Virtua Fighter 3(Shenmue) on Sega Saturn

x

 

This is what Chris Coffin and her team were working on for the Saturn. They dubbed it Project Condor. Meanwhile, Chris Senn and Ofer Alon were working on their own version on the PC. Turns out, most pictures and videos we ever got of X-Treme came from the PC Version. So we know relatively little about Project Condor. What's interesting is that this engine used a 3D modeled Sonic, instead of the classic sprites. Chris Coffin said that it took the Saturn the same time to create the model that it took it to decompress the sprites, so they went with the 3D model instead.
I'd love to see more of this, but Chris Coffin was disappointed when she saw her work being sold by someone here on the forum for whatever money it was sold (2000+ $).
Perhaps someone should try to get her to release more screens or videos of this?

 

x

 

I sincerely doubt that. Ofer Alon had apparently little to do with this version anyway. And anyway, the guy doesn't want to have anything to do with Sonic X-treme since 2005 or so when Sonic fans pestered him about it. He probably has moved on and doesn't want to remember a project that he probably spent a ton of time on and in the end got nothing out of it.
Those screenshots were released by HXC with the permission of Chris Coffin (she's the one who supplied him with them in the first place). They were indeed taken using SSF.
More here : http://forums.sonicretro.org/index.php?showtopic=18538&st=150&p=383323&#entry383323

 

All these years later, and I've only just noticed that last screenshot is actually from a stage of the "Sonic Pool" bonus stages rather than the main Project Condor game engine (duh)! From what I remember about this period in Sonic Xtreme's development, Peter Morawiec originally pitched his concept for the pool areas as a special stage, but when STI abandoned the project he then suggested it for inclusion in the Saturn port of Sonic 3D. However, since Travellers' Tales were working under the supervision of Sonic Team, anything to do with Xtreme had little chance of making the final cut - especially when it's claimed that Yuji Naka himself was behind the polygonal update of the Sonic 2 bonus levels.

Considering the politics between Naka and STI at this time, it was little surprise that his idea was chosen over Peter's work, though I still hope Sonic Pool finds its way out one day as it appears to be far more complete than the game he originally created it for. As for those Condor build pictures, I'd always suspected they were running in SSF, and I've even heard in the past that copies of this may be in the hands of some very lucky people out there. Also, it's great to see the curved eye surface trick works through emulation, though I'd totally forgotten this exploit was documented so please forgive my earlier mistake on the subject.

Regarding the performance figures of Shenmue mentioned in another of my previous messages, these exact numbers came directly from a member of staff within Sega Europe who knew far more about Virtua AD (as it was known back then) in contrast with the very limited information found on AM2's official Japanese page, which can still be found via the Wayback Machine, I should point out. 500-750k polygons a second is supposedly very much possible and was quoted as being a genuine representation of what Keiji Okayasu's team could produce. In terms of speed, the number passed on to me was 12-15 frames per second, though I continue to have a strong feeling that video footage was manipulated in post production somehow.

Despite there being very credible internal sources to suggest that Saturn Virtua Fighter 3 was either completed - twice, no less - or at least in a preliminary yet playable* state, AM2 has still to show anything of this one as we all know. If this highly anticipated conversion was anything like the Shenmue prototype (and let's not forget that it was in development over a year after Yu Suzuki's epic saga had already made the jump over to what later became the Dreamcast), they'd surely have given some kind of similar demonstration by now, even if it was just a handful of static images. So, why the maintained secrecy?

This may be a long shot on my part, but I'm willing to bet the shroud of secrecy is a result of direct orders from Yu Suzuki himself. There's simply no way colleagues such as Hiroshi Kataoka would be allowed to openly discuss the 32-bit roots of Shenmue and deny that VF3 was ever in the works if both projects existed on an equal level. The fact remains that while Shenmue did eventually ship to its maker's satisfaction, Suzuki allegedly felt rather undermined when it became obvious he'd spent considerable time and resources on giving Saturn owners one last installment of the VF series, leaving responsibilities for the Dreamcast version in the hands of Genki. For those not aware, this outsourcing team was founded by a large number of former AM2 members, and I'm sure that Suzuki once trusted their skills when it was clear his team couldn't spare the necessary personnel for a high profile game (see their earlier Hang On GP for proof of this).

Even before the first reviews of VF3tb focused on minor graphical details being inferior when compared with the Model 3 source material, Sega was strongly emphasising the Dreamcast's ability to flawlessly reproduce titles from this high-end arcade board. In reality, it soon became painfully obvious that despite boasting great improvements over the Saturn and its complicated architecture, those looking forward to the Dreamcast's arrival were soon concerned by the pre-release media of VF3tb and the infamous Tokyo Game Show '98 video of Sega Rally 2, which was still in a 40% complete, non-playable form and looking more like a mid-range PC game just weeks before its intended debut alongside its host console.

Fortunately, although Sega Rally 2 was eventually the victim of significant delays, VF3tb was never actually that bad. On the other hand, with the promise of pixel-perfection still fresh in the memories of journalists and consumers alike, minor graphical shortcomings simply couldn't be excused - no matter how brilliantly the gameplay had been carried over from arcade to console. Sensing they might perhaps have been capable of better if not pre-occupied with a conversion that was supposedly once again turned down for release in September '98 after the implementation of many improvements, AM2 and Suzuki in particular felt the small failures of Genki's Dreamcast work the hardest.

To this day, the former AM2 leader has yet to openly confirm or deny even the existence of Saturn VF3 with the exception of his announcement at the Power Up press conference in November '96 to reveal that he would be taking full responsibility for the conversion following months of extensive technical research into whether or not such an ambitious game was within the Saturn's capabilities. Assuming that the character movement demonstrations I mentioned earlier date from anywhere up to this point in time, why would they be associated with any of AM2's work going right through to September '98? This just doesn't make any sense to me! I still believe there's more to this particular case, but like so many aspects of Sega's history, it seems getting a straight answer is virtually impossible... and that pun certainly wasn't intended.

How can it be that some anonymous contact or a few people here are prepared to swear that VF3 existed in at least tech demo form, yet those still with AM2 who were there back in the day maintain they saw nothing and that it never existed in the first place? Come to think of it, just how long can a non-disclosure agreement hold out for, and could the "never saw anything" attitude be a sign of continued loyalty to a certain former producer within the department who's since taken the fall for a chain of epic fails, including his pet project, Shenmue?

*According to several contemporary website and magazine articles, Saturn VF3 was confirmed as being in a working state around the same time Genki's treatment of VF3tb was first previewed for the Dreamcast. If my timeline is correct, this would have been May 1998, and it's reported that on July 8th Sega of Japan officially denied the first revision of AM2's work a release, even though it was supposedly finished. Of course, there's a world of difference between having a completed product ready to launch and basic tech demos. For every person who's claimed to have ever seen a final build, there's easily a dozen more that can verify the existence of what have been independently described as incredibly simplistic character animation tests. Any accounts of the former are all third-hand information, while one thing we can be certain of is that the latter has too many descriptions for this to be co-incidental.

 

By the way the PS1 Version of Dead or Alive looked quite poor(even with... or because of the go around shading...poor excuse for less texture work). They weren't even able to reproduce the arcade backgrounds!
by the way die hard arcade is also a technically great achievement!

 

x

 

Can you please draw out how you have arrived at these numbers? Is it all about the fill rate, and if so what is the Saturn VDP1's fill rate? I remember seeing a discussion on it once claiming that the PS1's fill rate was something over 30 mega pixels per second and the Saturn is around 28. But the highest polygon count either system reached in game was around 200,000 polygons per second. As far as I know Iron and Blood is the only PS1 game to reach 200,000 and the game looks just like Fighter's Megamix or Shining Force III's battle scenes.

Anyway, I'd love to see further discussion on this, especially any math involved in figuring out these system's real world capabilities.

 

x

 

Where was the fill rate posted in this thread? Using 32x32 polys with 28Mpix/s at 15 frames per second is over 400,000. I'm not sure that is how it works, but it would be awesome to know for sure.

 

I mentioned it was about 24Mpx according to some unofficial tech sites, I've heard it been guessed as low as 7Mpx, but that was a few years ago when people where reading the wrong clock for the VDP1, accepting that VDP1 can write 1 pixel per clock 28mpx seems about bit right assuming its clocked at 28Mhz like the SH-2's are, I was being a bit conservative with 24Mpx.

Lets not forget how much fill-rate VDP2 saves though, two infinite planes and 3 rotating backgrounds, would massively eat into any renderer of that generation, but VDP2 wasn't always useful depending on the kind of game you where making. I thinks thats probably where the Saturn could really equalize agains the PSX in performance provided you could intelligently use it.

PSX I thought was anywhere from 33-66Mpx?

Anyway, lets get straight to the point. So you've basically got someone who probably dreams Super Hitachi RISC code in his sleep, who somehow manages to get both SH-2's roughly working in tandem (no mean feat), pumping an insane amount of geometry (note: the SH-2 is faster than the R3000A at matrix math, but then the GTE is faster than both). You've also done some clever bandwidth saving optimisations, culling etc, to save as much bandwidth as possible for drawing.

You've wasted a good chuck of VRAM for your display-list, meaning spartan texturing.

Heavy and clever use of VDP2.

And you've got VF3 running at a fluctuating frame-rate of 15-25fps.

Thats not even taking into account basic AI, gameplay housekeeping.

But the biggest question that begs to be asked is, whats the point? The Virtua fighter series from 2 onward pushed a rock solid 60fps (57.5 VF2 arcade), why bother having a game that relies on precision timing running at a lacklustre framerate. The game would be laughed at in competitive circles, and would be a poor representation of the game, you would'nt be able to count the frames properly since they would be all messed up, your timing would be messed up, VF prestige comes from the love it gets in the hardcore fighting scene. The fluidity that makes VF3 VF3 wouldn't be there.

AM2 doing a faithful home version of VF3 to me would mean 60fps and nothing less, otherwise its just a glorified tech demo. And I doubt Yu Suzuki would compromise on that.

Regarding the SCU DSP a 14Mhz co-processor really is'nt going to be some magic turbo-boost to the system either, people look to it longingly as some secret enigma that once unlocked could make the Saturn dominate the competition. I imagine one of the SH-2's is faster at matrix multiples, and its not likes you could get the slave SH-2 and SCU DSP to sync up easily, its hard enough balancing two SH-2's surely, let alone a third processor. The SCU DSP seems like its left-over from an older slower design, where it was used to generate small amounts of geometry before the Saturn was beefed up. Thats just speculation of course.

Phew, I'm off to bed. Its getting late here in blighty.

 

Sega-16 might be anathema here, but it does have some very knowledgeable members. At the link is instructions for converting the bump mapping code to a usable format, and a picture of the results.

 

AM2 only started the conversion process after Yu Suzuki himself was satisfied the results would not only meet his own expectations technically, but impress the average consumer and hardcore VF players alike. As he said during the Power Up conference, several months of extensive technical research were conducted, with the real time Aoi character movement demonstration previously mentioned a probable key element of this preliminary work (which is all the more likely when you consider how many accounts there have been of this material - clearly those responsible for it were proud of their achievements).

I'm not entirely sure where you got those frame rate numbers from, since it's been documented that Saturn VF3 ran at a consistent 30fps in both of its complete revisions. Assuming that my understanding of the descriptions I've read in the past is correct, polygonal backgrounds only existed in the later finished build and were extremely basic in their construction. Just a quick look at Fighting Vipers or Fighters Megamix proves the Saturn was more than capable of generating interlaced 3D characters and surrounding walls at 60fps, and when you think about it those didn't exactly use large numbers of polygons.

Although this is pure speculation on my part, I've no reason to doubt that most of the levels in VF3 could have been reproduced by using a mere handful of polygons, and there are enough similar accounts to assume that the character models weren't too much of a leap over those seen in Vipers/Megamix. Then again, if Digital Dance Mix was a glorified tech demo from early in the research process for Saturn VF3, how come this utilised the system's highest quality display mode when the "main event" ran in the same resolution as the Shenmue prototype?

My best guess is that the detailed model of Namie Amuro was specifically a test bed for the rumoured switching method used in VF3, where only scenes featuring one visible character (such as during their winning poses) used textures comparable with those seen in VF2. For the rest of the time, VF3 wouldn't have looked too much better than Megamix, though as I said before it's claimed the frame rate was half that of its console predecessor and boasted the most simple of geometric shapes to represent the Model 3 original's incredible stages.

I can think of so many games on the Saturn with decent-looking scenery, only to discover upon closer inspection that these locations were built up from very few polygons. To name just one example, I found the overall texture quality made up for any poor geometry in Panzer Dragoon Saga, and VF3 was believed to have rendered background elements at a quarter of the resolution of its characters, similar to how Burning Rangers handled transparent effects. While this approach would certainly have made it harder for AM2's textures to look any good, it definitely wasn't theoretically impossible.

Yes, the odds were stacked against VF3 being an arcade perfect conversion, though on the other hand few people cared about the general reduction in graphics that was necessary to realise Tekken 3 on the PlayStation. Despite its many compromises, this particular game is still held in high regard because of how well it played, and I've absolutely no reason to doubt that AM2 was following Namco's approach with Saturn VF3. I'm sure most would have appreciated this one last showcase of the relatively untapped power Sega's 32-bit machine possessed in the right hands, plus the door would then have been open for a truly flawless Dreamcast sequel.

We all know what happened instead, and I have a feeling Yu Suzuki took this any kind of failure rather personally, as seen with the slightly later demise of any plans to conclude Ryo Hazuki's journey in Shenmue. From the cancellation of Saturn VF3 to Shenmue III's disappearance and the finally the delay in producing a fourth Virtua Fighter, AM2 lost all sense of consistency very quickly. Naturally, their once untouchable leader seemed a liability to management within Sega, eventually leaving having been undermined after years of being unable to deliver hit products that were once second nature.

Plenty will consider the ill-fated Shenmue III his final straw, but I prefer to think it was the massive loss of momentum in the VF series that ultimately spelled the end of Yu Suzuki's reign at the top of AM2. Then again, you could argue that the entire fighting genre was in a decline by the point VF4 eventually struggled out, even though in Japan it seemed very much a case of business as usual in the arcade scene, regardless of declarations the industry as a whole was on its proverbial knees.

In my alternate universe, Soul Calibur still stole the limelight as the flagship fighter around the Dreamcast's launch, while Genki's flawed VF3tb was left on the drawing board. Meanwhile, VF3 would have brought the Saturn's life to a triumphant close as an exclusive for the machine, fronting one last wave including Sonic Adventure and the first two chapters of Shenmue. With the Dreamcast finally unleashed in mid 1999 as opposed to its rushed late '98 launch accompanied by a pitiful number of half-baked AAA titles, VF4, Shenmue III and a bug free Sonic Adventure 2 would then have really introduced the world to the next level.

Having learned major lessons from the Saturn era, Sega would have been in an ideal position to defend its newly-restored fortunes against the PlayStation 2, and I honestly believe the chain reaction started back with VF3 might have just proved enough to maintain desperately-needed financial safety well beyond the 2001 cut-off point that eventually led to its restructuring into a third party software-only company. I must apologise for slipping into the theoretical, but I simply cannot accept how easy Saturn VF3 is for so many of you to dismiss. After witnessing that Shenmue footage, I'd love to see what else AM2 was capable of during its peak.

 

Some of Virtua Fighter 3's backgrounds would have been impossible on the Saturn while maintaining higher than FMM level character models. Here are the ones that I think would be impossible on the PS1 or the Saturn. Note that these all have fully 3D tiers to them that the fights frequently end up on.





Pai's stage might be possible with VDP2 floors:

 

I once prepared a few mock-up shots for this very purpose, but since you've already uploaded a few Dreamcast screens let's take a look at each one and see just what a Saturn version might have been like...


In this instance, the floor, roof and distant background would probably have been bitmaps handled by VDP2. Even the railings and moving trains would probably have been easy to process the same way, with plenty of polygons being saved by having the upper stairs a background layer as well. So, what does that leave to be drawn as polygons (apart from the characters, of course)? I'm sure the Saturn could have managed a few low polygon stairs.


Apart from the handful of stairs at the back and those two low walls, everything else - including the rock walls and building to the right in this picture - could easily have been more hard work passed over to the VDP2. Yes, this approach would have meant that a lot of the subtle ground undulation would be lost, but at least the essence of this stage remained.


Rather than a scrolling element for the rooftop itself, I supect the floor was a handful of polygons, along with perhaps the two back fences again. I suspect everything behind that would have been more for the VDP2 to deal with - let's not forget that similar results were achieved in Last Bronx, where the 3D effect on plenty of background details was actually quite convincing, despite most of these being scaled sprites!


While there's no doubt the Great Wall stage would have proved one of the more challenging to accurately reproduce on Saturn, I reckon AM2 had some idea of how to bring over the basic feel using a small number of polygons. The walls at either side of the two fighters would definitely have remained, and I suspect at least some of the undulation was possible. As with the previous examples, my best guess is that the VDP2 was put to a lot of use here, too.


Those trees and the floor could easily have been done as separate bitmap layers, but what of the two rock walls (not pictured) and fence in front of these? As with before, this is probably where the majority of polygon would have been used, though I'm not sure how the flowing water was dealt with. Maybe this was removed entirely, or even changed to a frozen state?

[/QUOTE]
For this area, I can't see why a lower polygon recreation of those stairs couldn't have been possible, especially if the rest of the stage was a scrolling background. Even the fences at the top of those stairs would be relatively easy to implement, and for the record it's been suggested the character models during matches weren't much better than those of Fighters Megamix.

As for levels not featured in these images, you only have to look at D-Xhird to see that even third party programmers knew how to push the Saturn enough to generate a circular-floored stage, such as the one seen in Taka Arashi's Sumo stage. Sure, the background details would more than likely again be reduced to a scrolling bitmap, but with a little imagination VF3 would easily have been within the reach of Sega's 32-bit machine.

On a purely technical level, my greatest concern is how AM2 intended to deal with the transparent floor panels of Jacky's construction zone, since the familiar dithered mesh effect would have looked dreadful. Then again, the aforementioned Takara game had transparencies, which I believe were the result of VDP2 blending.

Also, we should consider that any background polygonal data in Saturn VF3 was supposedly running in a lower resolution, presumably to further save on the overall workload. Burning Rangers utilised a similar technique, and I'm sure AM2 was keeping a close eye on methods being employed by other Sega divisions. After all, Shenmue "borrowed" its camera system from Panzer Dragoon Saga!

 

@seath Absolutely right! With any Hardware upgrade non of this STAGES WOULD HAVE BEEN POSSIBLE!
Saturn could only create impressive 3D fighters with 60 frames thanks of zooming and rotating playfields! It would have been easier to port VF4 or 5 at least the flat stages!

By the way D-xhird was only a 30 frames game....that means more polygons and power for the graphics...

Polygon employment only for fences in Last Bronx already pushed the engine to its limits(glitches)

transparency effects on jackys stage wouldn't have been a problem at all....because saturn hasn't got any problem to create transparent playfields

i know AM2 is an incredible gifted team.....but there are hardware limitations that even they couldn't surpass. There are no saturn games giving any evidence that such a big leap would have been possible...

 

Nobody is dismissing the existence of VF3, a few others and myself are trying to ascertain a few concrete facts on the technical side of the game. The whole problem as yourself must be keenly aware is pinning down anything, unless we all miraculously have the golden master of the original playing on our Saturn’s for a detailed technical breakdown, all we have to go on is your own suggestions; a few nuggets, a bit of industry gossip here, some contacts there.

You’ve provided a rough outliner from what details you’ve managed to piece together that provide a good estimation of what VF3 could look like, we also have the comparative evidence of Shenmue that shows some of the ground breaking technical work AM2 was coding on the Saturn at this period, there is no doubt in my mind that VF3 would have pushed just as many technical boundaries in its own way with the Saturn hardware.

The problem with some of your suggestions is that you seem to create a narrative history about AM2, inflected with your own emotive discourse. You’ve probably never met Yu Suzuki (I’m not totally ruling that out though!) yet you frequently create a discourse about how all these grand projects at Sega, such as VF3 and Shenmue where all very personal failures for Yu Suzuki, yet either due to the impossible transit of Japanese emotional values, these feelings are never made felt in interviews, if anything Yu Suzuki seems to have a great deal of gratitude and love for Sega, pre-Sammy, and his adoring fans. You suggest this information is all of course behind doors, which exonerates you from provided any factual evidence yet how could you be privy to the emotional feeling of one of Sega’s contemporary or previously withstanding directors. On the other hand perhaps you should look at Sega’s perspective; Yu Suzuki had vision, but his ideas where increasingly becoming more and more costly, a systematic problem also at the heart of Sega at the time, great vision without any grounding in the facts of reality such as cash flow and revenue.

I will be perfectly clear about this, I’m not denying and suggesting your information on the subject is false, but until concrete evidence and sources are provided this information is just here say and Chinese whispers on the internet. I don’t think you realise how many times I’ve stumbled on preposterous claims about Sega and the whole Saturn period from what seem to be very young and naïve posters on many video game boards, who’ve read some of your technical posts and the narrative you’ve created about AM2 and Sega and go on to extrapolate and turn to mush what was probably a well meaning gift of information you’ve been giving out. Most importantly a few years back I too dabbled in this game of internet Chinese whispers, but I went out and started asking question on boards like Sega-16 to find out concrete technical facts about Sega’s hardware, and also to understand via the interviews of somebody like Melf over at Sega-16 the culture of Sega as two distinct transatlantic division wrestling for Sega’s direction.

I think you need to start being increasingly careful about what information you present and give out, you certainly give out nuggets that allow us Sega fans to keep the faith, yet as someone who has studied academic history all this propagates is a never ending stream of unquantifiable and un-source-ble information. You’ve created a snowball of expectations that provided that research piece is ever-finished will only lead to disappointing people.

I have no technical background in video game programming or design methodology; the information I know and understand is gleaned from technical documents, discussions on the web and direct questioning toward people who do know. You yourself seem to be in the same position, which is why the validity of your technical information is sometimes vague, hence the reason the thread has continued, as it had done. I cannot and will not accept VF3 nor Shenmue push 500,000 polygons a second regardless of the detail Shenmue displays. The PlayStation geometry engine is a very fast matrix math processor that exceeds the speed of the matrix multiples on the SH2’s, yet no commercial piece of software on the PS1 ever pushed much further than 200,000 polygons a second (Iron and Blood). I can’t see how any of the systems three main processors (Twin SH-2’s and SCU DSP) could have been eked out to provide that amount of performance even calculating in parallel. Besides I think its been discussed VDP1 also simply doesn't have enough fill-rate or bandwidth.

I find your re-reading of Sega and VF4’s success fascinating, VF3 was by all means not the arcade smash Sega’s hoped it to be, there are discussions on fighting forums about how for a long period people couldn’t even get to grips with how to play the game competitively and ditched the game, until a Korean player ShinZ created the two-step technique that improved the games tournament standing. VF3 was a disappointment to many Japanese players, and was not the smash success VF2 was, and it seems to me the game was more a display of Yu Suzuki’s hubris at the time, along with along the money funnelled into his grandiose project, Shenmue (a game which was a hard sell to the public, and not the DCs killer app) and research into exotic and overly expensive arcade hardware. VF3 provided a new game mechanic that was poorly implemented and removed some of the purity of the original 3 Button systems, which subsequent instalments returned to.

The arcade cabinet had far lower distribution simply because of its immense cost comparatively to other hardware, and this only curtails user base and also interest, hence why VF3 was usurped by Tekken 3. Virtua Fighter was just not a hot ticket anymore, coupled with a basic and phoned in Dreamcast port, its no wonder people turned and saw Soul Calibur as a defining title. Virtua Fighter 4 comparatively was a hit even in the dwindling and fading arcade scene in Japan. Whatever you may think about the game stylistically, technically it played much better and was met much more warmly from the competitive scene. Sega’s success being tied to VF3 performing well on Saturn, is a fairly mistaken argument. If anything Sega’s success was tied to VF3 being a different game than what it turned out to be running on cheaper hardware, VF3 lost Sega mind-share in the maturing arcade market at a time it needed it most.

 

Funny you should say that...








I found these screenshots of VF3 Saturn online, they look pretty legit in my opinion.



I think this is from VF3 saturn too.

 

Those shots look very convincing. For everything but the levels with the stairs there would be an easy work around to avoid too many polygons being wasted on the backrounds. Even games like Mechwarrior and Gun Griffon show that the Saturn was capable of using polygons to add graduation to VDP2 floors. It was a simple matter. So, most of Kage's and Aoi's stage could be replicated with a VDP 2 floor with polygons or even a second floor making up the difference.

The Great Wall and Lion's stage may have been possible, it just would have been technical wizardry to the Nth degree. None of it would have ever surpassed the Dreamcast port besides.

 

Oh for pete's sake, those are screenshots from a decade old Model 3 emulator, theres about 4 or 5 threads on different forums on the net where posters have posted those very same shots and had them debunked by various individuals. I'm sure theres probably one on this very forums.

No way would even the Saturn version look as good as those borfed pre alpha emulator shots.