• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Sega Saturn - Metal Gear Solid (fan made) Prototype

I might be crazy, but I remember some source saying the Saturn version could have handled larger rooms are more than 2 enemies per room.I read it a long long time ago so I cant remember where.

If it was capable of such, that would have likely been down to have two CPUs in the system, while consoles like the PS1 had just the single CPU. I'm assuming thanks to VDP2 offloading a lot of the background rendering, the 2nd CPU would be freed up from issuing command lists for polygon generation on background objects, and could use that budget for VDP1 instead.

Much needed context from the creator of this:



More context from the video description...



He does go further into how difficult this was for him and how difficult it would have been for the official team to port to Saturn in a patreon post, but the post is paywall locked.

Well, FWIW even the PS1 had to "fake 3D" for games like Resident Evil where the backgrounds are prerendered 3D bitmap images. Between games like that, and MGS which have mainly static overhead-angled cameras, the Saturn could mostly "keep up" with PS1 in those types of games.

Though for when playing MGS and switching to cover-camera or first-person with the binoculars, the Saturn's 3D limitations would have been immediately more evident. It'd of likely needed to use VDP2 for background objects simulating 3D the way Sega's Superscaler 2D sprite games did. Which really, was the philosophy of Saturn's 3D in a nutshell, just turbo-charged to a maximum, logical conclusion of that rendering methodology.

That's putting it mildly. I remember reading how devs had to dump the devkit and code their game in Assembly(!) in order to get maximum performance out of the Saturn.

TBF you had to do this to an extent with PS1 and N64 as well; no games on systems that gen maximized the hardware without coding big chunks in assembly language.

The difference was that PS1's hardware was less convoluted (less chips to program assembly for in unison), and Sony's APIs were much better than Sega's even for low-level, making assembly on PS1 easier. The Saturn's DSP alone was a beast in its own right and IIRC had to be completely programmed for in assembly; C language wasn't an option.

I'm recalling GameHut's video on it, in his Sonic R coverage, FWIW. He put out some really good documentation/proof of programming for the DSP in Saturn.

yes I know. In short, depending on the game design, this could give advantages to the Saturn but there are limitations, in 3D racing games this technique cannot be used, nor in the main 3D platforms that we have come to know such as Tomb Raider 2, Rayman 2, Gex Enter The Gecko (advanced game) even Croc 2.

With 3D racers, you could somewhat still use VDP2. It'd just be for stuff like the skybox, distant background objects, and certain trackside objects like trees, bushes, the crowd etc. Maybe even parts of the track you'd drive on, but that would have probably required clever programming of VDP2 and VDP1 (polygons) for a smooth effect.

At least, I'm thinking that would have been possible. If the track has elevation changes it probably becomes less feasible.
 

SomeGit

Member
Dreamcast = 4 Saturns = 4 x (2 32x) = 4 x 2 x (2 MD) = 4 x 2 x 2 x (2 MS) = 32 MS

Hmmm...

The Dreamcast is the Master System 32x!!!

Master System’s melded together is a much better metric than GameCubes duck taped together. It should be the new industry standard.
 
Shenmue is one of those “ how the hell!?” Deals. I looks better than most games at the time on any system.

The Saturn was a big victim of “ why should we put in any effort? “ when it came to third parties.

Its architecture was so different from the way 3D graphics were heading and the failure in the west meant that there was little incentive to put in any time into the machine.
So some of the few multiconsle third party games were said to be ported by just adding an extra vertex overlapping a vertex to a 3 sided polygon to make it a quad .. that meant The Saturn had to push more vertices and draw more pixels than what was actually needed.

This isn't necessarily true. Nvidia's first GPU was a joint effort between them and Sega, based on the Saturn architecture. Honestly, the future of 3D could have gone either way, between quads and triangles. Both had their benefits and drawbacks.

Saturn's 3D was a natural evolution of the Superscaler 2D stuff of Sega's arcade line of games. It's literally taking quadrilateral sprites and applying perspective distortions to position them in a 3D space. The problem is it only worked with quads, and didn't do much to be more efficient for use-cases that would have naturally favored triangles. The Saturn didn't really have a way for discarding wasted vertices when a quad was 'folded' to form a triangle, for example, that didn't also impact processing performance.

That was really on Sega, though. They should have foreseen it and ensured the APIs and support tools were robust from Day 0. When even their own arcade teams had issues with Saturn development early on, you know a major problem was at hand. Those teams were used to both quads and dual processor setups, so that was saying something.
 

Moses85

Member
Yeah. The Saturn could do 3D. It could have done Metal Gear Solid if it were ported to it, albeit a bit worse than Playstation. This isn't all that impressive, lol
How about doing it better?

Your Turn GIF by Chattanooga Mocs
 
With 3D racers, you could somewhat still use VDP2. It'd just be for stuff like the skybox, distant background objects, and certain trackside objects like trees, bushes, the crowd etc. Maybe even parts of the track you'd drive on, but that would have probably required clever programming of VDP2 and VDP1 (polygons) for a smooth effect.

At least, I'm thinking that would have been possible. If the track has elevation changes it probably becomes less feasible.
just things like minimap. Tracks, crowds and trees are all made on VDP1 and VDP1 is weaker than the PS1's GPU so if you see any racing game where the Saturn is close to the PS1 know that a lot of effort was made by the devs
 
just things like minimap. Tracks, crowds and trees are all made on VDP1 and VDP1 is weaker than the PS1's GPU so if you see any racing game where the Saturn is close to the PS1 know that a lot of effort was made by the devs

Are you sure about that? IIRC the crowds were just bitmaps, not polygonal models. Same with the trees. VDP2 had five layers, and the UI would've used one layer. The distant background would've probably been another layer, then the skybox. That is still two layers left for bitmaps which they could've different types and sizes of 2D assets on with per-layer scaling and shifting.

Since the POV is 'fixed' (i.e it's not like an open-world racer where the player can drive off-track), there's a certainty in knowing what the player would see at any time on the track, and you could program things from there.
 

cireza

Member
Are you sure about that? IIRC the crowds were just bitmaps, not polygonal models. Same with the trees. VDP2 had five layers, and the UI would've used one layer. The distant background would've probably been another layer, then the skybox. That is still two layers left for bitmaps which they could've different types and sizes of 2D assets on with per-layer scaling and shifting.

Since the POV is 'fixed' (i.e it's not like an open-world racer where the player can drive off-track), there's a certainty in knowing what the player would see at any time on the track, and you could program things from there.
You cannot put content rendered with a VDP2 layer between polygons rendered by VDP1 (both in front and behind) which would be a major blocking point for what you suggest.

Also you would have to calculate where is the limit between front and behind polygons. In the end, the only choice that would work would be to use the technique in Burning Rangers which limits the framerate.

VDP1 is very fast at displaying and scrolling 2D sprites anyway. So why bother ?
 
Last edited:
I mean... It's probably harder to appreciate the fervor that it can drum up now if you hadn't participated in that generation in any real way. Going from 16-bit, to essentially having this bad ass and at the time what felt like "Movie Quality" presentation...there just wasn't anything like it really. It was like playing a bad ass spy movie and cheesy as all hell, but with some actual deep esoteric messages. Not always great, and if you aren't in the mood for it's specific kind of "je ne sais quoi" there's so many great offerings these days that it's easy to not fall into the excitement the series has a history with.


You're not wrong by any means, but I think there's a massive gulf in difference in presentation from RE1/Castlevania on Saturn and Metal Gear Solid on PS1. They are kind of on the same track but at massively different levels. I'd be interested in what would have had to been changed/cut to even get the game running acceptably on the hardware.

Castlevania is an outlier in that discussion. It needed a proper port job to take advantage of the Saturn's capabilities.
 

AREYOUOKAY?

Member
They made a six button controller for Street Fighter II.
Even so it would play awkwardly with the button layout. It’s like playing Mega Man X 3 (Which also was planned for the 3DO) on the Saturn using that controller over the PlayStation or SNES versions and their layout. It just feels wrong but who knows maybe someone enjoys it that way? I’d have to try playing MGS with a Saturn controller using an adapter myself, but I can’t imagine it’s worse than playing it on the Vita with those back L and R touch buttons which you absolutely need to buy buttons for.
 
Last edited:
Top Bottom