/vr/ - Retro Games

>>6103987
Thats pretty cool.
Link to video?

>>6103987
People talk shit about N64 having blurry video output but look at that difference. It really cleans up the edges.
>>6103989
https://www.youtube.com/watch?v=mznc8dhijTo

>>6103989
https://www.youtube.com/watch?v=mznc8dhijTo

>>6103987
>Thanks to Digital Foundry
Or, you know, the debug option that’s been available since the game launched in 1996.

>>6103997
We've got video now though

>>6104000
I think I was playing a video game in 1996.

In 1996 I purchased a PSX. Came with a demo disc and I bought Warhark and Raiden Collection. Wasn't many games at launch. We laughed at the N64 with it's "paltry cartridges" ahahaha.

soul/soulless

>>6103987
>Thanks to Digital Foundry
No
>undeniable evidence
LOL
>>6104000
>We've got video now though
LMAO. We had video back then dumbass.

>>6103987
We've always known this. It's discussed in the N64 programming manual. The N64 used coverage-based AA which was cheaper than MSAA but lower quality. Kudos to DF for exploring it, though.

DF should do a video on Factor 5's N64 work in general. Rogue Squadron, Battle for Naboo, Indiana Jones, plus all the games that used their codecs.

>>6104009
No load times

Honestly, the N64 would have been a lot better if it didn't force AA and bilinear filter.

Developers ended covering shitty and lazy texture work with blurriness all over the place. Pic related. With sharp pixels visible, the devs would have to take extra care of the final look of the texture work.

The AA of the N64 was pretty rough anyways with jaggies still visible.

>>6104047
And with this I mean that AA and bilinear filtering should have to be optional.
Certain games automatically look superior when you turn off of the cataract filter, like Doom64.

>>6104035
Small nitpick, but coverage based AA is not necessarily lower quality than MSAA. (2x MSAA at least).
In fact coverage based AA when applied only to polygon edges is actually 100% correct where 2x MSAA, 4x MSAA, etc. is only approximating the mathematical correctness of coverage AA through brute force.
The problem with coverage AA is that unless your hardware has an alpha buffer, overlapping partially transparent line edges will combine slightly wrong, and depending on how z-buffering is handled with the partial pixels, you could get ugly ringing if you don't pre-sort.
My guess is that the N64 was the last major graphics chipset that didn't massively parallelise everything so draw call ordering by the developer was still practical. Compare with the PS2 where AA was only practical when using full height buffers so pretty much no one used it when faking it with a framebuffer effect was simpler and consumed less RAM.

>>6104047
>The AA of the N64 was pretty rough anyways with jaggies still visible.
It was 320x240 for the most part. Even if you 8xMSAA that you're going to see the pixels that make up the edges. It's a contrast thing rather than an AA thing.
Personally I think the AA and bilinear (biliner+?) filtering was fine, it was just that final post-proc smearing and the shitty composite that ruined everything.

>>6104136
Also 16bit dithering messed with the perceived clarity something awful. Banding is preferable to dithering at low res IMHO.

>>6104009
its bullshit how they didn't include a full game with the playstation

also thanks to DF retro we have had more conjunctive adverbs and transitional phrases to last multiple life times.

>>6104136
>Even if you 8xMSAA that you're going to see the pixels that make up the edges
Not on an appropriate CRT. It effectively blurs pixels together.
Perhaps a PVM is too sharp for 240p, but 640x480 with MSAA looks almost flawless on a Trinitron.

So how does the N64 do anti-aliasing?

>>6104494
It doesn’t. It does something like FXAA. It does a filter post processing. DF is retarded as fuck

>>6104009
No self respecting man would ever play N64. Whenever I see an N64, I always joke with my girlfriend, "There goes another gayboy." I always fuck with N64s, I get right next to them, and rev the shit out of my straight piped V8

>>6104047
You're making exactly the same mistake that DF Retro disproved. AA != texturing on N64. Edge AA doesn't apply to texture pixels, only edge pixels.

>>6104509
>It doesn’t. It does something like FXAA. It does a filter post processing. DF is retarded as fuck
Jesus. Literally brain-dead poster.

>>6104132
>you could get ugly ringing if you don't pre-sort. My guess is that the N64 was the last major graphics chipset that didn't massively parallelise everything so draw call ordering by the developer was still practical.
You don't need sorting to make edge AA work on N64. That's what makes its implementation so special. Literally every other implementation of edge AA needs sorting to work, which is why even though Voodoo and PS2 supported edge AA they barely every used it. It's way too expensive to sort.

The way the N64's GPU avoided the sorting requirement is splitting up the edge AA process in two, anti-aliasing inner edges in the middle of GPU pipeline and outer edges at the end of the pipeline on the VI unit.

IIRC the only sorting requirement that remains is separating the draw order of transparent from opaque polygons into two different groups, and that's it.

>PS2 where AA was only practical when using full height buffers so pretty much no one used it when faking it with a framebuffer effect was simpler and consumed less RAM.
That's just the interlace blending, which isn't a genuine form of AA, but it does filter out flickering. The PS2's only true hardware AA is edge AA exactly like N64... but without the sorting fix, it basically made it worthless, which is why developers were forced into trying other things like supersampling.

>>6103987
Nintendo always wins baby.

>>6104570
So similar in concept to what CSAA does? Looked almost as good as equivalent MSAA at about half the cost.

>>6104546
You're not getting it. The joke is pic related

>>6104047
>>6104052
It doesn't "force" the filtering. N64 filtering can be enabled and disabled on individual polygons. Most devs just left it on though, because at the time the blurry textures were considered way more high-tech than pixellated textures.

Actually I've had trouble finding instances of any devs turning it off, ever. The only times I've noticed the blur disabled is certain textures in Paper Mario 64, like here near the beginning.

>>6104047
>With sharp pixels visible, the devs would have to take extra care of the final look of the texture work.

>Extra development time is BETTER

fucking retard

>Sonic is too slow for him

What an absolute fucking madman

>>6103987
>Thanks to Digital Foundry we now
Know about something any dumbass could look up in the game's debug menu or find early youtube videos of.
>used actual anti-aliasing
We've known that for years as well dumbfuck.
https://en.wikipedia.org/wiki/Nintendo_64_technical_specifications#Components
>Hardware features: texture mapping with perspective correction,[5] anti-aliasing,[4] Z-buffering,[2] bilinear filtering,[6] trilinear filtering,[5] Gouraud shading, 8-bit alpha blending, level of detail management.

Real talk, why is everyone here retarded?

>>6105682
Another one is the distant skyscrapers in the first level of Perfect Dark.

>>6105927
People have been falsely claiming for years that the N64’s anti-aliasing was just a lying marketing term for vasline filter

>>6103987
AA and texture filtering on, blur off is the ideal imo

>>6106013
>People have been falsely claiming for years that the N64’s anti-aliasing was just a lying marketing term for vasline filter
That's mainly because the AA is basically a vaseline filter when applied to resolutions like 240p. This is also exacerbated by most people playing games in composite which the N64 outputs poorly.

>>6106027
>That's mainly because the AA is basically a vaseline filter when applied to resolutions like 240p.
You are like a flat-earther. No amount of evidence (whether its video footage or even the fucking N64 manual) can seem to convince people like you that the N64 uses a coverage form of AA which is LITERALLY antithetical to a screen-based filter (coverage meaning the polygon edges are specifically marked and stored in memory)

I wasn't aware that people thought AA was a myth, lol.

>>6103994
>It really cleans up the edges.
This isn't needed on a crt though.

>>6106092
I mean, it had the first incarnation of a GPU in a home console, so stuff listed as being supported in >>6105927 shouldn't really be a surprise. But people aren't really that versed in when arcade/consoles > PCs on graphics got flipped so a lot of assumptions about how the RCP can't do it and etc.
But in my opinion, AA can probably stay at 2x for most conventional 3D and you get great bang per buck. But in my opinion, texture filtering is much more important for visuals, and it really sucks that even today, it has gotten the shaft on console games because game consoles are always CPUs and memory bandwidth limited so even though stuff like the Dreamcast supports ansiotropic filtering, most games used bilinear until the current generation, where they still aren't maxing it out and most use 4xAF. PCs can literally 16xAF on even low end graphics cards and see less than a 1 FPS hit.

>>6106421
The funny thing is that trilinear filtering is basically "free" on N64 because z-buffering congests memory bandwidth so much that by the time memory access becomes available again, the GPU has enough time to fully load two mipmaps and trilinear filter them. That's why almost every N64 did it. Only reason to not use trilinear filtering was either you didn't use z-buffer so your rendering was faster than trilinear, or you didn't want to load mipmaps into the texture cache.

>>6103987
I knew this since before it came out and false advertising is illegal so it pretty much had to be true. Have fun with now having your epic proof from literal who's, and using that to 'win' e-arguments with shut-in's on internet forums, you dumb fucking faggot.

>>6106425
The N64 not really have a trilinear filter, and it's a misnomer for what was their custom 3 sample bilinear filtering instead of the 4 sample standard. I remembered from this blog post from a while back explaining it and it links to the patent filing too.

http://filthypants.blogspot.com/2014/12/n64-3-point-texture-filtering-in.html

So Wikipedia is technically wrong here in >>6105927, and looking at its source, it quotes a gaming magazine, which isn't really the greatest of authoritative sources especially when in hindsight. Trilinear seemed a bit too early anyways for 5th generation consoles.

>>6106456
Actually it does have trilinear. 3-point bilinear has nothing to do with it.

The way trilinear works is this: two bilinear filtered mipmaps are themselves bilinear filtered with each other. That generates a trilinear filtered pixel.

Whether the bilinear filter used in that algorithm takes 3 samples or 4 samples doesn’t matter. The post you linked doesn’t contradict that at all.

>>6106460
How was that possible with the 4k texture cache on the RCP? Can you can fit large enough mipmaps and enough textures on there to do this to an effective degree over regular bilinear? I would probably just rather take the extra space for more textures instead but what do I know.

>>6106467
The way it works is (assuming you use the whole texture cache for one texture) is that the base texture goes into the first 2 KB half and all of the different mipmaps at every possible smaller size go into the remaining 2 KB.

Of course you can have mipmaps without trilinear on N64 but there’s no point, because the way the N64’s GPU works, the ‘cost’ of enabling mipmaps gives you free trilinear (but mipmaps themselves are not free).

Some games like Conker only use mipmaps sparingly to get higher resolution base textures. But without mipmaps, you get shimmering on distant textures of course. It’s a trade off.

>>6106092
Tards aren't people

>>6106421
>it had the first incarnation of a GPU in a home console
Literally retarded. Are you OP or do we have two downies ITT?

>>6106851
wahah

>>6106851
N64 can't compete with the PS1 when it comes to wiggling jiggling polygons

>>6104027
you didnt have HD video back then dumbass

>>6103987
>Thanks to Digital Foundry
https://www.youtube.com/watch?v=QDiHgKil8AQ

>>6106905
Except the morons in this video don't know the difference between anti-aliasing and dither filtering

>>6103987
you mean a fucking crt

god your dumb

>>6106872

>>6106851
The crispiness of PS1 is fucking kino. Shame about the loading times

>>6106872
>HD video for something rendered in 240p

>>6104509
nigga what lol

>>6104315
The N64 only did in western countries.

>>6103987
You could do the same with PS1 games using the smoothing filter on a PS2 and PS3.

>>6107451
wtf is this picture supposed to even be? why do people attach random pictures that have nothing to do with anything when they post

>>6104047
>lazy texture work
They didn't have the space for actual texture work, it was about making most of what they did have.

>>6104546
>I always joke with my girlfriend
Things that never happened: The post.

>>6108842
>why do people attach random pictures
makes their post stand out

>>6103987
You're a fucking retard, OP. Fuck off with your youtuber cancer.
We've known it had AA for years. There's even a document in the oman archive that spells it out for you. It's likely in the normal developer documentation as well.

>>6109670
If its not on youtube, zoomers don't know about it.

>>6104040
No load times and no games. Based.

>>6109671
I'm the one who posted that oman pic.
I was born in '96 so by some accounts I am a zoomer. That's not why; it's because they're too retarded to look for anything that isn't spoonfed in video form.
More recently, Phil Gossett (the guy who wrote that doc) gave some sort of presentation where he touched on AA on the 64. So again, not DigitalFoundry.
https://archive.computerhistory.org/resources/access/text/2019/08/102781588-05-01-acc.pdf

>>6109680
Thanks for posting that link, anon.

>>6103987
we've known that since the Everdrive on N64.
https://youtu.be/QDiHgKil8AQ?t=280

>>6110157
Left easily looks the best.

>>6110160
I don't actually disagree. On a CRT as the dev intended looks better too, plus a lot of N64 games used dithering, any game that did looks like shit with the filters removed.

>>6110164
SF64 has one of the most aggressive dither filters out of any N64 game. As soon as the game starts you can see an almost film grain like effect on the black background. Now I see why.

>>6110157
see
>>6106921

>>6110160
No it doesn't, it looks terrible.