paper in an attempt to make it look older.
It's like removing grain from movies because they weren't present in the studio. They may be an artifact but you aren't going to get rid of them without sacrificing intended details.

its like those little faggots who burned the edges of their projects to make them look old.
what kind of a fuckin paper burns perfectly around the edges

>>1602128
No, that looks like a Game Boy Pocket. I should know, I have one, along with an original Game Boy. The Pocket is superior in every way.

>>1598209
>I heard of someone's kid referring to SMB1 as "Lego mario" because Mario looks like he's made of legos.
That's how he looked in the box art.

>>1602148
Books that have been in a fire, but not wholly consumed.

>>1598209
I still think the left image looks better than the right image. It looks cleaner. I like cleaner. And I owned and played AoL back when it was first released.
The right side looks icky and bad.

>>1601403
>It's really nice not having to deal with the fuzzy garbage of a CRT anymore.
And yet you're using a flawed panel technology with a disgusting amount of motion blur and a fixed resolution + refresh rate.

>>1600609

I remember helping the author of that shader extract the colors from some pictures of my Game Boy Pocket I took. One thing that's lacking from that shader is the "grainy" appearance of the reflective back layer, the author tried to replicate it with a background but it seems like you would need an insane resolution to make that look convincing.

>>1601403
I think you misunderstood me. You said there are better CRT filters. I was asking 'like' as in "like, which ones are you using that " brighten and saturate colors and makie brighter pixels bloom resulting in variable width scanlines". I don't know shit about retroarch.

NTSC_X2 or something for NES emulation.
Using the fceux emulator.

>>1603569
Hyllian's CRT shader is the best one. It is very easy to adjust to suit your tastes

https://github.com/libretro/common-shaders/blob/master/crt/crt-hyllian.cg

dot matrix master race

>>1603741
That looks like it has compression artifacts.

>>1604529
That's not a dot matrix. That's a regular old low-res LCD.

>>1604529
how did you achieved this? I want to play Mario Land 2 and Link's Awakening like this.

>>1603741

NTSC needs horizontal linear filtering or else it looks badly aliased

>>1602139
>Filter fags are suggesting we should put it on canvas using actual paint because that's what the original was.

No, those are CRTfags.

Filterfags want to reproduce the painting on poster paper, after applying a canned 'brushstrokes' effect in Photoshop to make it look extra painterly.

>tfw when i spend far more time tinkering with filter stacks trying to make games look good than actually playing them

>>1598209

Idk what kind of shitty TV's you used to have, but no CRT will make mario look like a "smoothed sprite". I have a shitty CRT from the early 80s that still produces some surprisingly crisp sprites.

And all of that shit about the developer creating the sprites with the displacement of graphics from CRT's in mind only applies to some forms of fine ass texture dithering.

Also

>>1597010
Jesus fucking christ how horrifying.

>>1597010

>Using filters
>Playing the worst SF3

Who'da thunk, huh?

>>1602148
>its like those little faggots who burned the edges of their projects to make them look old.

AHAHAHAHAHAHAHHAHAHA

We actually had a project in high school where we were forced to write with a feather and shit; we also did that too.

Best analogy ever

>>1610058
I was like this as well. Then I got a CRT monitor and discovered superwide 480p and interlacing.cg, and that put an end to that.

>tfw finally satisfied
>tfw actually playing games for once

>>1598715
and scaled to to 16:9. good job

Whats a good shader for higan on the snes? is there a pack?

>>1610186

>>1610190

>>1610192
>this is real
http://research.microsoft.com/en-us/um/people/kopf/pixelart/supplementary/comparison_super2xsai.html

>>1610165
Anyone?

>>1610201
>>1610192
>>1610190
>>1610186
My god that shit's ugly

>>1610229
Anything looks like dogshit zoomed in that close.

>>1610192
Is Peach having a gizgazm there?

>>1610087
Same, but I've been playing around with different horizontal filtering methods (240p xBR, NTSC, blend shaders)

Then again I play around with shaders because I'm interested in what it will look like, not because I'm "not satisfied" with what I'm using currently.

>>1610917
My final setup basically applies gamma correction first (from 2.2 to the 2.4 that TVs supposedly used), then prescales horizontally by 2x nearest, then applies linear horizontally the rest of the way. This way you get slightly rounded pixels, and it avoids the picture looking like nearest neighbor with black lines, without looking like blurry shit.

Sometimes I use a CRT filter for the hell of it. Just for a bit of nostalgia. I give zero fucks about what the developer intented, because no one fucking knows what the developer intented. And if I contacted them about it, they'd probably say "just play the fucking game you pedantic asshole"

And honestly, if people want to use shit filters that make their games look like a mess, I don't care. I don't care if they fuck up the aspect ratio and stretch the emulator so it covers their whole widescreen monitor.

I don't see the point in getting anal about how another person plays video games.

>>1598459
>not using a dithering smoothing filter like mdapt as well
full pleb.

>>1610201
>vector art
>the filter

>>1612185
>actually thinking this looks good
nigga u dumb

>>1612517
How does that not look good?

>>1612517
Gross.

We need a filter for handheld emulators that simulates dust particles that get between the screen and the glass over time, as well as dead pixels, dead rows, and dead sections.

>>1598459
are there filters for higan?

>>1612716
They were removed along with just about everything else back when it was called bsnes, use bsnes in RetroArch instead.

>>1612721
but inst the newer version better? also what is this RetroArch thing about about? can it play everything? does it have all the best emulators or something?

>>1612780
Its emulation has improved but the UI was drastically simplified, and now only offers a bare minimum of features. It also requires you to import ROMs into a weird folder setup that doesn't work with anything else. RetroArch is sort of a frontend for different emulation cores, including up-to-date bsnes ones. It sometimes works better than the original emulators (bsnes for example) but not always.

>>1612721
Higan has many cg filters and more are posted on byuu's forum. It's been that way for several versions now. Funny how he's gone completely 180 on that philosophy since bsnes.

>>1612821

No, it does not and cannot work better than the stand-alone emulators. It's simply a front-end. It does not change the code of the emulation cores, which is something that would be necessary to improve upon their performance in the first place. It's not a "sort of" front-end, it's a front-end. Period.

Don't waste your time with it. Just download and use the actual emulators. They all have their own native GUIs, and this often means you get more configuration options (and easier access to these options) than you would by using a front-end. Front-ends enforce their own configuration parameters. They don't always permit you access to everything in the emulation core that can actually be tweaked. RA is particularly bad about breaking things for this reason.

>also what is this RetroArch thing about about?

It's a front-end that sucks, and that RetroArch's developers keep spamming on /vr/ for some inexplicable reason. Ignore them.

>>1612835
I cannot say frontend because it's far too tightly connected for that, closer to a plugin interface. Also yes it can do better, and does in respects to syncing and netplay, and having a UI designed for controllers, and portability far better than many of the original emulators.

>>1612835

You have no idea what you're talking about, literally nothing you've said is true.

>doing this in every emulation thread

Holy shit just kill yourself already

>>1612835
>No, it does not and cannot work better than the stand-alone emulators.

Yes it can, when the standalone emulator's frontend is poorly coded and does things like vsync in a suboptimal manner. In the case of bsnes/higan it is objectively better with RetroArch than with standalone higan simply because it doesn't force game folders down your throat, and it also does syncing better (dynamically adjusting audio rate instead of statically like higan does). RetroArch has its own methods for blitting video so it definitely can perform better if the standalone frontend was unoptimized.

>It does not change the code of the emulation core

Often times they do make modifications. bsnes is modified to load normal ROMS, Mednafen cores have some modifications, Mupen64plus is a fork that has been modified extensively for portability, performance, and some extra accuracy in regards to texture filtering. Some emulator cores have a shitty internal API (which has a tendency to change every release and break ABI) that may need hacking or modification to make it work with libretro.

>and this often means you get more configuration options

9 times out of 10, most of these options are unnecessary or even detrimental cruft from older versions. Some of the options aren't even part of the emulator, just its frontend. The important options can be exposed in Core Options and everything else can be auto configured to best settings for a particular game.

>>1612835

>>1614242

VBA-Next runs even faster and doesn't have that slightly annoying vertical pixel shift

>>1612835
>and that RetroArch's developers keep spamming on /vr/ for some inexplicable reason

http://pastebin.com/sebtNbUw