Sega's Hiroshi Kataoka mentions Saturn VF3. never heard of any 3D-upgrade cartridge

http://www.1up.com/do/feature?pager.offset=1&cId=3149135

 

the first of april is a weird day for people in the "obscure"

 

That interview is from 31 March ;-).

 

I believe you Assembler - I have faith in the existance of the upgrade.

 

I had always hoped that the Saturn 3D upgrade was the
Lockheed Martin Real3D-100, with or without a PowerPC CPU

as discribed in Next Generation's Saturn 2 article, which at the time, speculated that
Saturn 2 would either be a new stand-alone console, or an upgrade to the existing Saturn






though I realize the many other possibilities for what powered the Saturn upgrade, including:

*Nvidia NV2
*3DO - Matsushita M2
*first generation PowerVR ~ PCX1 / PCX2
*first generation 3DFX Voodoo
*low-end Lockheed Martin Real3D - Intel 'Auburn' aka i740

 

I say that it could but of course not in the same way as the arcade. There are many 3D games with variable stage elements. With some serious work they could have pulled off something that looked like VF3 on the Saturn.

Take a look as this shot I took of Saturn Last Bronx just before.

Notice that it's in a train station with both a roof and wall ! True these are not made of polygons but high resolution 2D mode 7 effects and sprites and do look the part. This stage is quite similar to Sarah's stage on VF3. With the upgrade cartridge it would have been able to produce a pretty good looking replica I think.

Here's another stage with floor, roof and walls from Last Bronx.


Yakumo

 

I'm pleased that we've finally got official comments from AM2 regarding Saturn Virtua Fighter 3, but was this game really held over for the Dreamcast and would a 32-bit version really have been impossible? There's a part of me that wants to consider Hiroshi Kataoka's brief discussion on the subject as gospel, though on the other hand I want to hear from the one person who had the most to do with this project. Sadly, when Yu Suzuki was questioned as part of a recent interview he requested that any mention of VF3 in its console guise be left out of the published article - even to this day, he's upset by Sega's decision to reject his team's amazing Saturn effort in favour of Genki's rough port (to me this sounds like "some serious work" was indeed done, as Yakumo suggested would be required).

There definitely was a cartridge upgrade for the Saturn, and various reports suggest that it even made it to a working prototype stage. However, just because such technology existed doesn't mean that it was being used for VF3. In fact, the one thing we do know from Suzuki's own mouth is that both VF3 and Shenmue were running on stock hardware. Come to think of it, perhaps Kataoka should take a closer look at something like Last Bronx or even his team's early work on what became Shenmue and then reconsider his view that the undulating terrain of VF3 would have been out of the Saturn's reach!

P.S. Hasn't anybody noticed the presence of VF3's desert stage in Fighters Megamix? The characters might not look all that great, but you can imagine the general effect of how VF3 might have appeared. Of course, you'd need to add some extra polygons for the sand dunes - surely the Saturn had enough proverbial horsepower left to include these and the other stages in a basic yet playable form? For every answer we get, there's always new questions to be asked regarding Saturn VF3. Unlike many other cancelled games I could think of, this one seems destined to remain elusive forever. *sigh*

 

yes, the VF3 desert stage is in Fighters Megamix. I think that game looks awful though ! Last Bronx may not have light sourcing or shading but it does run at a much higher resolution making it look crisper plus the characters are better defined. Oh and it has the Radical Parking Lit as ASSEMbler pointed out ;lol;

Yakumo

 

While its characters and walls run in a lower mode, the backgrounds of Fighters Megamix actually use the Saturn's high resolution. I didn't think the desert stage looked too bad, though of course it would have benefitted from a few graphical improvements had it ever been shown in its VF3 guise. To be honest, I wouldn't have even minded if Saturn VF3 resembled Megamix with regard to the character model quality - after all, isn't gameplay more important? Just as long as the undulation and "organic" arenas could have been maintained without compromising the frame rate too badly...

 

honestly, I never liked Fighters Megamix.

Saturn was just not suited for Model 2 translations - it couldn't even handle Model 1 translations perfectly.

it's a damn shame Sega did not opt for one of the 3D chipsets they had oppertunity
to use ( SGI, Lockheed, M2, 3Dfx, PowerVR)

Sega wasted themselves on FMV SegaCD games, the 32X and Saturn. dispite all the incredible games that did get made for Saturn. it wasn't enough when 3D had gone mainstream. it's ironic that Sega was so far ahead in 3D technology in the arcades, but fumbled MISERABLY in the home market until the Dreamcast -- but by then it was already too late, no matter how well Dreamcast sold.

 

Well, the one thing I really loved about Saturn was its sheer number of coin-op conversions. Few were truly arcade perfect, but at least in most cases the gameplay remained intact - for me that's always what mattered. There's no doubting that the Saturn's hardware was fatally complicated, though at least some programmers stuck with the system and proved just what it was truly capable of in the right hands. Unfortunately, as with most of Sega's post-MegaDrive output these classic titles were a case of "too little, too late..."

P.S. Would the Saturn really have been that much better if it had used the same NVIDIA technology behind the Diamond Edge card? If anything, why didn't Sega just create a home version of the Model 1 board or even wait and produce something more like that board's successor?

 

no, it might have been worse, I don't know. the Diamond Edge 3D card used the NV1 chip. this was NEVER concidered to be used in a Sega console. the Diamond EDGE 3D card with NV1 chip actually had Sega Saturn controller ports on it, and several Saturn games were ported to it.

what Sega *was* working with Nvidia, was the-then next-generation NV2 chip, which would've been used in a new Sega console. IIRC that was cartridge based. but that was not Jupiter-- since Jupiter was pre-Saturn in 1993, whereas the NV2- based Sega console was after Saturn. but the NV2 chip didn't work well at all, so it was scrapped.

what would have worked (maybe) would've been a true triangle-based 3D upgrade for Saturn using Real3D-100 GPU (with geometry co-processor!)
or even the lesser 3Dfx Voodoo or Videologic PowerVR 3D accelerators (which had no geometry processor). one of these three architectures are the most likely candidates for the 3D upgrade that Assembler knows of.

one other possibility was the 3DO M2. but Sega pretty much rejected M2 after evaluating it. Sega, or 3DO's Trip Hawkins walked away from a deal that was put together in late 1995 or early 1996. even though Sega rejected the M2, it would've been VASTLY superior to the Nvidia NV2, and even a little better than 3Dfx or PowerVR first generation chips.

all of this went unreleased, and then Sega began work on two totally new consoles in parallal, the BlackBelt with next-gen 3Dfx and Katana with next-gen PowerVR.

 

Would it have even been possible to create an upgrade for the Saturn that used triangular polygons as opposed to the base system's quadratic geometry processing method? As for the idea of a Saturn follow-up with similar power to the M2, that sounds all too much like what we got (not in terms of specifications but definitely performance-wise) with the Dreamcast. I'm quite happy with how the Saturn turned out, though I can't deny that a SGI or 3DFX-powered console sounds like a better solution than what Sega ultimately rushed onto the market.

 

thank you so much for posting those scans, that was an awesome read! =)

do you have any more sega arcade-related stuff, especially concerning the model series?

 

not really offhand that I can think of
( I havent had my coffee yet, not thinking clearly)

I mean, not anything more than what you could just find yourself on the web. If I do find anything particulary interesting, be it from the internet or magazines, I'll be sure to put it here.


okay here is more about Lockheed Martin entering into the PC 3D graphcs industry with Real3D, with what was said to be a $180 graphics card. the article talks mainly about Real3D-100 (it's more in-depth on Real3D-100 itself, than in the later Saturn 2 article) a GPU that is more powerful than MODEL-2 arcade board, though not as powerful as the very high-end Real3D/Pro-1000 GPUs in MODEL-3.

Real3D-100 was supposedly going to enter the PC graphics industry at that affordable $180 ~ $200 price point (if it had, Nvidia, Rendition, 3Dfx, PowerVR and all the others would've been dead), not only was Real3D-100 supposedly going to be for consumer PC cards, but also, as the later article (that I posted before) reported, would be the graphics subsystem for Saturn 2.



the problem was, the big Lockheed Martin Real3D announcement in 1995, which this article is apart of, announced a $180 price, but showed off a chipset that would cost several thousand dollars with its max amount of memory, and still many hundreds of dollars with less memory, still far more than $200.

The Lockheed Martin Real3D consumer-grade PC chip would not come out in 1995 or 1996, and it would *not* be Real3D-100, but instead would be a low-end, 3D accelerator, without a geometry engine (thus it was not a GPU) with performance closer to that of the original 3DFX Voodoo graphics. It would come out much later, in 1997. It was codenamed 'Auburn', better known as i740, a joint-venture with Intel (used in Intel's crappy intergrated graphics) and the same chip that powered Lockheed's crappy consumer-gamer-level PC cards, the Starfighter series. I honestly don't think Auburn / i740 was even as good as MODEL-2, let alone Real3D-100, and not to even mention the awesome Real3D/Pro-1000 powered MODEL-3. the Auburn / i740 was closer to Voodoo1. though I'm not sure how it compared to 3DO M2. I supposed it could've made a decent 3D chip to upgrade Saturn very cheaply. but by 1997 it was already outdated with PowerVR2 already selected to go into a totally new Sega console, the Katana / Dreamcast.

but I always wanted to believe that Real3D-100, along with a PowerPC CPU and some RAM, could've been the basis for a Playstation-obliterating, N64-crushing, 3DO M2-beating new Sega console, or Saturn-upgrade, called Saturn 2.

the Real3D-100 had three main processors:
geometry processor
graphics processor
texture processor

in say, a 3DFX Voodoo Graphics, I assume the PixelFX chip is equivalent to the R3D-100 graphics processor, and the TexelFX chip is the equivalent to the R3D-100 texture processor. 3DFX Voodoo had NO equivalent to the geometry processor, that was the key difference between true GPUs such as R3D-100, and all the other PC 3D accelerators of the 1990s. consumer-grade gaming graphics chips for PCs did not gain a the geometry processor until Nvidia brought out the NV10 in fall 1999 -- the original GeForce.

in Playstation1, which is not nearly as good as R3D-100, the equivalent of the R3D-100 geometry processor would be the GTE (geometry transform engine) embedded into PS1's CPU. the equivalent of the R3D-100's graphics and texture processors in PS1 would be PS1s single graphics chip, which was severely lacking in features, although pretty damn fast for its time and console price. In the Nintendo64, the equivalent of R3D-100's geometry processor would be the RSP (reality signal processor) portion of N64's RCP (reality co-processor). the equivalent of the R3D-100's graphics and texture processors in N64 would be the RDP (reality display processor) portion of N64's RCP.

3DO M2 - okay this is more interesting. the M2's equivalent of R3D-100's graphics and texture processors would be contained within M2's most custom chip, the BDA (BullDog ASIC). one of the M2's twin PowerPC 602s would be doing the geometry processing, and thus be the equivalent of R3D-100's geometry processor.

All in all, the Real3D-100 is more powerful than Playstation, Nintendo64,
3Dfx Voodoo, PowerVR PCX1, PowerVR PCX2 and 3DO M2. a 3Dfx Voodoo2 paired with a fast CPU might be able to rival the Real3D-100. of course, the Dreamcast with its SH4 + PowerVR2DC smokes the R3D-100, but that's a 3 year gap in technology (1995 to 1998).

ahh never mind me, babbling on in the moddle of the night about an 11-year old GPU.

 

I don't see why not, as long as the bus to Saturn's motherboard could support it. the VDP1 and VDP2 chips could just be forgotten about, or relagated to 2D sprite and parallax scrolling which they were designed for. if you real the Next-Generation article on Lockheed Martin based Saturn 2, you see that there was the idea of using the existing Saturn for it's CD-ROM drive, I/O and power supply, and the Lockheed Real3D (and PowerPC CPU) do the rest.

As for the idea of a Saturn follow-up with similar power to the M2, that sounds all too much like what we got (not in terms of specifications but definitely performance-wise) with the Dreamcast.

well, features-wise, yes, kind of, but not performance-wise. Dreamcast was easily 5 times more powerful than M2 and very likely more. M2 could handle between 300,000 and 500,000 polygons with all effects on (N64 was ~160,000) Dreamcast could handle AT LEAST 3,000,000.

An SGI console could've been done in the timeframe that Saturn came out, at least in the U.S. a 3Dfx console could not, because 3Dfx was founded in 1994, did not announce Voodoo Graphics until the end of 1995, and did not release Voodoo until mid to late 1996. Lockheed Martin was far ahead of 3Dfx not only in performance, the ability to deliver 3D in time (but unwilling to do it at consumer prices until a few years later in much scaled-down form (i740) . Even though the later Model 3 tech was late, Model 3 still came out around the time of the original Voodoo.

PowerVR was even later than 3Dfx with their first gen chips which did not hit the market until early 1997, right around the time that news of the 2nd gen PowerVR hit the web and magazines.

Sega's best choices would've been SGI or Lockheed Martin. next best choice would've been 3DO M2. I'm talking about a console for the 1995-1996 timeframe. instead of Saturn or to quickly replace or upgrade it.

 

I'm sure Assembler knows a lot more than I know, or think I know.

I'm sure he knows of things that I am not aware of and have not even thought of.

 

See: ST-V ports
Lots of other ports (mainly 2D stuff). Recap wrote a good post on these... gotta find it.