/vr/ - Retro Games

They use clever tricks to only render what is exactly onscreen. That's why every stage has either a ton of turns (to obscure stuff up ahead) or fog (to not render stuff too far ahead).

Remember those demos of whenever a console gets released? The PS2 demo with the ping pong balls, etc? Well, it's easy to make good graphics when you only need to render a single room.

It's probably something due to the artists, can't really put my finger on it... umm maybe well designed not sure

>>5455946
>>5455941
There's a cool blog series by one of the devs that takes about everything in detail. Fascinating read.
https://all-things-andy-gavin.com/2011/02/02/making-crash-bandicoot-part-1/

>>5455941
>The PS2 demo with the ping pong balls
wasn't that xbox?

>>5456014
meant for >>5455945

PS also has trouble showing 3d accurately and things become warped/curvy. Most things in Crash are made intentionally curvy so this warping effect is not noticeable.

>>5456017
The warping effect doesn't occur in Crash because the camera is on rails iirc.

Lets see:
1. Figured out polygonal coloring didn't come with the eye cancer of textured polygons. Most characters are not textured, they are colored by polygons, because it scales where textures can't do that on PS1

2. Ran at a higher resolution, reducing aliasing problems

3. Precalculated LOD because of fixed camera. Which means you have a lot of background detail
4. Same procedure is used to work against the perspective warping being visible

There’s two things that are the biggest cause of low performance in these old games.

Point one is overdrawing too many polygons whether they are actually visible on screen or not. Why this is bad is obvious. It’s a waste of time and resources for the hardware to do shit the player can’t even see.

Point two is using depth management to calculate which polygons aren’t visible and just don’t draw them, solving point one. The problem is that depth management is itself quite demanding in hardware.

Where PS1 falters it’s either because it’s doing too much of point 1 and thus slowing down, or doing point 2 in a really sloppy inaccurate way leading to ugly glitchy visuals. Where N64 falters, it’s by doing point 2 too hard by using a z-buffer which blows the console’s memory bandwidth budget and is usually responsible for slowdown but resulting in an accurate 3D image.

Crash solves point 1 and point 2 by precomputing depth management on a powerful workstation and not on the PS1 itself. This means that the PS1 console will never overdraw a polygon and also doesn’t have to waste resources on sorting depth. Performance is accordingly maximised from the user’s perspective.

Why doesn’t even game do this you might ask? Because you can only precompute depth if the polygons are displayed on screen in a predicable way. It can only work for games that are on-rails, like Crash Bandicoot’s corridors. Giving the user any camera control whatsoever would break the technique used.

Wasn't there a N64 game that gained a massive advantage by not using the Z buffer?
How do you avoid artifacts without it?

>>5456075
the massive advantage was it ran at 30fps instead of 20 and looked like a game that would normally run at 15. The N64 had too many limitations to compete in 3D unfortunately.

>>5456075
>Wasn't there a N64 game that gained a massive advantage by not using the Z buffer?
There’s a few. World Driver Championship. Another one seems to be Indy and the Infernal Machine.

>How do you avoid artifacts without it?
Sorting by polygon or by mesh, like the PS1 does. Nintendo demanded any such algorithm be of higher precision than what the PS1 used though or they’d bounce the game. That’s why polygon draw order errors are quite rare in WDC.

>>5456081
>The N64 had too many limitations to compete in 3D unfortunately.
Given the N64 is entirely limited by memory bandwidth, Crash with its pre-computed depth management would probably run at 60 FPS on the N64.

>>5456081
>The N64 had too many limitations to compete in 3D unfortunately
Yeah...no. Bullshit m8
Get some technical background before you spew uneducated BS

>>5456084
>By mesh
It's hard to imagine that this is even close to being sufficiently accurate. How do intersecting objects and complex non-convex geometry not break down?

>>5455941
1. Every PS1 game looks better than N64 game.
2. Crash graphics were sort of pre-rendered.

>>5456120
>How do intersecting objects and complex non-convex geometry not break down?
They do. That’s why it looks like flickering crap. By mesh is probably an extreme example. By clusters of polygons is probably the most common.

>>5456123
>1. Every PS1 game looks better than N64 game.

>>5456142
What game?

>>5456142
world driver championship
actual console capture

>>5456163
world driver championship
actual console capture

>>5456093
Perhaps you should substantiate your opposing viewpoint before launching a full-blown attack.

Could the N64 have been a killer system with the price tag it carried, or was it just too cheap for its own good?

>>5456191
It was ultra cheap hardware. N64 only used a single 9-bit memory bus. PS1 used two 32-bit buses and an 8-bit bus.

I don't think it looks that good.
Also the whole game is corridors, and very samey.
I prefer Jumping Flash.

>>5456176
I just tried it out on my everdrive and your webm is accurate. It looks nicer than roadsters which I had thought was the best looking car game on 64.

>>5455941
Crash games look the best on ps1.

>>5456087
Not to mention low texture cache, filling polygons (blurring), low draw distance (the notorious fog), it'd be perfect for N64. It's rare the PS game that can look as good as a N64 game.

>>5456123
>1. Every PS1 game looks better than N64 game.
Did your brain fall out of your head somehow?

>>5456013
Here's the real answer, and nobody has paid attention to it.

>>5456438
>Not to mention low texture cache
Which is twice as big as the one in PS1.
I don’t even understand the rest of your post.

>>5456210
The n64 was weird. It was really cheap in some places, but in others it was way better than anything else out there at the time. It's not really very fair to compare it other systems because it's not necessarily more or less powerful than them, just really good at some things and really bad at others

>>5456470
If the N64 had two 9-bit memory buses instead of one, then it would have thrashed even top PCs, let alone the PS1. That’s all it needed. Nintendo REALLY cheaped out.

>>5456142
This is a beautiful game but it doesn't help you posted one that shows fog

>>5456492
WDC doesn't use distance fog but atmospheric perspective fog, which is more demanding on hardware than no fog.

>>5456013
This.
Rather than going with textures, a lot of Crash utilizes colored polygons.

Crash games strain the Playstation hardware a lot. If you play the games too often, you'll have to constantly repair your PS1.

>>5456646
Lol weak system.

>>5456646
This is true because Crash games are constantly reading from the CD, and the developers knew how many reads the CD player was rated for in a lifetime ahead of time.

>>5455941
It runs at almost double the horizontal resolution of most ps1 games, also good art direction helps.

>>5456465
texture cache is a wholly different beast on PS1 -- on N64, textures must be in cache (really, it's just texture memory, referred to as TMEM). The programmer manages TMEM.
On PS1, the texture cache is just a cache, which gets checked when you draw a texture, and loaded into when it's not there. It's automatic.

>>5456646
Emulators win again, baby.

The Crash developers really, really knew what they were doing when it came to programming for the Playstation. I'd go as far to say it's one of the most impressive things on the entire console from a technical standpoint.

>>5457139
Nope the remakes win. Bye.

>>5457108
That’s just a difficulty of programming issue then. The fact is that N64 can render 4 KB textures without reaccessing RAM, while the PS1 can only render 2 KB textures without reaccessing RAM.

>>5456018
that's not true. if the camera moves relative to an object at all it becomes a problem

>>5457812
>That’s just a difficulty of programming issue then.
well, yes
damn near everything about the N64's effective performance is related to difficulty of programming
It's not really a secret that the N64 is dramatically more powerful than the PS1...
...except the framerates, texture quality, and model detail of many N64 games seem to obscure that fact. They shouldn't be merely on the level of PS1 games but with more accurate rendering (and getting by a bit better because you don't need to subdivide polygons to reduce warping), they should be absolutely chewing the PS1 to bits. The graphics system is real SGI hardware, compared to the cheap hacked up "speed above all else" shit in the PS1. The CPU is literally a more advanced version of the PS1's. Same family, but a 64-bit model and at several times the clock rate.

and there are several N64 games that do properly trounce the PS1 in terms of texture detail and model quality, but they're much too far and few between

>>5456142
like, damn that's pretty
closest thing on PS1 is Ridge Racer Type 4... and you basically have like 2 cars on screen at any point there

>>5457857
Almost every programming problem on the N64 is related to the unified single channel memory. How well developers manage this singular issue has a drastic impact on performance.

>>5456142
Woah Woah! Thats an N64 Game???

>>5457876
yep

>>5456081
>The N64 had too many limitations to compete in 3D unfortunately.

>>5457171
Not retro

>>5457880
1 in 5 N64 games is a racer.

>>5457903
>>5456093
>>5456087
>framerate isn't important gang
even the saturn has better 3D.

>>5458380
>even the saturn has better 3D.

Man some underage seems to be samefagging a lot in this thread, "Saturn has better 3d graphics than N64", like that is any way based on reality.

>>5458383
Fuck you too, dude.

>>5458383
If it was on the N64 it would be running at 17 FPS

>>5458420
Yes, Mario 64 really demonstrated the N64 really struggling with those 2D billboard trees

>>5458330
1-1 graphical remaster of 1996 game in mentioned briefly
GGHHNNOT RETROOO TRIGGGGGEEEEEERED

>>5458330
>1-1 graphical remaster of 1996 game is mentioned briefly in a thread about that game's graphics
GGHHNNOT RETROOO TRIGGGGGEEEEEERED

>unironically defending the saturn
you have to go back.

>>5455941
Incredibly limited field of view, narrow space, and limited areas.

with low fps even 3DO had the capability to do great 3D with small narrow fields.

granted this was one of the most visually impressive 3DO games, it still ran fine. What PS1 did was not that impressive for the time.

>>5456093
>>5457903
Seething nintoddlers

>>5458420
The N64 was slow, but not THAT slow.
Saturn is unambiguously inferior.

You look better than most 64 games

>>5458613
You have to be 18 or older to post here

>>5456084
the PS1 didn't use an algorithm for that. programmers had to write their own. certain games did it extremely well (crash bandicoot, for example, because it didn't do the sorting at runtime)

>>5458447
People defend this place rigorously because it's not like they have many other options to turn to if they want to talk about retro games. If you want to talk about shitty fucking remasters you can go ahead to the thousands of options available for that.

>>5456013
I've read all of it.
It's a fascinating read.

>>5460679
>talk about
*briefly mention
this is why people hate this board
people like you give us a reputation for being extremely sensitive and easily triggered

Extremely aggressive occlusion culling.

If you load it up in an emulator and try to force a widescreen hack it looks awful because the game outright refuses to render anything not directly on camera.

>>5462350
it's even more evident on the side scrolling sections

>>5456018

>>5456018
You don't understand graphics, GTFO please

>>5458447
>1-1 graphical remaster
>1-1
lol

>>5460836
I do it just to make you mad.

>>5456014
Yes, the PS2 was rubber ducks.

>>5460653
>the PS1 didn't use an algorithm for that
Actually the GTE does built in polygon sorting functions with set algorithms and levels of precision. Of course how those functions were actually used were up to the programmers of each game.