Sega Model 3 & The Dreamcast - WTF Went Wrong?!?

Just on the theme of this thread, wasn't Virtual On Oratorio Tangram also a model 3 game with a dreamcast port? I bought it back in the day and was quite unimpressed with the graphics compared to other more mature DC games of the time. Definitely think I would have liked SCUD race more.

It seems a little strange sega would have spent the time and resources porting a marginal game like VOOT, although I guess it has its own following and/or was popular in Japan.

 

I always forget about Oratorio Tangram's conversion for the Dreamcast, mainly because it was never released over here in Europe. I've just had a look at its page on Sega Retro, and it seems this project was handled by many of the same developers that also worked on the Saturn effort, including CRI lead programmer Toru Kujirai (who supposedly got a basic prototype of Sega Touring Car running on the console in just a matter of days - his earliest playable build is in the hands of a known collector).

 

So there is a proof concept build of Sega Tour Car Championship for Dreamcast? Or are you taking about a Saturn build?

 

Sorry for the late reply - I was very much referring to an early Saturn prototype build of Sega Touring Car (that is supposedly just 5% complete!) with flat shaded AI vehicles and an empty white box serving as a placeholder for the eventual rear view mirror. My understanding is that only one track was playable at this stage in the conversion process, and by 60% the team at CRI already had most of the final game's planned content fully implemented, the remaining time being something they hoped to spend on further optimising their base engine's performance. I wouldn't be surprised if the end result was similar to The House Of The Dead, which suffered as a result of Sega advising Tantalus to release whatever they had or risk missing out entirely, since the company wanted to kill off the Saturn and focus its limited resources on preparing to launch the Dreamcast.

 

Yeah HotD on the Saturn was so poor toward the end. You can tell that Tantalus just released it in an unfinished state. As one of my favorite arcade games I did a little bit of research regarding the Saturn port. From what I remember a lot of the textures towards the end of the game were just placeholders. The game starts out okay but towards the end you can tell they just didn't put certain textures and assets in. I guess they wanted to work on the framerate as well but due to the rush didn't have time to optimize anything. I'd like to be corrected if I'm wrong on any of that. I played so much of HotD when I was younger.

 

At the point Tantalus were told by Sega to release whatever they already had done of Saturn HOTD, the conversion had reached a stage where the arcade original's four levels were only just functional. Based on preview coverage in SSM, it's clear the team had implemented each area in their seen order, and the last two sections especially suffered badly from the use of placeholder textures and slowdown. From what I can remember, Tantalus had planned to further refine their work in the last 40% or so of development, with removing the need for breaks in the gameplay to stream new assets one of their biggest goals that remained unfulfilled. More than Jon Burton's recent "director's cut" of Sonic 3D, I'd love to see what Tantalus may have intended for The House Of The Dead on the Saturn, if only they'd not been given such an unrealistic deadline by the powers that be at Sega. Of course, there's little chance of those final textures surviving all this time, plus I can't imagine them wanting to revisit such a controversial project... I once had occasional contact with one of the programmers who worked on this game, only for all communication to abruptly end when I started asking questions about its incomplete state.

P.S. During the same meeting where executives from Sega Europe demanded that HOTD be released in what is technically a prototype form, they also signed up Courier Crisis and Deep Fear (while neither of these ultimately made it to the US, at least a final build of the latter's NTSC variation had since leaked). Sadly, negotiations between Sega and Capcom to possibly give PAL Saturn owners X-Men VS Street Fighter with the necessary 4MB RAM upgrade cartridge failed because Virgin didn't want the whole package going to retail with a suggested price of £49.99 - Sega's argument was that few players would have paid Virgin's suggested £79.99 figure even when the Saturn was at its commercial peak, depriving us of a further late classic to add to our software collections!

 

Oh wow. Definitely a sore spot. Was anything else said about the port prior to that?

 

I'd been sending questions in batches and receiving lengthy replies via e-mail (for a Saturn-related project I long since had to abandon - blame a computer giving up the ghost before I was able to archive nearly a year of work), though no further contact took place after I brought up the subject of HOTD being released in an unfinished state. Prior to this, I learned quite a lot regarding the conversions of Manx TT and WipEout 2097, which both used the proprietary "Duck" engine that was developed primarily by Tantalus with assistance from AM3 staff, who flew over to assist with the former.

Thinking back, it was another member of the team who first mentioned what Tantalus had wished to do with the remaining 40% of production time they'd scheduled for HOTD, or at least that was until an executive from Sega Europe demanded they have something on shelves by Easter 1998 or risk it never coming out. Curiously, there was never any talk of Tantalus moving this over to the Dreamcast, but it's likely there were already plans to port its sequel over from the more technically similar NAOMI board, meaning there was little need for the original HOTD as well.

P.S. Tantalus may not have completed anything commercially for the Dreamcast, but they did produce a tech demo for a rather impressive-looking FPS engine that could support up to six b... I mean four players!

 

Isn't the DC more powerful than Model 3?

 

There was an article in one of the later issues of SSM that had members from Sega of Japan's hardware design team insisting their newest console was more powerful than its flagship arcade board, but I never felt convinced - especially not when Sonic Adventure was mentioned around the same period as some kind of great technical benchmark. Sure, on paper the Dreamcast was capable of higher resolutions and had more advanced features like texture compression, though I don't ever recall any of its Model 3 conversions being truly arcade perfect. NAOMI was obviously a much closer match, with the console only having less memory (and even that didn't lead to many problems from what I can remember). I wouldn't be surprised if Sega pushed its development groups to concentrate more on NAOMI and subsequent various revisions of this as a cost-cutting measure rather than because anyone genuinely believed these offered an improvement over the Model 3, especially when you look at what happened to the company soon after... Just compare something like Sega GT to the visual quality of virtually every Model 3 arcade racer and try telling me you don't prefer the coin-ops!

 

I've heard conflicting things about this. There are quotes from people at Sega suggesting both that the Model 3 is more powerful and that the Dreamcast is more powerful. I think, though I'm not sure, that the Dreamcast can render more polygons/second than the Model 3. The Model 3 has some hardware features the Dreamcast lacks, like edge anti-aliasing. The Dreamcast doesn't have hardware T&L exactly like the Model 3, but its CPU has a graphics unit that does geometry and lighting calculations, so I guess it's kind of similar. The Dreamcast has hardware bump mapping, which the Model 3 doesn't have. Like you mentioned before, the Model 3 has more graphics memory.

Regardless of the differences in the hardware, I think the Dreamcast should've been able to handle pretty good ports of Model 3 games. Some good ports were made, like Fighting Vipers 2. That port was fairly accurate, though it's not the best Model 3 game, and the arcade version ran on Step 2.0. I think the fact that a lot of Model 3 ports weren't very good is because not enough effort was put into them. The Dreamcast should've easily been able to do a near-perfect port of Virtua Fighter 3 - I own the arcade version of this game, and it definitely seems to me that the Dreamcast should be capable of handling it. The same applies to Scud Race.

Sega Rally 2 had performance issues because of the fact that it used Windows CE - this was a common thing with games that used Windows CE. Half Life suffered for the same reason.

I think that if enough time and effort was put into it, most Model 3 games could've had good Dreamcast ports. With games that ran on the later steps it would've been more difficult, but still possible, I think. The effort was much less when porting Naomi games, so that's probably why they focused more on that. The Naomi was also much cheaper to produce than the Model 3, so that was another reason to focus on developing for the Naomi.

Here's a question: would the Model 3 have been able to run Shenmue perfectly? Assuming storage weren't an issue. I don't think so. No Model 3 game was as complex as Shenmue, though that doesn't necessarily mean it wasn't possible.

The Dreamcast also had a short lifespan. I think that if it had lived longer, developers would've had a better understanding of the hardware and could've produced better graphics.

Correct me if anything I've said is wrong.

 

I'd definitely have to agree that Fighting Vipers 2 showed what could be done with a Model 3 conversion when the programmers weren't limited by short deadlines or having Windows CE pushed on them, but for some reason it's still not a game that I've ever really been able to enjoy, which is quite odd considering how much I love the original. As for whether Sega's high-end board or any of its later revisions could have recreated something like Shenmue, I'm equally not convinced. Yu Suzuki once mentioned that Shenmue was built from the ground up - after an initial phase on Saturn, of course! - to take full advantage of the Dreamcast hardware, and it probably speaks volumes that only Shenmue II has so far turned up on another platform. For those unaware, much of Shenmue II was already completed before the release of the first installment in this series, explaining why a lot of early screenshots previewed scenes that wouldn't be played until the sequel. AM2 made huge advances over the course of developing Shenmue, and while a lot of the improvements were only superficial, it's telling that Shenmue II doesn't ever show details such as realistically moving cords on telephones, not to mention its textures suffer from a lack of MIP mapping. Behind their similar visuals, Shenmue was doing things its supposedly superior follow-up didn't, including huge amounts of simulation that proved so subtle few appear to have even noticed these calculations taking place, never mind spotting their results.

 

Yu Suzuki also looked up Japans weather data from the same time period and used it in Shenmue so all the weather is the same as how it actually was in real life on those dates. That was something that wasn't in Shenmue II.

 

i wouldn't say you're wrong. but half-life on dreamcast actually has a crapload of graphics cvars that aren't set, like for fog, only rendering a certain distance in front of you, shadow dithering, a whole bunch of stuff. none are documented, so you pretty much have to find them all yourself and tweak. but if someone did and added a CFG file to the game, you could boost performance quite substantially. these probably aren't set because the build we got was still in development, and they didn't get to the polishing yet. but i can assure that it is possible to get Half-Life to run around 45fps with a lot of tweaking, if not a full 60fps

 

I thought the game was basically finished, and they only canceled it after it was finished.

I like it, personally. My only real criticism is that maybe the single player mode is a little too easy. Also, certain textures in some levels aren't filtered, which is strange for a DC game. But that isn't a big deal.

Do you know of any examples of that? I know a lot of what was in Shenmue II was developed first, but there are definitely things in Shenmue II that exceeded what was in the first Shenmue. The most obvious thing is that the areas were larger, and there were more of them.

 

My understanding is that while Half-Life for the Dreamcast was cancelled after the game had been finished, the version that leaked isn't a complete build - instead, this was supposedly a late beta with some missing optimisation implemented during the very final stages of development, including modifications to strip out the networking functions planned earlier. As for Shenmue, the first game did indeed make use of the so-called Magic Weather System, yet its immediate sequel didn't, further proving that the original was more technically advanced.

 

The one we got is likely a pre final build, as it gives a crash error sometimes.

The MP component as far as I know was meant to be released later on, so wether or not they only had that code or much more is unknown at the moment.

 

I wasn't aware there were plans to release the multiplayer elements of Half-Life separately after the standalone disc, but that does make sense. Unfortunately, by that point there would have been even fewer developers still continuing to support the Dreamcast, so we'd probably have been stuck with just the single player experience and its Blue Shift add-on, which I believe was always intended to be included with the main game.

P.S. Does anyone know if split-screen was ever in the planning for Half-Life on Dreamcast? I'm surprised there has never been any mention of this as a possibility (to my knowledge, at least), especially since both Quake III Arena and Unreal Tournament both had such options, meaning its absence probably wasn't due to any technical issue on the console's part.

 

Unreal Tournament has slowdown even in single-player mode, though. I haven’t played it in split-screen mode, but I’d guess that it would have even more slowdown.

 

Yep, as the game was nearing the completition, they decided that the MP component would be released as a standalone, likely because of size issues. Blue Shift was always planned as the DC exclusive content, but after the many delays in 2000, Sierra decided to bring it to PC as a standalone expansion pack that has the HD models patch.