/vr/ - Retro Games

>>10125043

Shiny glow

>>10125052
KINO

>>10125043
World 8 for me is when I usually stop having fun.

>>10125052
YOU GOTTA ASCEND NIGGA

>>10125043
The Final GP with the Dig Racing Team filtered the shit out of me.

>>10125329
ARE YOU NOT ENTERTAINED???

>>10125475

>>10125475
>>10125494
>check 'em

>>10125475
>>10125494
kek

>>10125545
My eyes are getting so bad that this is actually comfortable for them. Any bright backgrounds make them well up in tears nowadays.

>>10125575
It's not that bad for me, but if I stare at a bright screen for more than give seconds, it starts feeling uncomfortable, and if I close my eyes it's like I see a bright oval or something. It's why I now run my monitor at minimum brightness at all times.

>>10125052

I am new to /vr/, do why actually respect namefags here?

>>10125663
I typically despise namefags on any other board but the ones here post cool screenshots and webms so I'm okay with them.

>>10125663
I want to say welcome but the vast majority of population increase here over the past few years has been negative, so I won't.

>>10125052
Haha my TV had this problem back then. Nice

>>10125663
I didn't even notice, who the hell cares

>>10125579
what you need is dark mode. it makes sense for additive color spaces. dark screens and light walls is the way.

>>10125052
soul

filters:
>they mess up the pixels and look like shit

sharp pixel filters:
>too sharp and it's not "what developers intended"

CRT filters/shaders:
>doesn't look like the real shit, especially on motion

real CRTs:
>all the CRTs that you can pick up already come with some sort of problem just for the simple fact that it is old technology and they need maintenance
>they don't last forever and will eventually die(very soon)

I want to kill myself

>>10127409
There just needs to be less focus on emulating masks and scanlines, and more focus on emulating composite signals. Really, that's all you need. Most pre-gen 5 cores on RetroArch already have a built-in lightweight filter for composite/s-video/rgb to trick the dithering patterns.

>>10127420
finally someone that gets it
I don't understand when I see someone showcasing their fancy shader that doesn't even try to blend dithering or pixels. like dude you're just adding garbage to the image and making it worse for no purpose

>>10127427
Because most shaders are based on RGB CRTs. Have you ever seen a CRT with RGB? They don't blend the pixels as much as Composite and RF signals.

>>10127420
If all you want is to blend dithering, there's better ways to do it now. The old Blargg filters will do the trick, sure, but they also shit up the image in numerous ways.

>>10125475
IMMA COACHIN MAH MAC

>>10127696
>>10127420
>doesn't have the fucked up rainbow effect

>>10127696 better than >>10127420. I get NTSC-shaders chads, they have a point, but as part of the whole. If you have access to higher shaders like guest-advanced-ntsc, then you can use it to blend dither, interlacing artifacts and still maintain a cohesive and decently sharp picture. Otherwise, I'd rather control the blur level to balance a right amount of sharpness (with something like zfast-crt and variations) than to outright deface everything with only NTSC shaders just to blend a dither.

>>10127420
There are tv-out shaders that don't apply scanlines.
And as >>10127696 said, there are better ways to blend dithering.

Filter thread.

>>10127696
>blown out over saturated eyesore
>better

>>10125494
Add flare to the mouth

I'll probably be flamed to death, but I actually like the xbrz filters.
Also, not dunking on SMB3, but I like the Super Mario Advance 4 version of it best. I remember spending literally months playing the NES version as a kid.

I used the Virtual Console rom. The Bob-ombs, the "?" Orb dropped by Boom-Boom, or the wand dropped by a Koopaling use yellow effect instead of rainbow colors.

The train stage in Little Nemo. A like 4 minute auto scroller with obnoxious ass music and if you die you go all the way back to the beginning. Fuck that level screwed me over as a kid.

>>10127908
Oh I thought we were talking about levels that filtered you. Sorry I'm gay.

>>10127889
Just use whatever you like. Personally, I don't like it, but it impressed me a lot in 2012, by showing how much it improved over HQx, which was the standard upscale filter back then.

>>10125575
Leave your house. What are you a fucking mole person?

>>10127934
I go outside quite often in an area that is shaded by tall trees and with foggy beaches.

>>10127696
so what is this filter called?

>>10128326
based and comfy pilled

>>10128210
Guest-Advanced-NTSC

>>10127467
I don't normally hang out in these threads, but what does "CRT with RGB" mean? I thought the major inputs were composite, s-video, RF and eventually component? It was stated in this post as if it were an alternative input type.

>>10125052
>when I watch tv wearing dirty glasses

>>10129058
(euro)scart, scart can carry all signals(including RGB), just about all European TV's had them, Japan had its own variation of it(JP21), same cables and connector but the pinout is different.

>>10129207
Ah, I see. Thanks anon.

had this idea for a shader: plasma tvs change their subpixels on and off instantly, but they can't do in-between brightness levels, so they blink faster than what's perceptible. this results in motion clarity worse than a crt, so i've heard. the goal of this shader would be to have every pixel display some color that only uses fully on or off subpixels, i.e. RGBCMYKW 3 bit color. to do this, the source is scaled up 5x in either direction, resulting in 25 pixels per source pixel. then, according to some gamma curve, of the 25 pixels, some are lit up based on the rgb components of the source pixel. e.g. if the original pixel has a red value of 139 out of 255, then 19 of the 25 pixels would have a red component since (139/255)^0.45*25 is about 19. pic related is a mock-up of what this might look like. i used a floor value instead of rounding to the nearest number, so the result is a little inaccurate. even if it were accurate, there would be black crush and less color depth overall. this is as good as i think it could get for plasma displays, though. i guess the same could apply to led (not the lcd type) and mled displays, too. i dunno how to implement a filter or shader, but is the idea sound? you could also place the pixels horizontally to get a scanline effect if you're into that sort of thing

>>10125494
Just like I remember it from my childhood.

>>10127795
Which emus have checkerboard dedither? it does look nice.

>>10131420
It's a RetroArch shader. It uses an algorithm to detect checkerboard patters and blend them into a single color. Unfortunately this can lead to false positives, as sometimes checkerboard patterns aren't used for that purpose.

>>10131425
Not what the developers intended.

>>10131520
Indeed. It's certainly an option if you want to blend dithering but don't want composite artifacts or a blurry image, but the threat of false positives isn't worth it IMO.

>>10131559
I'd rather get some false positives instead of those composite and s-video color artifacts.

>>10129701
I didn't get it. You want to suggest a shader that only uses colors that don't strain a plasma screen, so it can focus on better motion clarity?

>>10131873
You can tweak some shaders to only do blur without the artifacts. TVOut-Tweaks and GTU can do this with a simple parameter change, and if you want more bells and whistles, guest-advanced-ntsc can also do it with even less blur than those two.

mudlord oldtv or get the fuck out of my face

or whichever the fuck one this is. I don't remember

>>10131901
that only uses colors which don't flicker

>>10125329
just as the developers intended

I'm trying to experiment with a very simple shader preset intended to only smooth the picture, not necessarily blend dither. Something like a better bilinear filtering, just to see if it's possible and worth.

So far, my best results were achieved by tweaking zfast-crt-composite. I got rid of the scanlines leaving only the mask, and controlling the amount of blur to smoothen sprites better than bilinear. For the lulz, I later combined it with ntsc-adaptive and found out a combination that blends dither while being a tiny bit sharper than guest-advanced-ntsc, though image quality is different.

I'll try to post some screenshots tomorrow. Hope someone can help me improve that smooth-only shader.

If only I could emulate the 90's economy and buy myself some snacks now

>>10125043
The only filter anyone needs is smooth bilinear.

>>10132239
Chad

here's your blended waterfalls bro

>>10131559
But I'm a law abidiep cuck goy slave and what the developers intended for me is what I have to accept to be happy.

>>10125329
The average 80s kitchen tv

>>10129069
>when I watch tv wearing dirty glasses
kek i just my glasses off to clean them cause i thought they had smudges on them cause of that anon's pic

>>10132332
Too blurry!

>>10127420
The ringing on that checkers pattern in driving me nuts!

>>10132749
There's no perfect solution, unfortunately. Your picture may be sharper, but I can see several places where the algorithm is blending where it shouldn't. It's a case of picking your poison.

thread status: filtered

Just gimmick filters plz.

>>10134657

>>10134979
Fuck off! You're not Batowl! Stop impersonating him.

>>10135679
Huh what u mean

Sorry, it took me too long! I'll post 8 pictures: four for Sonic 1 and four for Streets of Rage 2. Basically, I went for an almost no-scanline look, with a faint mask. I wanted to simulate the softening that two CRTs of mine did on the pixels, but nothing else, since it doesn't translate well enough for LCDs. For comparison, I also included pics using popular parameters for guest-advanced-ntsc.

I used:
* Blurry and blurrier (no blending) pics: zfast-crt-composite, tweaked.
* Custom NTSC pics: tweaked ntsc-adaptive + tweaked zfast-crt-composite.
* Guest-Advanced-NTSC pics: the titular shader, also tweaked.

If someone wants to help me improving them, I'd appreciate.

>>10136014
>Newfag doesn't even know Batowl posts here for years and thinks we wouldn't notice the difference

NO MORE SONIC WATERFALLS PLEASE GOD

I DONT WANT TO TALK ABOUT DITHERING ANYMORE

>>10132653
this nigga knows

>>10137330
Here's a Cool Spot instead

>>10137545
You never taught us the shader you used.

>>10137582
I've been dicking a lot with my presets lately. I'll see if I cook up something in a while.

Trying out some setups without bloom or mask mitigation of any kind save for a touch of Magic Glow and Halation, since I find that a lot of times these things kind of fuck with how the mask looks on bright scanlines. It'll probably look dim to most of you, but I usually have my monitor set to the lowest backlight brightness setting, so blasting it to full brightness really makes up for it and overall looks more natural and closer to a CRT, at least in my eyes.

>>10137923
It really is dim. But, if it doesn't bother you, it's fine, I guess.

>>10138867
This drawing is quite awesome. A refreshing take on the meme.

>>10136554
I like this one.

>>10139641
It's a popular preset here. If you want to try it yourself, it's guest-advanced-ntsc with the default parameters except:

* NTSC Preset to 0 (S-video)
* NTSC Phase to 3-phase
* NTSC Resolution Scaling to 0.80
* NTSC Chroma Scaling to 2.25

You can also tweak other stuff, but that's the baseline.

I was playing around with BSNES and found a RF filter. Probably the only one that actually feels nostalgic when playing on a 1080p monitor.
Yeah, it would probably look like total ass on a 480i CRT, but without a filter everything just looks way too sharp.

>>10140151
The latest changes from Guest appear to have made 3-Phase slightly sharper from the look of things. There's also a new "Mixed" NTSC Phase which, aptly, is sort of a mix of 2-Phase and 3-Phase, in that it's a little blurrier than 3-Phase but it doesn't have as much color bleeding as 2-Phase does. That one seems to be a little bit better for blending dithering.

>>10125494
>IT'S TIME TO AWAKEN YOU MUTHAFUCKAS

I found out you can make 3D objects blend really well into pre-rendered background games if you turn on SSAA and linear upscaling (at the original resolution)

>>10140159
Thanks for the info. It's been a few months since I updated shaders. Gonna try that.

>>10138672
Tweaked it a bit more. Trying to make the scanlines look less artificial at the moment, with more deconvergence seeping in between the gaps. It really does look bretty gud with the backlight maxed out. I bet with HDR with that Sony Megatron shader it would look fucking phenomenal.

>>10140808
Why so dark!?

>>10141171
That's better. What you did?

>>10141415
This preset: https://forums.libretro.com/t/new-crt-shader-from-guest-crt-guest-advanced-updates/25444/3751?u=hyllian

>>10141171
Read my posts. The gist of it is functions such as Bloom employed by shaders like Guest-Advanced are, in essence, a way to make up for the loss of brightness that you inherently get from imitating phosphor masks and scanline gaps on relatively dim LCD displays, and while it can definitely look passable or even good at a distance, it's not what CRTs actually do. Bloom and other such features artificially wash out the mask, whereas a CRT naturally does so through sheer brightness, and only to our eyes. So the ideal solution isn't to erase the details of the mask and phosphors, but to fully draw the phosphors and instead have the display itself output as much brightness as possible.

I readily admit it's not for everyone, or even most people at this point. Your display needs to be really bright for this method to make sense, and most aren't there yet. Even on my display with the backlight maxed out, it only truly pops with all the lights in my room turned off. This is where you want a modern display with proper HDR support, combined with the Sony Megatron shader. This shader doesn't have Bloom, Glow, Halation or any of that, because the idea is the display, with HDR enabled, will provide all of that naturally through its high brightness output.

>>10141797
Ok, my tentative using Lottes:

shaders = "2"
feedback_pass = "0"
shader0 = "shaders_slang/interpolation/shaders/quilez.slang"
filter_linear0 = "true"
wrap_mode0 = "clamp_to_edge"
mipmap_input0 = "false"
alias0 = ""
float_framebuffer0 = "false"
srgb_framebuffer0 = "false"
scale_type_x0 = "viewport"
scale_x0 = "1.000000"
scale_type_y0 = "source"
scale_y0 = "1.000000"
shader1 = "shaders_slang/crt/shaders/crt-lottes.slang"
filter_linear1 = "false"
wrap_mode1 = "clamp_to_border"
mipmap_input1 = "false"
alias1 = ""
float_framebuffer1 = "false"
srgb_framebuffer1 = "false"
scale_type_x1 = "viewport"
scale_x1 = "1.000000"
scale_type_y1 = "viewport"
scale_y1 = "1.000000"
hardScan = "-12.000000"
hardPix = "-20.000000"
warpX = "0.000000"
warpY = "0.000000"
maskDark = "0.000000"
maskLight = "2.000000"
shadowMask = "2.000000"
brightBoost = "1.000000"
bloomAmount = "0.000000"

>>10141429
>Hyllian
Damn, I didn't know he was still around. We used to talk quite a bit in the past, mainly because I gave him feedback about the xBR shader he was developing back then. I know "looking good" gets its share of hate, but, in 2011/2012, it was a hell of an improvement over HQx. It got me pretty curious to see how far he could take that upscaled look without distorting the original art.

>>10142263
Looks like Lottes is indeed powerful!

shaders = "2"
feedback_pass = "0"
shader0 = "shaders_slang/interpolation/shaders/quilez.slang"
filter_linear0 = "true"
wrap_mode0 = "clamp_to_edge"
mipmap_input0 = "false"
alias0 = ""
float_framebuffer0 = "false"
srgb_framebuffer0 = "false"
scale_type_x0 = "viewport"
scale_x0 = "1.000000"
scale_type_y0 = "source"
scale_y0 = "1.000000"
shader1 = "shaders_slang/crt/shaders/crt-lottes.slang"
filter_linear1 = "false"
wrap_mode1 = "clamp_to_border"
mipmap_input1 = "false"
alias1 = ""
float_framebuffer1 = "false"
srgb_framebuffer1 = "false"
scale_type_x1 = "viewport"
scale_x1 = "1.000000"
scale_type_y1 = "viewport"
scale_y1 = "1.000000"
hardScan = "-9.000000"
hardPix = "-20.000000"
warpX = "0.000000"
warpY = "0.000000"
maskDark = "0.200000"
maskLight = "2.000000"
shadowMask = "2.000000"
brightBoost = "1.000000"
bloomAmount = "0.000000"

>>10142361

>>10142263
Ok, not bad. I will say I don't particularly like Mask 2 in Lottes for a VERY autistic reason, namely that the phosphor layout isn't evenly spaced. Mask 2 is a Trinitron-style RGB layout, and while you'd think it makes sense to use this, as that's what CRTs used, in practice what happens when you use that layout in shaders is the red and blue subpixels on your LCD panel end up close together to form magenta, like so:

RxGxBRxGxB

What you ideally want is for all three subpixels to be evenly spaced out so none of them merge together like that. An alternative layout would be Magenta-Green-Black (found in Guest-Advanced as Mask 9), which at the subpixel level results as follows:

RxBxGxxxRxBxG

So while the shader is technically drawing a magenta pixel, followed by a green one, and then finally a black one, the way it works out is you get evenly spaced subpixels that end up looking like an RBG phosphor layout, with a little bit of space in between.

Like I said, subpixel layouts get severely autistic, and I've gone down my fair share of rabbit holes with this shit.

>>10142369
Shit, I fucked up the layouts. The first should be like this:

RxxxGxxxBRxx

Where each x is a blank, unused subpixel. Again, compare it to the second layout, which should be like this:

RxBxGxxxxRxB

Alternatively, using a BGR layout, Mask 2 can look like this:

xxBxGxRxxxxB

This one looks a bit too spaced out to my eyes, though, for some reason.

>>10142395
Too much imperceptible detail to take into consideration! Guest's has an option to use BGR instead RGB, which, I admit, in some color regions you can spot some differences, but overall, the look is basically the same.

>>10142454
I did say it was autistic. At the end of the day, all of this is highly subjective, anyway.

>>10140180
I find that turning on supersampling with a shader yields better results.

>>10143050
supersampling on

>>10143056
supersampling + shader (doesn't need to be this specific preset)

>>10133296
That's why I then put the crt filter on top, to hide the impurities

>>10143060
You're right it looks great

lel

>>10146790

>>10125663
idk its been a while since I've posted on here. but they aren't too annoying imo so who cares

>>10127420
>more focus on emulating composite signals.
Retard, blargg did that 20 years ago

I unironically hate /vr/ shader threads
Every single post is shitty settings, you never have anyone use 100% mask strength etc...
just garbage

>>10149193
>noooooooooo you're not allowed to use things you like, you have to use the things i like!

>>10149193
There's one right here >>10144383

>why yes I prefer pixelated backgrounds frfr no cap how did you know ong

>>10125274
>World 8 for me is when I usually stop having fun.
Its Mario, its not meant to be fun its meant to be comforting or at the very least numbing
https://www.youtube.com/watch?v=oIZUJ4Msv7U

>>10141797
Bloom is a fucking gay artifact due to the imperfect nature of CRTs
The CRTs devs used to make games had very little bloom compared to consumer shit

>>10149210
Shit settings with awful fake bloom applied

>>10149216
2nd one looks great because it blends that disgusting dithering

>>10149193
>100% mask strength
Thanks for outing yourself, zoom-zoom! Real CRTs used 2000% mask strength.

>>10149223
>Its Mario, its not meant to be fun its meant to be comforting or at the very least numbing
>>10125475
>ARE YOU NOT ENTERTAINED???
>>10125274
>World 8 for me is when I usually stop having fun.
>>10125043
>Super Mario Bros. 3

https://www.youtube.com/watch?v=QrWZNAQrkf4

>>10149417
Ah, a fellow Sonic 3D Blast chad... what are those shaders though? I really like the 1st and 4th ones in this.

Shaders are good. Crt's are good but not many exist anymore.

>>10149530
unfiltered > NTSC preset by anon here (the one with the sonic waterfalls) > modified royale > some preset some anon posted in a previous thread > modified lottes

>>10149583
Thanks, anon.

>>10137330
>NO MORE SONIC WATERFALLS
>STICK TO TAILS ADVENTURE AND LAKE CRYSTAL YOU'RE USED TO

>>10125043
>What are some filters

>>10149357
I don't think we're talking about the same thing. Bloom as used in CRT shaders is a feature that more or less reduces the visibility of the phosphor mask the brighter a pixel is in order to recover some brightness to the image. Normal CRTs don't do this, whether they are consumer CRTs or PVMs or whatever. If we can't see the mask, it's because it gets drowned out by the brightness of the CRT, not because the mask actually goes away.

>>10149364

My favorite filter is hiding threads on 4chan.

>>10137330
ok

>>10131960
beautiful so amaze i nostalgiae'd hard