/vr/ - Retro Games

This one actually looks pretty good.

>>10805482
I'm glad the obsession with filtering pixel art has been left behind

I wonder how Yoshi's Island would look. Probably good

>>10805482

>>10805482
Is pixel smoothing a lost art form? Are there high fidelity modern algorithms or are we stuck with whatever ZSNES had in the mid-2000s?

>>10805490
since we realized crts blurred the image and raaw pixels were never the developer's intended way of viewing the art, we have moved on to replicating crts via filters. but these filtered pixels of old are closer to the developer's vision than any clean pixels ever were.

>>10805582
And I'm sure one day you'll move on to the next fad.

>>10805487
If you're a simpleton maybe.

>>10805606
And if you're not a simpleton definitely.

>>10805582
"Raw pixels" don't exist on a screen. Pixels are point samples of zero area. It is arbitrary to expand them to big squares with absolute boundaries, and it's arguably less valid than other methods of extrapolating them to an image.

http://alvyray.com/Memos/CG/Microsoft/6_pixel.pdf

>>10805603
CRTs/shaders really are a good way to maximize apparent resolution of SD content while not making up information that isn't there

projection is another option, but too high latency for gaming. i often go to screenings of DVD sources on a 2K cinema projector

i also go to CRT events you bumpkins will never know about :-)

>>10805626
man i was just larping idk wtf you’re talking about. but as long you’re happy.

>>10805612
Sounds like something a simpleton would say.

>>10805642
no u

>>10805482
Someone photoshopped the dolphin on the right, there are too many artistic liberties that a post-processing shader would not do.

>>10805626
>Pixels are point samples of zero area
Only true when you're aiming for realism. The defining feature of pixel art is that it does in fact treat pixel as having area.

>>10805482
That cherrypicked example is like the only instance of that filter producing a decent result.
The author of the filter acknowledges as much as well.
https://johanneskopf.de/publications/pixelart/

>>10805684
It's an edge detection algorithim. Edges are processed as splines and smoothed in the vectorization process. The elements determined as not being edges are blurred instead to produce color gradients.

https://johanneskopf.de/publications/pixelart/paper/pixel.pdf

The goal of the specific filter in OP's picture is actually vector graphic conversion with the specific goal of removing visible pixels, and is best suited to anything with black outlines.

If you've ever used the "trace bitmap" option in Macromedia / Adobe Flash, it's a competing implementation of that.

>>10805717
kek lookin good there marine

>>10805740
i'll just never understand why people try to fix what ain't broke. as a thumbnail, all three look the same. and it's basically the resolution you're meant to be seeing and playing these games. it's a small sprite, supposed to be on a small screen, interpreted by your eyes as one element in a bigger picture.
when you expand the image and see it blown up to fuck, then for me personally the left pixellated image becomes jarring to look at. it's not how you're meant to see the sprite. it's expanded by 16x pixels and occupying the bulk of your vision cone. of course it looks like shit. it's only at this point that the bicubic result looks more... interpretable? like i can easily see it's the viking man instead of a mass of cubes. but i also would never want to fucking play a game at this resolution. am i playing lost vikings on a 4k 50 inch flatscreen? i really fucking hope not

>>10805626
>Pixels are point samples of zero area.
Incorrect, digital equipment cannot create this. Your NES doesn't make "Point samples of zero area." Your point here is useless unless you're playing video games on analog equipment.

Ever since NASA invented digital photography, a pixel is a digital square, deal with it.

>>10805568
Lol, just use CRT shaders, that’s literally what they’re for
>>10805626
Maybe on the analogue output signal
The consoles are still producing a set amount of “raw pixels”

>>10805568
Newer ones are developed all the time. ScaleFX is probably the best one atm, followed by xBRZ. Can't say I like them myself, but they're an option.

>>10805487
>>10805606
>>10805612
>>10805642
>>10805649
keep up the high level intelectual discussion, lads

>>10806192
Still looks like dogshit, but kind of impressive the edges it manages to detect and group up certain objects

I still can't get over how popular this shit used to be. Like every other emulator screenshot use to have this filter

>>10808073
Smooth filters came when there weren't any crt shaders. Before 2010 only shit scanlines filters were available.

>>10805734
It's the humps at the top and bottom that really kill it.

Heh, I remember liking 2xsai.

>>10805482

>>10808158
>blends the pixels
>still leaves the dithered pattern of the water

>>10806189
Love Bowser and Geno merging into the ultimate lifeform over there on the left.

>>10808116
Keep in mind they also came out when everyone was still using CRT monitors, and they were meant to be used at 640x480 or 800x600. Pic related is kiiiiiiind of what it was supposed to look like with a filter such as Scale2x. Needs much more resolution to properly look like a CRT monitor running one of those vintage filters.

>>10805717
honestly that thumbnail doesn't look too bad, it kinda works for me but only far away/small

>tfw I love to add a lot of bloom/glow/halation to my shader's settings and people mock me for it despite it being how I recall my childhood awe at those big CRTs I loved so much back then

>>10805717
>That cherrypicked example is like the only instance of that filter producing a decent result.
It looks like it would work on any sprite which was "cel-shaded" (for lack of a better term) in that fashion. Doom Guy has pretty detailed lighting squashed into his mugshot, so it's obviously going to look like shit. I'm impressed that it actually does hit the geometry for a few of the striking facial planes though.

>>10806189
finally, how developer intended..,.

>>10805717

>>10805626
>"Raw pixels" don't exist on a screen.
False, how can we see them than?

>>10810321
gooking lood

Minus the rainbow.

>>10810070
they have to be cooked first

>>10805509

captcha: MORON

>>10805734
>>10808130

>>10811204
>>10811205
>>10811209
>>10811221
Is this a joke? Does anyone actually use this? Did they threaten the authors they would lose grant money unless they publish something? I'm so confused.

>>10808181
when you show it like that, i can kind of see the appeal

Still good when using a crt shader to mask their weaknesses

>>10811221
more like this

>>10808181
I played a bunch of shit like that on ZSNES in 2004 and loved it.

Sharp, square pixels on a design document for a Famicom game from 1990.

tfw you can't stop me from using SuperEagle on every game every time forever
Keep seething chuds

>>10811505
Why the fuck do you fags think filters are terrible but then having a grid on top of the filter makes it great?

oh no...my video games are FUCKING RUINED

>>10810321
How do you make it look that way? I'd actually like to play some games like that.

>>10805717
It's not surprising that it fails with highly detailed life like art. The results with cartoinish low high frequency detail art are much better.

>>10811505
This is how I wish it looked when I was a kid.

>>10811684
CRT shaders do not put a grid over the image. They draw the virtual phosphors and then draw the image onto the virtual phosphors. It has the effect of significantly increasing apparent resolution while not creating information that was never there.

Pixels go much further when they're not presented as square units of area and are allowed to talk to their neighbors. You've probably paid to see a movie on a 1080p projector and assumed you were seeing 4K

>>10811845
It's just another math equation blending the pixels only it's more complex than old filters and mimics what a CRT would look like
There is no more increase in apparent resolution than what filters do and there is no less creating of information than what filters would do
Even a real CRT "creates information" with the way it works

All of these go about this all wrong.
They should be working with a CRT shader first to get a version of the sprite as it would be seen on a CRT, then applying the filters and upscaling.

>>10806189
quinoa

Working from the raw pixel image is throwing a way a bunch of information that is supposed to be there from the variable intensity of the CRT image. Apply the CRT shader first to get a broad idea of what the sprite is supposed to look like on a CRT as the artist envisioned, then work with that to do whatever upscaling you want to do. It makes no sense to start with the raw pixels when that is not the highest quality source you have available.

>>10811156
>>10805509

>>10806189

Oh look another thread where literal virgins, literal manchildren who live in basements and will never touch a woman, shriek and argue about how other people are playing video games wrong.

Yes, it was definitely important to waste more board space on the diaper shittings of these children.

>>10812113
i'm not a virgin, and i often touch women multiple times a week, mostly moshing, high fives, PLUR, and hugs, but vintage gaming is ABDL-adjacent if we're being honest

these threads are harmless fun, you uwu smol bean :3

>>10812113
Why are you projecting so hard against scarecrows you weirdo

i am immensely thankful that i was too retarded as a kid to figure out how to get filters to work when emulating for the first time

>>10811204
>spook mon
>smoke mon
>needle mon
>mooo mon
>pico man
>voo mon
>gemini mon
>mogney mon
>snooow mon
this filter looks like the kinda shit a normalfag would have on their wall

>>10812886
>i'm special for playing extremeley mainstream video games

lol

>>10811204
>>10812886
>spook mon
>smoke mon
>needle mon
>mooo mon
>pico mon
>voo mon
>gemini mon
>mogney mon
>snooow mon

>>10811505
The entire title text looks like shit

>>10813582
better?

>>10813582

>>10813683

>>10805582
>since we realized
I've "realized" that your a niggerfaggot zoomer that never knew what a tube tv was until a youtube video told you about them.

>>10813340
im not sure whats supposed to be funny here. this looks pretty normal

>>10811209
Liars. They also very likely just upscaled and then smoothed while skipping the upscale on VM.

>>10812113
I've had sex but yea fuck you

Where's the issue?

>>10806192
not gonna lie, looks good enough to put on a tshirt

>>10805717
>>10805740
>>10806189
>>10806192
>>10810321
>>10811198
>>10811204
>>10811205
>>10811209
>>10811221
>>10812103
>>10815001
i just wanted to bump the thread

>>10816242
Same

>>10811682
We can't stop you from eating steak well done, and sucking dicks behind dumpsters either. Your choice, your life senpai.

>>10816480
WAKE??

Dolphin at one point used to have a bunch of meme filters pre-installed they they clearly never expected anyone to use. One of them was an ASCII filter that turned the entire game into ASCII and that shit slowed my PC to a total crawl. I'm pretty sure it took me like 10 fucking minutes to close the Dolphin process and in that time it rendered like 5 frames.

It was funny but holy shit.

>>10816976
Extremely based. kek

>>10811221
Funniest image I've seen to date on this board

>>10811221
>let me assist you, doctor

>>10806189
Pablo Picasso's Super Mario RPG

>>10807795
simpleton

>>10812103
needs a "If only you knew how good looking things really are" edit

>>10805717
Yknow, if I hadn't clicked on it, the thumbnail made it look kinda good.

>>10805886
>a pixel is a digital square
Except not a single retro console used square pixels so you are full of shit

>>10805708
All of what you said is buzzwords
A display doesn't care about realism or fantasy

>>10805568
Pixel smoothing is retarded
If you're gonna do it just use bilinear or bicubic and call it a day
Yeah you get a pretty blurry picture but with bilinear/bicubic you definitely get a picture that resembles the actual artistic intent

The best way really is nearest neighbour/integer scaling into an absurd resolution screen like a 10000/20000PPI display which we will have someday

>>10818576
SO you'll play on a 2" display with 20000PPI! :P

>>10811845
>It has the effect of significantly increasing apparent resolution
No it doesn't
Depending on the CRT, for example if it is a higher resolution one it just results in the shitty blank scanline effect
If it is a lower resolution one then it can blur the image because the scalines overlap on top of each-other, this has the effect of reducing aliasing in exchange for losing some detail and sharpness
Dot pitch also plays a role

>>10811896
>Even a real CRT "creates information"
No it doesn't
A CRT is a dumb device
It doesn't create anything

>>10811917
>Working from the raw pixel image is throwing a way a bunch of information that is supposed to be there from the variable intensity of the CRT image.
stupid motherfucket

>>10818584
20 inch 4:3 screen 200-400K resolution
It's doable
I BELIEVE

>>10818628
And how you'll fill up 200k x 400k resolution with only 320x240 samples (common retro game resolution)?

>>10818634
Did you somehow completely miss my post?
I mentioned integer/nearest neighbour upscaling

>>10818678
Well, then it'll look the same as a Full HD display. A lot of blocks on screen.

>>10813683
shader?

>>10805568
zsnes has blarg ntsc filter so is the closer to playing on a real crt you can get if you make the proper setup (i did it comparing a snes connected to a crt tv and then made the setup on zsnes with a plasma TV so is the closest you get crt).

>>10819139
No need for ZSNES. SNES9X also has the Blargg filters, both the standalone and RetroArch versions.

>>10818730
Wrong
You don’t know what you’re talking about
Look at how perfectly how straight horizontal and vertical lines upscale(check the ---- LIFE ---- font on the HUD of Zelda ALTTP for example)

When you have massively higher PPI ranging from 10000 to 30000PPI the same thing occurs for curved and vertical lines
And this effect is also drastically increased from a normal viewing distance + the size of the display(though this depends on the game and sprite sizes e.g NES/SMS 8-bit games and some SNES/Genesis games with smaller sprites will always look better on smaller screens whereas you can get away with playing Arcade games like Street Fighter 2 on a bigger screen because of their higher detail/bigger sprites)

Why do you think retro games integer scaled on a smartphone with a PPI of 300 to 800 have significantly fewer jaggies compared to Monitor and TV? It’s because of the combination of high pixel density and a small screen.
Of course not even smartphone screens are perfect but they look significantly better than Monitors and TVs.

Ultimately you need a lot more pixel density to properly upscale curves and diagonals compared to straight vertical/horizontal lines. That is all it is.

Is anyone else laughing uncontrollably from this thread or is it just me that have autism?

>>10811204

>>10819872
Now not sure if you're dumb, joking or just thinking you're serious.

Trans Rights

>>10819872
What the fuck, genuinely
>Why do you think retro games integer scaled on a smartphone with a PPI of 300 to 800 have significantly fewer jaggies compared to Monitor and TV?
Because it's fucking smaller so you don't notice? Play them on a vita with only 2x scale and the effect is the same.

>>10817374

>>10819139
> is the closest you get crt
No it’s not
Cyberlab Megatron shaders are way superior
Especially on a QD-OLED Monitor

>ZSNES
Poorfags and autistic boomers need to be shot at this point

>>10820025
>>10820054
Kill yourself you fucking idiot

>>10818563
They all used square pixels except for Capcom's arcade boards. Some devs compensated for the way images would be freely stretched by CRTs; most devs didn't bother and just let the image be stretched. If you really want to argue that circles are meant to be experienced as ovals, go right ahead.
>But GameCube's framebuffer used non-square pixels!
Barely retro, but that is technically correct.
>Wikipedia says that the SNES used non-square pixels! https://en.m.wikipedia.org/wiki/Pixel_aspect_ratio#:~:text=In%20that%20generation%2C%20square%20pixels,etc%20had%20non%2Dsquare%20pixels.
That Wikipedia article cites a source that explicitly says it's not intended to be a citable source. https://pineight.com/mw/page/Dot_clock_rates.xhtml
The cited source also makes no such claims. That Wikipedia article is made-up slop.
>Who cares? It's meant to be experienced in 4:3 because that's what everything was stretched to be on consumer displays of the time.
Enjoy your oval suns and moons.
>Thanks I will.
Based

I don't mean to strawman you by the by, these aren't necessarily your opinions (unless they are, that's up to you). And it's good to keep in mind that, while not the majority, there's a very fair share of games that do account for image stretching.

My personal onion is that you should stretch on a per-game basis to match what the game considers to be a circle. Where it gets confusing is, for example, the many PS1 games where there's *no fucking way* the button prompts were made for any aspect ratio at all, they just squished the hell out of them, you'd have to stretch the game to 16:9 or beyond to make them circles. You have to use your best judgment and pick objects that make sense. Some games really do have oval suns and moons when everything is accounted for; find another circular object and see if the game looks better when adjusted for that object.

>>10819886
It wasn't just you, some of these pictures were way too funny

>>10805492
nah not really
tbf the game already looked great. filtering that is just gonna be a step down no matter how you slice it

>>10820940
NTA but
>They all used square pixels except for Capcom's arcade boards.
None of them did.
SNES used 8:7 pixels
Genesis used 32:35 pixels for 320px mode and same as SNES 8:7 for 256px mode and they both had a couple of other modes
PSX had the same modes as Genesis and SNES + boatload of other video modes
CRTs were analog so it didn’t matter what pixel aspect you used. I remember reading that building hardware that coould render square pixels was expensive back in the day so it just wasn’t generally done.

It is insane how you can claim that any consoles from back then used square pixels and then contradict yourself by posting
>https://pineight.com/mw/page/Dot_clock_rates.xhtml


>Where it gets confusing is, for example, the many PS1 games where there's *no fucking way* the button prompts were made for any aspect ratio at all, they just squished the hell out of them
They didn’t squish them
Many lazy devs just didn’t account for the squish is more likely
This was common with retro games
You find many many retro games with assets that were accounted for DAR output and assets that weren’t all in the same game
Not even worth thinking about
Just play whatever the fuck makes you happy

Only one thing to consider is that if you want accurate Display Aspect Ratio(DAR) output then you have to remember that consoles never outputted a perfect 4:3 ratio so if you output 4:3 it will look a little fatter than original hardware output
How much fatter depends from hardware to hardware but it almost is never perfect 4:3/1.333333:1 output except for I think 3DO

>>10822585
I think we have the same idea but have different ways of defining it.
From my definition, aspect ratio and pixel ratio are independent measures. If there is a simple aspect ratio mismatch, it is up to the developer to decide how they want to deal with that, and it looks perfectly fine to simply not deal with it. But if there is a non-square pixel requirement, the stretching is much more severe and the dev *must* keep it in mind.

It sounds like splitting hairs, but consider it like this: On SNES, whether the dev accounts for the 8:7 PAR on any given game is a complete toss-up. They can treat it like non-square pixels, but most chose to simply treat it as cropped borders on the left and right sides of the screen and whatever happens, happens. Conversely, on CPS2/CPS3, 100% of developers accounted for it; every single game looks stretched out if you play with 1:1 PAR. This is where the distinction is. Non-square pixels can either be a soft requirement (per-game) or a hard requirement (every single game) depending on how severe the stretch is when not conforming to NTSC / PAL standards.

>They didn’t squish them >Many lazy devs just didn’t account for the squish is more likely
For PS1 games? I suppose. I don't think the devs were lazy, but it sure is a weird choice for them to leave button prompts super squished.

>>10811221
just like i remember them

It's pointless to post CRT screenshots since digital cannot reproduce it.

>>10808125
HURRRRRRRRR IMMA DURRRRRRRRPH

>>10810321
Stunning...

>>10806192
See. now this is a filter that's worth a playthrough. I like shit like this, not better than the original but an interesting interpretation

bump

>>10818621
The CRT adds its own complexity to the rendered image through analog means

>>10822803
>aspect ratio and pixel ratio are independent measures.
Correct
Aspect ratio = Display Aspect Ratio(DAR)
Pixel ratio = Pixel Aspect Ratio(PAR)

The point is that neither console or arcade hardware rendered or outputted 1:1 PAR
There is this misconception that CRTs stretched the image but this is wrong, the hardware itself used a different PAR and CRT simply painted what the signal told them to

Of course CPS1/2/3 took great care to ensure everything was as perfect as possible because they were in Arcades where presentation was second to none.

But when it comes to console games otoh, like I said, there are plenty of games where assets are inconsistent in the same game with some assets taking DAR into account and some assets not doing it. I think part of this may be that it was extra work so some sprites were small enough to where they thought they could get away with flubbing it here and there.
IMO just choose between original hardware aspect ratio or go with 1:1 square pixels depending on the game and what you think looks better/more correct and call it a day... as long as you don’t do generic 4:3 scaling which is never ever correct with some consoles looking fatter and others skinnier
>NTSC/PAL standards
No console(apart from maybe 3DO or CDi) really ever adhered to NTSC/PAL spec. And Arcade games even less so(53hz Mortal Kombat completely off-spec compared to PAL/NTSC for example).


>For PS1 games?
You’d be surprised.
In 5th gen more games were being made because of the cheaper price of the CD.
Now granted, you probably mostly play the AAA games on PSX like most people but even there, the focus for many PSX games were the FMV cutscenes and the CD Quality Audio and whatnot. I feel like many PSX and 5th gen games in general pay even less attention to how font and UI looks because of that.
Also the PSX had many video modes with many games utilizing different resolutions(mostly 480i for crispier menus) so that may have added even more complexity.

>>10823943
>since digital cannot reproduce it.
Actual retard

>>10827707
All agreed. It's why scaling retro games can feel like such a crapshoot sometimes. Try getting the correct aspect ratio for Castlevania SotN, there's like 4 different output resolutions (main menu, FMV, gameplay, pause menu) and if that wasn't enough, each of these resolutions has its own ideal aspect ratio. No display or emulator could possibly adjust for all these, and I have a hunch that it doesn't scale ideally on real hardware either. If it does that'd be amazing, and it would make a strong argument for being able to set aspect ratios not only per game, but also *per resolution used by the game*.

Knowing how "fuzzy" all of this can get does explain a lot about how some games look almost correct at 1:1 PAR but need slight stretching to look right.

>>10806192
At least the Sigma still looks like a Sigma, I'll give it that.

>>10811198
DIME :DDDDDDDDDD

>>10827732
I could be wrong but I think all the various aspect ratios are documented somewhere. So it could be done(as in output like OG hardware did).

>>10828517
Honestly looks pretty accurate apart from the smudge making it look waxy

>>10818624
You have no idea what you're talking about.
None.

>>10805734
>>10808125
I think these two work but that's about it

Filters like 4xbrz are racist because they manipulate and distort the images that the original Japanese artists intended for their audiences. It would be like if you watched a movie about Donald Trump and Donald's character was a Black man.

>>10818628
>20 inch 4:3 screen 200-400K resolution
For that to happen we'd need to start seeing 4:3 AR monitors & TVsmake a comeback.
That's my dream, even as they are currently. A 28" 4k monitor is equivalent to a 22.88" 4:3 monitor but they're not going to sell you a nice 22-23" 2160p monitor or TV even though the industry has a soft spot for that 27-28" 4k sweetspot in 16:9. Even the 32" 4k monitors equal out to a comfy 26" in 4:3. At appropriate viewing distances we're getting somewhere good.
My problem with the current 4k monitor trend is that besides working in strict size profiles with little variance (look at TV sizes currently) is the dead space if you use them for 4:3.
I get they're flexible but that nice 26" 4:3 display, there's a width difference of almost 7 inches that's dead space.

>>10810321
This really intrigued my pingors

>>10810321
hot new indie remake

>>10816301
That's some very pregnant goats.

>>10805482
>HQ
works fine for cartoony games, but SuperEagle is faster.

IMO 2XSai is fine for 3D textures. for 2D Raw 3X is optimal, though NTSC or scanline filters work fine for console games too.

>>10806189
I actually kind of like this. Can this be done in real-time?

>>10811221

>>10805482
Anything higher than x2 is just being greedy.

I wanna see this at 8K.

I feel like people who like these pixel smoothed visuals gotta be the same phenotype who post AI art to twitter and get into arguments about it.

It doesn't make the pixel art look like the concept art or promo art that it was based off of. It all has this weird blurry look like a kids finger-paint drawing seen through a vaseline smeared window. Even if you don't particularly like pixel art as a style or have any attachment to the old visuals it just looks objectively shitty.

>>10805812
Yeah nailed it perfectly.

HEAVEN FORBID I SHOULD SEE A SQUARE PIXEL WHEN I'M RUBBING MY FACE AN INCH AWAY FROM MY THEATER SIZED GIGA PLASMA GAMER HD SCREEN I must put a garbo filter over it so that it simulates the feeling of being ray charles staring through a vaseline smeared window underwater.

my wife is already perfect with her crt filter