/vr/ - Retro Games

>>8941008

>>8941008
2D>3D

>>8943048
This is only “bait” to discuss the VDP2 chip, friend. Not enough /g/ talk around here.

>>8943056
This.

>>8941008
If the 5th gen had focused on 2D more, it would have been better remembered. It's the shittiest generation outside of the 1st and 2nd. Wave Race, and Chrono Cross notwithstanding.
Making Neo-Geo graphics "affordable" at home would have won a lot of hearts.
5th gen is to 3D what 2nd is to 2D.

>>8943063
While there is no lack of fondness for gen5 and it was most certainly not shitty, I do agree that more games should have looked like Metal Slug, Symphony of the Night, and Final Fantasy Tactics, rather than janky, twitchy, blurry 3D. But that would have been a terrible business decision, as games like Final Fantasy VII and Tekken 3 were massive system sellers. Ultimately, emulation cleans up early 3D a lot, so we won in the end.

>>8943063
Even faux 3D needed more time to shine. No one ever made a mode 7 kart racer with 100,000 scaling sprites on the field and multiple mode 7 planes. Could have been dope.

>>8943063
>If the 5th gen had focused on 2D more, it would have been better remembered. It's the shittiest generation outside of the 1st and 2nd. Wave Race, and Chrono Cross notwithstanding.


Out of all the companies, Sony was pushing 3D the hardest with their early PS1 releases. They did prioritise 3D over 2D in the NA market. Even though they weren't completely against 2D releases. Especially if they were well known arcade fighters. But 3D was the major selling point of the hardware. Or at least elements of 3D (like models on pre-rendered BG's). I think that caught Sega off guard, because I do think they were aiming for a 2D and 3D machine. I think the PS1 put Sega of America into a position, where they had to also focus more on the 3D capabilities of the Saturn. The N64 is interesting in that it was conceived from the start as a 3D machine. MIPS/ Silicon Graphics went to Nintendo with their hardware. Nintendo really wanted the machine after the SNES to focus on 3D, and apparently they didn't know how to do it. They were probably really interested in what MIPS showed them. Especially after the success they had with SGI workstations and DKC.

>>8943091
>Even faux 3D needed more time to shine. No one ever made a mode 7 kart racer with 100,000 scaling sprites on the field and multiple mode 7 planes. Could have been dope.


https://www.youtube.com/watch?v=mDh3WOu0V48

>>8943115
The Saturn was designed specifically to do 3D, it was just bad at it. Relying on 2D instead was an unintended fallback.

>>8943143
Nah 3d was a last minute addition to the saturn where the saturn hacks 3d from sprites

>>8943206
no, stop this lie

>>8943206
Saturn had been a polygon-based console since the announcement of the Playstation, over a year before either of them launched.

>>8943120
the effect looks super fuckin nice. really smooth but this game is nauseating. if the tracks were bigger and better designed then it would look awesome. this just looks like a really cheap bootleg of power drift inbred with mario kart.

The early 90s was the worst time to release a console. Your product would be hopelessly outdated in just 6 months.

>>8943291
wtf does the MD, SNES and PC TG16 have to do with this thread?

>>8943291
>The early 90s was the worst time to release a console. Your product would be hopelessly outdated in just 6 months.
you're right. was a big time for change and competition was crazy. however, there was a lot of shit that was released and wasn't destined to survive.

>>8943295
>tried to be funny but made himself look like an idiot

>>8943307
>tried to be funny
what?
All those systems are early 90's systems in the west

>>8943312
The gaijin versions of the Mega Drive and PC Engine were released in 1989. Try again, champ. I'm sure you'll be funny next time.

>>8943319
and they're all the only consoles associated with the early 90s

>>8943291
You mean the mid 90s consoles. early 90s held up well.

>>8943360
No, I don't. The Sega CD, Sega 32X, Commodore CD32, Atari Jaguar, 3DO, PC-FX did not hold up well and were outdated within months.

>>8943373
>VDP2
>Anything to do with any of those consoles
?

>>8943373
>32x
Late 1994
>CD32
Late 1993
>Jag
Late 1993, 1994 globally
>3DO
Late 1993, 1994 globally
>PC-FX
1994

Those are all mid 90s consoles bro. Only the Sega CD released early and that one was half decent.

>>8943386
Do you think Sega pulled the VDP2 out of their ass or something? These are the consoles that were on the market when they were designing it.

>>8943390
>I really, really want to argue semantics. It's so fun, bro!
Sperglord

>>8943397
epig backpedaling m8

>>8943398
Autism

>N-NANI!? Did this ne'er-do-weller just suggest a year (A YEAR!) away from the middle of the decade was the EARLY part of the decade? On MY Laotian basket-weaving forum!? Oh the humanity! The sheer audacity! I must spend my day confronting this IMMEDIATELY! This injustice will not be allowed to stand on MY watch, knave!

>>8943445
stay mad

>>8943063
Cringe take that's always wrong.

>>8943056
Nope.

>>8943120
This is an enhanced port of the SNES game and it really shows. Just look at that draw distance.

>>8943230
Saturn was initially conceived without knowledge of the PS1 and N64. In the minds of its designers, it was going to be a next-gen SNES/Genesis.

>>8943924
work on the Saturn started in 91 or 92 so they certainly wouldn't have anticipated 3D back then.

>>8943924
It changed when the PlayStation was announced and again when Sony revealed (somewhat bogus) specs for it. The first change was not a last minute change given the lead time, and that's when it was decided the Saturn would be a polygon based console.

t. hideki sato

>>8943948
You haven't engineered anything for mass deployment. One year lead time is last minute, ESPECIALLY if it's after a reactionary pivot.

>>8941008

its quite weird to find someone on /VR that gets old hardware by what they where.

sega put too much faith that a strong "sprite + background layers" architecture was the best bet. (taking design aspects from previous arcade board designs of theirs)

VDP2 is AWESOME if your game design aims to have multiple background layers, and sega was probably still butthurt of mode 7 (that was implemented in a cheap snes asic at the time) so it tried to pull that off in sega cd (limited success) and fully in software in 32X (mixed result)

so yeah, finally we have mode 7! and most 3D games that have infinite flat planes use VDP2 to draw them, arguably helping with 3D... but its clear that looking back in history it was a waste of transistors that could have been used into a stronger unified VDP1.

the PSX GPU (GTE not included) has around 300.000 transistors
source: http://psxdev.ru/

VDP1 alone has more transistors than the entire PSX GPU, VDP2 probably had at least half of the count of VDP1... no wonder why it was more expensive!

so yeah, they spent more than twice the silicon area to pull off things like shitload of sprites and multiple background layers and infinite flat planes (mode 7).

>>8943964
Sega simply wasn't able to predict what the average 3D game would look like. A dedicated background processor is brilliant if your game has a limited perspective like a 3D 1v1 fighter or a rail shooter, but it shits the bed in explorable 3D environments with a free camera like Tomb Raider.

Also
>they spent more than twice the silicon area to pull off things like shitload of sprites
PlayStation can push more sprites than Saturn.

>>8943056
Yup. Nothing beats true hardware 2D in 240p. Nothing.

>>8943964
>>8944000
Great analysis

>>8944000
>PlayStation can push more sprites than Saturn.
And even in comparisons of some 2d fighting games where the PSX has fewer frames than the Saturn version it's not so much that its GPU can't animate them at full speed, but rather it just didn't have enough total RAM to hold all of the assets and also have room for the front and back buffers. The PSX's graphics chip was a fillrate monster compared to the competition, it seems; And that's with one IC and a single block of RAM dedicated to video.

>Nintendo troll Sega by putting Mode 7 in one small IC
>Sony troll Sega again by accomplishing usable 3D in one small IC
they don't think it be like it is but it do.gif

>>8944360
I do wonder what a PlayStation with a RAM expansion would have looked like. Both it's competitors had one. I think it could have matched the Saturn for 2D fighters.

>>8944373
Also didn't they have to limit the Mega Drive's color palette and drop sprite scaling because they couldn't fit it in the VDP? Seems like Sega didn't have the best engineers around.

The best graphical moment on Saturn, and perhaps one of the best of fifth gen, is pic related. This shitty gif doesn't quite do it justice, but it's easy to see how it's unique for its day. This is possible because of multiple mode 7 layers via the VDP2.

>>8944373
But the SNES PPU consists of two ICs.

>>8943206
Every hardware 3d renderer transforms 3d objects into 2d shapes, because they draw to a 2d screen. Educate yourself before you speak.

>>8944414
it's a lot better and more cleanly designed console on the whole than the SNES which is a lesson they apparently forgot with the Saturn

>>8944414
It was probably more of a limitation based on the transistor density of the nodes they had available at the time. The Genesis VDP was probably designed for a 1.5 or 1.2 micron die fab in '88, whereas the SNES was almost certainly on a smaller 1 micron fab for '90 and its transistors were originally spread out between two ICs like >>8944421 mentioned.
Nintendo was probably getting 50 to 100% more transistors in the same amount of area compared to Sega.

>>8944414
There was a big memory shortage right around then, which contributed to some of the restrictions of the console. There's a 128k VRAM mode that was disabled by default on the production console, you can enable it, which causes some interesting memory interleaving effects because the second 64k of memory isn't installed.

>>8944434
He's more correct than you think, texture mapping with distorted sprites can justly be called "hacking 3d from sprites" (because it sucks and you only do it if you're too fixated on sprite-based hardware engineering to design good polygon hardware)

>>8943056
true but not in vidya

>>8943063
PS1 has a lot of 3D games I fondly remember.
Spyro, Crash Team Racing, XenoGears...
There are a lot of rough games out there though.

>>8944414
The Genesis was going to have sprite scaling? Damn, they really fucked up not shipping with that, would have been a game changer.

>>8944453
While I do agree somewhat, I know that 99% of the time when someone says that, they think using quads is fundamentally not really 3d or something. Triangles aren't any more 3d, and the reason why saturn, and ps1, don't have 3d hardware is not because of their choice of primitives, but because their graphics chips only process 2d coordinates.

Actually, you might be able to argue vdp2 is the most "3d" chip out of the two consoles, because it does handle depth correctly.

>>8944414
>>8944465
From what I understand it was either that or Master System support. The VDP houses the PSG sound hardware so possibly that would've needed the boot instead, which also would've limited the sound on the Mega Drive, even if it's just the square wave/noise channels.

>>8944468
You're probably misunderstanding them, because clearly the point is that VDP1 was designed to push 2D, sprite-based experiences, which is the basis for using quads. If it was designed for 3D, it would have used triangles like everyone else did.

>>8944501
Not everybody else used triangles, Nvidia used quads. I'm not misunderstanding, as the post I responded to was too short to have any idea of what the poster knew, I just know that 99% of people on here don't understand what they are talking about, and why one approach might be better or worse than another. So it's possible they know why not having proper UV coords sucks (and is why you see a ton of squished textures on clipped polygons in saturn games), but I don't know that.

>>8944468
Quads vs triangles is a layman-friendly oversimplification of the distinction between forward and inverse texture mapping. The choice of rasterisation algorithm dictates the polygon primitive along with other important characteristics.
Designing 3D from the ground up without the bias of a 2D hardware designer, inverse mapping is the only sensible way to do it. Forward mapping is riddled with problems and only sounds reasonable if you're stuck in a 2D mindset or trying to retrofit a 2D chip for 3D.

As for what counts as "more 3D" - I'd say being able to UV map is a pretty big deal for producing convincing 3D. The exact set of features you need to count as "real" 3D is largely a matter of opinion anyway.

>>8944514
>Not everybody else used triangles, Nvidia used quads.
Nvidia used the VDP1!

>>8944518
>Forward mapping is riddled with problems and only sounds reasonable if you're stuck in a 2D mindset or trying to retrofit a 2D chip for 3D.

Probably true, but I don't ascribe that much understanding and subtlety to the avg chan poster.

>>8944527
Yes, as I said it often gets boiled down to muh quads. Which is related to the core issue, but not exactly the same.

>>8944524
From what I understand, the NV1 chip is very different from VDP1, handling curved quads and such natively, so I don't think they are the same chip at all. People probably came to the conclusion that they are the same because they both use quads, and certain saturn games were ported to pcs with NV1 cards.

>>8944538
For example
>NV handles not only four vertices per polygon, but also five additional points of surface to define real curving.

VDP1 definitely doesn't support that.

>>8943326
>moving goal posts

>>8944538
It came with a Saturn port. It was undoubtedly based on the VDP1.

>>8944614
We know that Sega and Nvidia did have some sort of partnership around that time, but the actual graphics chips are very far apart in terms of capability. Why do you insist things are true that you cannot possibly know for sure?

>>8944661
High-IQ pattern recognition and experience

>>8944691
list the similarities between VDP1 and NV1 that tipped off your pattern recognition.

sgi ruined gaming

>>8944696
>quads
>has a sega saturn controller port
>saturn games immediately released for it
>NV2 was designed to be the Dreamcast GPU
>NV2 is an upgrade of the NV1
keep saying the NV1 was developed in isolation of the VDP1, but it's just flat-out naive

>>8944465
It was mostly a simplified, cheaper variant of the System 16 hardware which did include scaling and a bigger color palette.

>>8944717
The saturn port is handled by a separate chip from everything else. And the saturn games had to be converted to run on it, it wasn't like they immediately ran on the graphics hardware once the code was converted to run on a pc.

I am finding sources that say both that Sega helped Nvidia with the design (from Sega retro of all places), and sources that say Sega had to be convinced that quads were still a good idea after their failure with the saturn.

>>8943915
>This is an enhanced port of the SNES game and it really shows. Just look at that draw distance.

There was also a version for the PS1 as well:
https://youtu.be/JCgqI2EwI7s?t=203

The PS1 game looks like it is using a polygon playing field, while the Saturn game looks like it is using the VPD2 chip for the ground. Street Racer was a Mario Kart styled clone made by UBI Soft for the SNES. But they also developed ports for the Mega Drive, Game Boy and even the PC.

>>8943285
>power drift

https://www.youtube.com/watch?v=04FtxXaeBnY&t=159s

There is a nice port of Power Drift for the Saturn. Though it doesn't use the Saturn's flat 3D playingfield. One saturn game that does come to mind for me is Shining in the Holy Ark:
https://youtu.be/RG-jaLAMXLw?t=17832

Which is a first person dungeon crawler. And it uses the Saturn's 3D playing fields for the ground, with 3D 'polygon' objects filling in the details.

>>8944909

Just about every 3D fighting game on the Saturn (that came out after VF1) used the VPD2 3D playing field effect:
https://youtu.be/d8Wr6zT_9QA?t=265

Virtua Fighter Remix, Virtua Fighter 2, Virtua Fighter Kids, Fighting Vipers, Virtua On, Last Bronx, DOA1, Battle Arena Toshinden Remix, etc.

>>8944000
> Sega simply wasn't able to predict what the average 3D game would look like
>SEGA model 1 and 2 exist before the saturn

>>8945073
Please read the post thoroughly.

>>8945225
Please play actual games, tomb raider does not shit the bed

>>8941008
You can actually use coefficient tables, windows, etc. to set up a pretty complex 3D scene with VDP2 alone. Grandia and Radiant Silvergun both do this.

Grandia uses a single VDP2 RBG layer and then defines windows within that layer. Each window then has different coefficient tables which allows the portion of the layer to have scale and rotate differently from the rest of the layer. This is how certain maps are able to have multiple levels without using polygons:

https://www.youtube.com/watch?v=bSHJ8x7L2rM
https://www.youtube.com/watch?v=AOsUdG8P9pg

>>8943206
The Saturn was intended to be 3D from the start, Hideki Sato has confirmed this in interviews. The reason it uses a sprite based approach is because no one on the design team had experience with 3D and they weren't allowed to ask AM2 for help. So they did the sprite approach as that's what people at Sega had the most experience with.

The last minute add-on wasn't VDP1, and it wasn't VDP2. It was the 2nd SH-2. And that happened around late 1993.

>>8945771
although the videos look kind of neat it was so much more work to produce decent results with the Saturn's balky architecture and it still looks chunky due to the square polys.

>>8945786
You realize you can still draw a triangle on Saturn right? Sure if it's textured you might need to do some work with how your texture is designed, but if it's flat or gouraud shaded it can do it just fine without any effort. And then there's the fact that during this time quite a few games were actually using Quad based renderers. The original Tomb Raider engine used quads on not just Saturn but PC as well. I'm pretty sure Quake was also Quad based on PC when running in software mode.

And the Saturn really isn't that complicated in comparison to other consoles before and after it. The PS2 and PS3 are both convoluted beasts to deal with, as was the Super Nintendo before it, and even the N64 isn't that easy to work with either. It's just that PS1 was an insanely simple and easy to use design compared to most other consoles.

>>8945817
The PS2 was deliberately convoluted because Sony responded to programmers' complaints that the PS1 was too easy and they didn't feel challenged, as there were quite a few Carmack types out there who got off on the idea of bit banging hardware registers.

>>8945821
Then why didn't those programmers develop for Saturn? The reality is that complexity doesn't mean jack shit when it comes to developer support. All that matters is if there's enough of an install base to justify supporting it.

At the end of the day the Saturn's hardware isn't really that big of an issue. It's more than capable enough to hold it's own for it's generation and it's no more complicated to work with than other systems before and after it. What really mattered was how horribly it was handled outside of Japan, which led to it not having an install base outside of Japan.

>>8945821

>>8945771
>they weren't allowed to ask AM2 for help.

Wait, really? Why?

>>8946132
because Japs are basically retarded

>>8946132
Because AM2 was busy making Arcade games that were actually making a boatload of money at the time. This was when that team was in the middle of making Virtua Fighter, Virtua Racing, Daytona USA, and Virtua Fighter 2.

>>8946132
>Not allowed to ask AM2 for help
>Not allowed to use Nights engine because Naka
>No developer tools or documentation
>No Anime games are allowed in America despite the huge surge in popularity boom.
>SOJ jealous of SOA success fuck you baka gaijin
No wonder the Saturn failed. Its like Sega deliberately sabotaged its own system out of hubris.

>>8946132
I'm playing Virtua Cop on my Saturn right now mate

>>8944490
I don't think its the PSG that is eating up all the die space, but rather the Master system vdp mode.

>>8945451
Saturn does shit the bed playing tomb raider, and that's one of the earliest and least demanding free camera 3D titles.

>>8946570
please play a game before you talk about it

>>8946573
Stop coping any time.

>>8946224
>Not allowed to use Nights engine because Naka
Sonic Xtreme was a doomed project long before this happened. The NiGHTS engine wouldn't have saved it.
>No developer tools or documentation
This was purely Sega of America's fault. Japan had that, but Sega of America was jerking off to the 32X instead of getting Saturn tools and documentation translated and out to western developers.
>SOJ jealous of SOA success fuck you baka gaijin
This is an old lie that doesn't hold up to scrutiny. Sega of America had Sega of Japan's full support. They let them do as they pleased. It wasn't until the 32X project failed that Japan started to come down on them a bit because of how damaging that miscalculation by Sega of America was. They then started to become more aware of a bunch of shit Sega of America was doing wrong that they weren't aware of. That's why the hammer started to come down during the Saturn years.

For example, after Kalinske left Stolar actually wasn't the CEO of Sega of America, it was Shoichiro Irimajiri from Japan. He found that Sega of America had warehouses full of unsold Genesis hardware and Games because they were overproducing them. They were convinced it was still selling like it was 1992 and refused to accept the 16-bit era was over. Irimajiri put an end to that and was forced to write off all that unsold stock around 1997/1998. This is why Sega posted such a huge loss in their Fiscal Year of 1998 report. It was writing off all that unsold stock from Sega of America.

>>8946587
The Demo mode isn't really a good thing to use. It uses levels that are some of the worst performing on Saturn. There's other levels in the game where Saturn performs on par with or better than PS1. This results in the games really being more on par with each other than people think. That's what the person you're replying to is trying to explain but failing at.

>>8943120
this game plays like ass, nice sprite work but just look at the absolute lack of momentum... it feels like youre driving at 3mph, zero sense of speed, the tracks look tiny.

>>8946587
>youtube video
>still not playing the game
nintendo fanboys are the worst

>>8946972
Tom Kalinski thought they could milk the Genesis until 1998.

>>8946972
>For example, after Kalinske left Stolar actually wasn't the CEO of Sega of America, it was Shoichiro Irimajiri from Japan. He found that Sega of America had warehouses full of unsold Genesis hardware and Games because they were overproducing them. They were convinced it was still selling like it was 1992 and refused to accept the 16-bit era was over. Irimajiri put an end to that and was forced to write off all that unsold stock around 1997/1998
really? all the manufacturing was done in Japan. you mean to tell me in all this time nobody in Japan noticed their production lines were making a glut of US spec consoles and shipping them?

>>8947093
The NES, PS1, SNES lasted a decade the Atari 2600 over a decade so...

>>8947190
which is unfair because the 2600 was seen as passe by 1982 yet Atari management for similar retardation were unwilling to let it go

>>8947098
Not all of it was done in Japan. A lot of the consoles were being produced in China and Mexico towards the end, and many of the games were actually being made in the US and Mexico.

A prime example of this over production is Shining Force II. Apparently Sega had tons of copies sitting in a warehouse because they severely over-estimated demand. A lot of copies of Shining Force II and other late US games if you flip them over will say "Made in the USA" or "Made in Mexico".

>>8947190
>Atari 2600
See previous.
>NES
in Japan it was considered passe by '89 once the PC Engine was out. the peak year for Famicom sales was '86.
>PS1
Budget titles still coming out until the mid-2000s but the peak years for the console were 97-00.

>>8947195
Atari also had major issues with overproduction.

Nintendo seemed to have better control of what was going on. They manufactured everything in Japan where they could keep an eye on it and NOA had a lot less autonomy than SOA did.

>>8947190
Sure, but they didn't keep selling at the same rate either. There's a point where they become a budget legacy system and sales drop significantly and the newer systems pick up steam. Realistically the Genesis was at this point, but Kalinske and the rest of Sega of America were deluded thinking it was still 1991/1992.

Look at actual shipments and sales by year for the NES, SNES, PS1, 2600, etc. With NES once the SNES comes out you see it drop from 7.5 Million systems sold to 2.5 Million sold. The next year it's down to 1.3 Million, the next 600k, the next 100k, then 10k, then it's done. SNES we see a similar thing. In 1995 with the Saturn and PS1 on the horizon SNES sales drop from 5.4 Million to 1 Million. The next year we see a slight surge back to 2 million, but the following year it's back to 1 million and it slowly trickles down until it's discontinued. We see the same pattern with the 2600 and the PS1. And we see the same pattern with the Genesis.

The Genesis was over and it was time to transition it to a legacy system. But Sega of America refused to accept that.

if you played a game like Spot Goes To Hollywood that's on Genesis and PS1 the latter looks so much slicker and more advanced that you know the former's time was up.

>>8947208
Nintendo had to scramble to develop a next gen console when the PCE and Mega Drive showed up, and ironically the NES was peaking in the West just as Japan was moving on to next gen systems.

>>8945835
> it's no more complicated to work with than other systems before and after it
t. Someone who has never done significant work on a saturn title

>>8947236
the Saturn is definitely a very awkward mess and the lack of adequate programming tools was an issue. the PS1 was so much easier, its programming model was more like coding a Flash game on a PC today and Sony supplied proper technical info to devs.

>>8946986
>nintendo fanboys
meds, now

>>8947240
AFAIK Sony didn't bother explaining how most of the hardware registers in PS1 worked and stuff like that, they expected you to use the API and pre-provided libraries. Naughty Dog basically said yeah fuck that shit and they went and did bare metal coding anyway.

>>8943074
>Ultimately, emulation cleans up early 3D a lot, so we won in the end.
No it absolutely fucking doesn't. Early 3D (and even many sixth gen titles) are some of the few that look absolutely terrible with upscaling and there is no actual way to make them look good short of rebuilding all the assets from scratch. The only real way to play them is native red on CRT or with CRT filter. 7th gen is when games started to upscale pretty well. About the only way emulation improves it is pixel wobble (and even that is arguable, and also requires manual per-game configuration)

2D games will always be timeless though.

Something like Spyro with an open world you go around in just wasn't happening on the Saturn, it couldn't do it. Not very easily anyway--even the best programmers in the world would struggle trying to pull it off.

>>8947236
I've actually done some Saturn development. It's not that hard and the libraries Sega provided aren't that tricky to wrap your head around either.
>>8947240
Yes PS1 is easier, but the point is that Saturn isn't THAT difficult to work with int he grand scheme of consoles. It's no worse than the SNES before it, or the N64, PS2, PS3, etc. Yet people who have never done any bit of software development or even touched the system act like it's this unprogrammable buggy mess that's impossible to work with.

The only systems I know of that fall into that category are the Atari Jaguar, and possibly the Frankenconsole that is the Sega CD 32X.

The main issue with Saturn development difficulty comes out of the US and the Europe because Sega of America didn't do their job and get manuals, libraries, etc. translated in time and didn't get devkits to developers in time. Japan had those things but the rest of the world didn't because Sega of America was convinced the Genesis and 32X was the future, not the Saturn.

>>8947273
man, a team of programmers literally almost died trying in vain to get Sonic X-Treme running on that shitfest.

>>8947276
That team almost died from their own retardation more than anything.

>>8947276
I'm going to blame that on the team having no fuckin idea what game to make more than the hardware itself being so bad.

>>8947276
That has more to do with incompetent management than the actual hardware. Sonic Xtreme's development is just a series for red flags that anyone with any project management experience would facepalm at:

> Has the lead developer work from home on the engine without ever looking at the Saturn hardware and has never done console game development before.
> Lead developer makes the engine entirely on PC thinking he can just recompile it for Saturn and it will be fine.
> February of 1996 comes around, they have to show what they have running on Saturn.
> Recompiled PC Engine runs at 1fps on Saturn. Absolute panic ensues.
> Team is split into main game and boss sections. Boss section is run competently with people who have some clue as to what they're doing.
> Main game section is completely lost and keeps failing. Sega calls in POV software to try and get things up to par. Lead developer loses his shit over this.
> Eventually Sega of Japan tells them to drop the failed main game engine that isn't going anywhere and instead focus on the promising boss engine.
> Team working on the main engine decide to instead of help the rest of the team, to go dick around on turning their dumpster fire into a PC game instead, leaving the rest of the team out to dry.
> Boss engine Team keeps trucking along but is having trouble. This is where the NiGHTS engine situation happens. It only lasts about a week and doesn't really amount to anything.
> Boss engine team keeps fleshing things out into a game but is clearly strained due to the other half of the team not helping and dicking around with their dumpster fire PC game.
> Key people working on the Saturn game almost die due to stress.
> PC game is turned down because it was still the same dumpster fire idea as before, just running at a higher frame rate.
> Sonic Xtreme is cancelled.

It's basically a text book example of bad project management.

STI clearly weren't up to a project of this size and they never should have been assigned to it.

>>8947308
No shit? SOJ should have handled it but they were burned out on Sonic and wanted to do something else.

>>8947249
Many PS1 games look absolutely stunning running in high res. Some PS1 games look better than modern games.

Sony's reach was enormous and Sega just couldn't match it. They could conquer all corners of the globe in a way prior console manufacturers never did. Nintendo focused on NTSC regions and Sega on North America and PAL regions but neither company had the resources to get their consoles everywhere. Especially when it came to conquering Europe where consoles had always lagged behind home computers, although Sony also benefited from the PS1 arriving just as Amiga et al were at the end of the road.

>>8947409
yeah Sony did figure out very shrewly that millions of Europeasants would buy a console just to play FIFA

there are people who've done some neat homebrew on Saturn with modern tools and unlimited time neither of which were available to developers in 1995

>>8941008
You aren't wrong. Bulk Slash was basically an SuperFX + Mode7 game on crack.

>>8947335

>>8941008
>you FAGGOTS like mode 7 huh

Never did, fucking eyesore

>>8944518
Hardware T&L, which is probably THE defining feature for a GPU and separates it from a graphics coprocessor/chip is probably where the technical line is drawn since the Nvidia GeForce 256 was the first GPU.

>>8948016
Then that's another way the Playstation's 3D is more "real" than the Saturn's.

>>8948236
Oh if we're talking about that, then no, the PSX's GPU didn't have hardware T&L, only the N64 out of all 6th gen consoles did and arcades. I would say the only thing in my mind that defines it here is the inverse texture mapping and triangles as primitives instead of quads.

>>8945835
>The reality is that complexity doesn't mean jack shit when it comes to developer support.

HUGE divide between:

"hard because with everyhing flexible, every register, step in the pipeline and blend modes can be tweaked"
VS
"hard because many things you do are hostage of a bad design choice constraint"

Pratical example of "good complexity"
* Psx could do framebuffer effects, with software driven blending and shading, reusing the framebuffer as a texture, but if you dont care about any of those things you can just write "simple and dumb" triangles to the screen using the libraries in the sdk provided.
* They didnt placed many "unneeded register limitations" so many tweaks that seem useless actually found great usage in some games. (i.e. the fast psx doom)
result: Many games look samey, some expert programmers can make really unique games.

Pratical example of "bad complexity"
* In the saturn, you can have transparencies, but not between objects draw in vdp1 but you can have them to be flagged to be transparent to bg layers of vdp2. You can draw a triangle, but it has to be a quad with the last vertice duplicated, what really fucks up some some aspects specially the final transparency to vdp2.
* You can use both cpus, but due to the day the data bus was designed, they interrupt each other to access the memory unless the access pattern is VERY well interleaved, limiting a LOT what you can do using both cpus. (many times its faster to just use one)
* You can have a very fast load time due to the SH1 dedicated ram (cdrom controller) but that memory is not mapped into the address space of the SH2's or the VDP's (that was REALLY retarded), why not just add more ram and leave to the developer to manage the read buffer, being a more complex work, but resulting into more available memory to work with. (good complexity).

post too long, spared of mentioning the non existent code examples sdk, and criminally bad debugger.

>>8943120
Thanks man, looks good.

>>8948373

as a minor follow up post, a link to an actual developer source talking about the provided "sdk" and debugger

https://youtu.be/BjhDXStrFdc?t=861

it is probably the cringiest shit i ever heard in software development, and he speculates that probably many games didnt used both SH2's because the debugger was not even available for that scenario.


https://youtu.be/FcwjgnaQXXU?t=3000

a sample developer insight on how the "dual SH2's" actually work in practice. Jim bagley (saturn doom port) did a amazing leap to use both SH2's running the same function with interlaved memory access by actually using the Motorola M68k to "orquestrate" them. Of course the M68k cant do everything else and both SH's dont run simultaneously on the non software rendering parts of the game logic.

source: https://archive.org/stream/gebbdoome/gebbdoome_djvu.txt

oh boy, i could hit another post limit shitting on that...

>>8948436
The development environment was improved significantly a few years into the generation, similar to the PS1 which was very obfuscated until better tools were developed. The pre-release tools Time Warner had available to them were the earliest development environment available and predates platforms like the Address Checker, Cartdev, Mirage etc. I'm not defending it but that was a key element of all consoles at the time, the earliest environments were not the most ideal - things got better as time went on and it was reflected in the software quality as developers and the platforms they worked on matured

>>8948747
Secondly Japan had better devkits and support at this time as well. Sega of America wasted time dicking around with the 32X when they should have been getting developers the tools they needed for the Saturn.

>>8948763
>Secondly Japan had better devkits and support at this time as well.
Absolutely, they also had direct access to SOJ while Tommy boy was twiddling his thumbs thinking the saturn wouldn't sell so why bother trying. It's not unusual for third parties to create their own dev hardware, but really the best tools were coming OUT in 95-96 after the console had been around for a year at least.

>>8948373
> In the saturn, you can have transparencies, but not between objects draw in vdp1 but you can have them to be flagged to be transparent to bg layers of vdp2.
No, you can have transparency blending to VDP1 objects or VDP2 objects, just not both at the same time without a bit of effort.
>You can draw a triangle, but it has to be a quad with the last vertice duplicated.
This isn't really an issue for the time. Especially when you consider that quite a few PC 3D engines were using Quad renderers, as were Arcade boards.
>what really fucks up some some aspects specially the final transparency to vdp2.
It only impacts VDP1 transparency due to pixel overdraw which an impact both triangles and warped quads. VDP2 is unaffected by this issue.
>CD-ROM block memory
PS1 has the same situation, additional RAM in it's CD-ROM block to act as a cache that's also inaccessible. I think it's around 128KB?
>non existent code examples sdk,
You can literally go download SBL and SGL disc images and Sega's DTS disc images and see they all have tons of examples, some dating back to long before the system launched in 1994.
>criminally bad debugger.
It's no worse than other debuggers of the time not including Sony. Sony just seriously changed the game and it's why they can be considered innovators for the industry here. That said, the really useful one that could trace your code and see where things were taking a long time wasn't a tool given to everyone. It was just one Sony had personally that they would allow certain high profile third party developers to use from time to time.

>>8948436
Most of the issues here again fall back on Sega of America dicking around with the 32X instead of getting things ready for developers to work on the Saturn.
>Jim Bagley
Is clearly a hack when it comes to Saturn Doom. I think the fact homebrew devs in less than a year have gotten Doom running in pure software mode at full frame rate and full screen on the same 2 SH-2s in the same master slave set up, but clocked slower with less RAM and no video hardware is proof he had no clue wtf he was doing.

>>8948292
The GTE counts.

>>8949330
How? It's just a coprocessor that's designed to do certain math calculations really fast. It doesn't really make it's 3D any more real or fake than Saturn. On Saturn its equivalent is the 2nd SH-2 and the SCU DSP. Both of which are used to help with 3D math and are quite fast and well suited for it. In fact the SH-2 is significantly faster than the PS1's MIPS CPU at doing multiplication and division operations which are quite useful for 3D math. Multiplication and Division operations on the PS1 can take anywhere from 8-12 cycles to do, while on a single SH-2 in the Saturn they take anywhere from 1-3 cycles.

>>8949350
>It's just a coprocessor that is designed to do T&L calculations really fast
Does hardware T&L have some different meaning?
>On Saturn its equivalent is the 2nd SH-2 and the SCU DSP.
SCU sucks balls and isn't fit for purpose. Bruteforcing it with raw software on the SH-2 is by definition not hardware acceleration.

>>8949372
>Does hardware T&L have some different meaning?
Having actual hardware T&L in the GPU. The GTE is again a co-processor. Sure it has functions that can help with software T&L, but at the end of the day you the programmer still have to tell it what to do and how to do it. The results of whatever you do on the GTE are then used to construct your 2D draw commands to send to the GPU. It's not really the same thing as what you see in a modern GPU.

At the end of the day it's still technically "fake 3D" as the GPU has no sense of a Z axis. It only knows of the X and Y axis and executes draw commands in that kind of space. The transformation is done in software using the GTE to help get it done quickly so it can then know what X and Y coordinates to give for each vertex of a polygon to then have the GPU draw it in the 2D screen space correctly.

This changes with 6th gen systems (and maybe N64?) as those systems are actually aware of the Z Axis. When you do a draw command on those systems GPUs you give it the X, Y, and Z coordinates of each vertex of your polygon, and it does the transformation on it's own when it executes the draw command. That's what actual hardware T&L is.

>SCU Sucks!
Not really, it's DSP is actually pretty fast and powerful. The problem is it has a limited amount of RAM and can stall the entire system if you're not careful with how you have it access HWRAM. If the Saturn just had one SH-2 it would make more sense. But since the Saturn has a 2nd SH-2 that 2nd SH-2 in most cases can do a lot of those calculations just as fast if not faster than the DSP without the drawbacks. That's not to say the DSP is useless or that it sucks, just that it really only has a benefit for very specific calculations that it can go off and do on it's own with out bothering the rest of the system.
> Bruteforcing it with raw software on the SH-2 is by definition not hardware acceleration.
And brute forcing it with a math co-processor still isn't hardware T&L.

>>8949482
Modern GPUs are just as fake as the PS1, modern GPUs are so fake they cannot even draw transparent polygons in the correct order. You have to sort the manually on cpu side.

>>8949489
Modern GPUs still have a sense of a Z axis and draw with vertices using X, Y, and Z coordinates. That makes them more "real 3D" than the PS1.

>>8949493
The only true 3D hardware that has ever existed is the Power-VR gpu that the Dreamcast had.

>>8949495
Not true, the NDS also has true 3D hardware.

>>8949482
>Having actual hardware T&L in the GPU. The GTE is again a co-processor.
Whether it's considered part of the CPU or GPU is an irrelevant distinction. By that metric you'd say the PS2 has no T&L, which is asinine.
>Sure it has functions that can help with software T&L, but at the end of the day you the programmer still have to tell it what to do and how to do it.
Most T&L units are programmable.
>At the end of the day it's still technically "fake 3D" as the GPU has no sense of a Z axis. It only knows of the X and Y axis and executes draw commands in that kind of space. The transformation is done in software using the GTE to help get it done quickly so it can then know what X and Y coordinates to give for each vertex of a polygon to then have the GPU draw it in the 2D screen space correctly.
Whether Z coordinates make it to the rasterizer also has nothing to do with whether you have hardware acceleration for T&L. But this goes back to my original point that what constitutes "real 3D" is a matter of opinion since there are many different features which add up.
>And brute forcing it with a math co-processor still isn't hardware T&L.
If it's a math processor specifically optimized for matrix operations then it isn't "brute forcing" at all. By that metric you'd have to say the N64 has no T&L. A reductive description of the RSP is that it's just a regular CPU with some functions that can help with T&L.

>>8949495
And Gamecube, Xbox, etc.
>>8949506
>Whether it's considered part of the CPU or GPU is an irrelevant distinction. By that metric you'd say the PS2 has no T&L, which is asinine.
That actually was a pretty divided topic back in the day. People argued that both Dreamcast and PS2 didn't have hardware T&L because they both relied heavily on their CPU's and coprocessors to calculate it. However I'm pretty sure both their GPUs do have a sense of a Z axis at least.
>If it's a math processor specifically optimized for matrix operations then it isn't "brute forcing" at all. By that metric you'd have to say the N64 has no T&L.
Which people have also argued i the past.
>But this goes back to my original point that what constitutes "real 3D" is a matter of opinion since there are many different features which add up.
Which is why this argument in general is just silly and pointless. The only people who care about it here are the people still console warring who just want to say "My console is more 3D than your console!"

>>8949506
>Whether it's considered part of the CPU or GPU is an irrelevant distinction. By that metric you'd say the PS2 has no T&L, which is asinine.
Hardware T&L means fixed function hardware to do all the steps of transform, clipping and lighting without any programmer explicit intervention to invoke it which is what the Geforce 256 and newer cards all the way up to the invention of vertex, fragment and pixel shaders several years later had. All 6th generation consoles don't have hardware T&L by that definition, because they all have coprocessors that have hardware accelerated instructions in the T&L pipeline but it still relied on the programmer to do that work. But they don't have vertex or fragment shaders either, you can't do any sort of programming on those coprocessors that became vertex, fragment and pixel shaders on GPUs. The PS2 was the last console to take this approach with its VPUs, when everyone else had shaders which would be unified.
>Most T&L units are programmable.
They were all fixed function hardware, the whole idea of having it in hardware was to speed it up. And hardware T&L went away anyways with modern GPU designs and sensibilities. The concept of having programmable pipelines and shader programs was to claw back that flexibility while still having the performance and that also gave rise to using GPUs outside of graphics work.
>Whether Z coordinates make it to the rasterizer also has nothing to do with whether you have hardware acceleration for T&L. But this goes back to my original point that what constitutes "real 3D" is a matter of opinion since there are many different features which add up.
This I can still agree with, the argument is dumb.
>If it's a math processor specifically optimized for matrix operations then it isn't "brute forcing" at all. By that metric you'd have to say the N64 has no T&L. A reductive description of the RSP is that it's just a regular CPU with some functions that can help with T&L.
See my first point.