32X=Destruction of Saturn?

[kgenthe]

I haven't been on these boards that long, so I'm not sure where you've seen my comments.

"or so" I just remember seeing your name more often than not throughout the past half year during various threads.

My main issue with dithering is as I said, emulation brings it out more than it ever appeared on a screen. Actually, lower color counts apparently don't stand out as much even in RGB as it does in emulation on a high resolution monitor from today.

So are you advocating that game developers from yesteryear should have used less dithering? Or advocating that retro-gamers shouldn't use emulators? Or something else?

The tech heads might find this a "duh factor" but I cannot tell you how many times I've seen well designed and graphically appealing games reviewed via emulation and trashed for "bad colors" or excessive dithering. Actually, the reviews don't always go that far, they just refer to the game's graphics as having "aged badly" and give it a thumbs down.

I haven't read many reviews that mention dithering, but my focus is also more towards video than written reviews. But I do agree that it bugs me when a game isn't reviewed with original hardware. That may be another topic altogether. All of my reviews are written based on original carts/discs on real hardware. Though I do use emulation for screenshots. The irony

I also find it interesting that dithered transparency on Saturn is well discussed, but full screen dithering on PS1 is so unknown it is doubted constantly on any board. The Saturn's dithered transparency is usually referred to as an eye sore and as evidence of its weakness at 3D. So, I find it odd that full screen dithering is neither of the above even in tech minds.

I think I can start to understand how this conversation ended up in a Saturn thread. And also why the composite versus RGB thing came into play. All of my retro games are played through a 19" TV with composite leads. So I don't see much dithering, and faked transparency tends to looks great. My TV also has some nice filtering built in, as I never see any rainbowing artifacts.

Does that mean that games should be played through composite leads? It reminds me of A/V talk where people argue to the death about aspect ratio's and "director's intent." For example, Avatar was released in theaters at 2.35, but 1.78 for the Blu-ray release (it's actually a little more complicated, as the aspect ratio depended on wether you were seeing it in 2D or 3D). In this case, the "director's intent" depended on how the movie was being seen. Prior to HDTV, most films were hacked down to 4:3 for VHS and cable channels.

Back on topic. How did the game producers intend for Genesis games to be played, composite or RGB? My guess is they didn't think twice about it. They intended them to be played on smaller displays from the comfort of a couch. From this vantage point, combined with the fact that games are not stationary, it doesn't matter.

The Genesis did have a limited color palette compared to it's 16-bit contemporaries of the time. And while some may attack the Genesis or imply it's an inferior machine based on technical aspect of the machine, it doesn't make them right. As the previous pages have shown, all pixel artists of the time utilized dithering in their art. It doesn't make them lazy, it doesn't make dithering bad, it doesn't mean "composite video" should be used when playing retro games.

I can't wait for their to be a cheaper RGB to HDMI solution so I can move some of the retro consoles to the living room and enjoy them on the Big Screen with Big Sound. Currently, I have a nice VGA to HDMI adapter for the Dreamcast, and love every single little jaggie that is revealed with this set up!

I agree, dithering didn't matter back in the day, it matters today though because people are talking about it.

With all of that rambling aside, thank you for your tactful response. I'm still a bit fuzzy on where you stand

[sheath]

I dislike dithering when I can see it, and consider it an artistic choice when it succeeds at blending into new colors.

I have been harping on this topic in various boards (and Usenet) for about a decade now. The reason I got into the discussion was simple, system fans started using emulation to "prove" one system inferior to the other. Dithering, or low color banding, were really all the evidence they showed.

I have said before that I hate dithering whenever I see it. I prefer art to be viewed in the way it was intended overall. I have personally drawn pictures in dots, in color and in gray. These drawings intent were understood before hand. To look at them under a microscope was far outside of any reasonable criticism.

I know what dithering and other color blending approaches are for, but in games they take on another context. There aren't many games that sell themselves on their artistic approach to graphics (Cell shading aside). Approaches like pointillism are not intentionally applied to games. That is, no game developer says "we've made a pointillistic game, come play it for that."

Games have had abstract settings. In fact that is the second biggest theme next to violence in gaming. Abstract settings are not the same thing as abstract art though. In my opinion, viewing most Genesis games in unfiltered emulation is equivalent to viewing them as abstract art. The dithering and banding was intended to be blended by something.

To be fair, I also find that my SNES library is far too aliased over S-Video to my tastes, and that is something I bought from Nintendo.

[kool kitty89]

I haven't been on these boards that long, so I'm not sure where you've seen my comments. My main issue with dithering is as I said, emulation brings it out more than it ever appeared on a screen. Actually, lower color counts apparently don't stand out as much even in RGB as it does in emulation on a high resolution monitor from today.

The tech heads might find this a "duh factor" but I cannot tell you how many times I've seen well designed and graphically appealing games reviewed via emulation and trashed for "bad colors" or excessive dithering. Actually, the reviews don't always go that far, they just refer to the game's graphics as having "aged badly" and give it a thumbs down.

I also find it interesting that dithered transparency on Saturn is well discussed, but full screen dithering on PS1 is so unknown it is doubted constantly on any board. The Saturn's dithered transparency is usually referred to as an eye sore and as evidence of its weakness at 3D. So, I find it odd that full screen dithering is neither of the above even in tech minds.

I agree, dithering didn't matter back in the day, it matters today though because people are talking about it.

You're wrong though: a good RGB display (be it a high end TV or dedicated monitor) WILL LOOK virtually intentical to what emulators show with zero filtering.

Also, you're screenshots are far more blurry than what I've seen on decent TVs with model 1s.



Anyway, the point is that dithering is useful and often necessary even without any sort of blending, but there are additional factors to consider: if you're optimizing for composite you'd need to take artifacts into account, if making it look best for RGB or all video in general, you have some different considerations, and if you just don't care and want to be lazy (and assume there will be some unspecified amount of blur), then you can make assumptions and not really optimize a game for any display.


Dithering is always a useful option regardless of color depth or resolution, but the question is how to optimize it. It obviously works much better at higher resolutions and is less necessary at higher color depths... and far more flexible when you have no inconsistent artifacts to worry about. (as you do with rainbow banding on the Genesis, Saturn, and PSX -at least early models of the latter 2).

Again, I'd still take a nice sharp image over a blurry mess in most cases. (and, of course, texture filtering is a totally different issue than full-screen blur/artifacting which is also why I like S-video on the N64 rather than blurry composite -even if the jaggies are more visible: like texture filtering was a CHOICE by the programmers, not forced like composite video artifacts -on that note though, it's a shame more -or any?- N64 games didn't opt for dithering as there's quite a few cases where it would be preferable to posterized/banded highcolor -for both filtering and gouraud shading; after playing around with dithering options in some PC games at low resolutions, there's soem definite cases where it's preferable and others where it's not)
Likewise, FMV on the MCD would be FAR more limited without dithering, even if I think error diffusion dithering should have been used more. (unless they needed pattern dithering for better lossless compression... I still don't like the posterized stuff on the MCD)

[tomaitheous]

I have said before that I hate dithering whenever I see it. I prefer art to be viewed in the way it was intended overall.

If this is meant as to view them over composite, then you probably have to concede one important fact. The Genesis high res mode (320x224) is no longer applicable as being higher than 256x224 (in relation to comparing it to other consoles). It's actually less. It roughly equates to about ~210 pixels horizontally (for that res mode) and sometimes less (really depends on the placement of pixels). And it's fairly dirty too, although newer TVs temporal clean the image better now than back they couldn't back then. Dithering was never completely solid. It always had a slightly dirty look to it. At least for NTSC. Newer filters for TV and capture, pretty much equate to the special filters in emulation. I know first hand the capture card stuff is definitely not realistic in representing these old consoles and how they used to work. Matter of fact, blargg's digital NTSC artifact filters are more representative than capture card images/video. My point is, you can't just pick the good and throw out the bad - when it comes to the systems original composite output (or RF which was probably even more popular than composite for Genesis gamer users back then, and even worst game image). That's just as miss-representing as the review sites showing completely raw shots.

Even though it's capture card footage, look at those shots of Ranger-X. The capture card one is muddy and blurry. The emulation shot is crisp and beautiful. I think the emulation shot looks much better like that, in comparison. The checkerboard mesh still works. The function of it isn't lost on me at all.

And there's also the very fact that RGB output was provided on the system. Developers had to know some gamers were going to view/play these games in RGB (yes, even that sharp as raw emulation shots. But even a little analog blur of a low res RGB set isn't going to 'remove' the dithering). They had to expect this. I think, if none of the systems had this output option - then we wouldn't be having this conversation (again, NES as a perfect example). That's not to say some games weren't made to benefit more over composite than RGB (a lot of us are already assuming as such with some game titles, and some are taking it far enough to say every game on the system which is where I have a problem with this), but that doesn't make viewing them (sharp or not) in RGB invalid. Not unless a manual specifically stated "do not play this game in RGB output" or "this game was intended for composite/RF display only" or some such statement. It's just not that clear cut.

Again, this is all in relation to comparing the graphics to other systems. If it was just for showing off games and such - then it really wouldn't matter.

Kool Kitty: There's also the fact of where games have dithering (just like these Genesis titles) and you can clearly see the dithering. Are we not supposed to see it? Was it not intended to be seen on the Genesis, but it's ok for any other system? SNES rarely has dithering, so I'll use the PCE as example. There are games with dithering on the PCE. And you can clearly see it over composite. Such, it's more rare but probably about the same number of titles as those on the Genesis that have vertical bar dithering (which some people seem to be making the exception/line as to where a game should be viewed over composite instead of RGB). How does that factor in? And it's major models don't even have RGB to begin with (nor did the original models even have out of the box RGB). Yet all sites that I've seen show them in raw shot (same with SNES too).

[sheath]

If this is meant as to view them over composite, then you probably have to concede one important fact. The Genesis high res mode (320x224) is no longer applicable as being higher than 256x224 (in relation to comparing it to other consoles).

I have no idea what you are talking about here. I was making a purely static image related comment. If I can see dithering I don't like it. If dithering actually blends, whether it be caused by screen size, distance from screen, or blurring by output, I can appreciate its purpose.

I know I have gone on about Composite before, but that was more due to it as a consumer accessible standard.

Even though it's capture card footage, look at those shots of Ranger-X. The capture card one is muddy and blurry. The emulation shot is crisp and beautiful. I think the emulation shot looks much better like that, in comparison. The checkerboard mesh still works. The function of it isn't lost on me at all.

My capture card shots of Ranger X are muddy and blurry because they were filtered by ATI's software. My only point here was that any degree of filtering created by the actual screens would have blended the dithering better than unfiltered emulation does.

[kool kitty89]

Newguy: I just remembered another thing for PAL composite: not only id the color resolution higher (and thus less compaosite artifacts in general and no rainbow chroma problems), but the MB3514 video encoder is also rather common on model 1s and some model 2s while it's pretty much absent from NTSC models. The MB3514 is a little dark, but sharper than the CXA1145, more like the 32x or CXA1645, so that would mean even less composite blur or artifacts in general.

Even though it's capture card footage, look at those shots of Ranger-X. The capture card one is muddy and blurry. The emulation shot is crisp and beautiful. I think the emulation shot looks much better like that, in comparison. The checkerboard mesh still works. The function of it isn't lost on me at all.

That capture card is blurred to shit compared to real hardware... on a decent TV... except maybe the KA2195D. I can clearly see the dithering in the target box on my TV. (though the composite artifacts and blur are still annoying)

I totally agree, I couldn't care less if I could see dithering so long as it serves a useful purpose and especially if there's a lot of detail in general that is gained at the expense of visible dithering. Tactful art design is always important of course, but compression and other practical tings play roles too (if using lossless compression, checkerboard/ordered dithering is favored over diffuse dithering -granted posterization is even more favored)

And I'd take dithered FMV in RGB over posterized FMV any day... with all else being equal. (one thing that makes that recent Jonny Mnemonic game obviously poor and why I dislike Novastorm's style as such including the 16-color limit -Soulstar's video is a nice compromise of modest -diffuse!- dithering and full per-cell color optimization)
Of course all dithering blends to some extent, though not completely.... even on an emulator sitting a couple feet away from a ~15-20" screen, it blends a fair bit.

It's like saying you can't stand pixelated low-res stuff in general, regardless of dithering, due to the blockiness... and I do prefer a bit of bilinear filtering at times, but I rather like the raw RGB representation as well.

Kool Kitty: There's also the fact of where games have dithering (just like these Genesis titles) and you can clearly see the dithering. Are we not supposed to see it? Was it not intended to be seen on the Genesis, but it's ok for any other system? SNES rarely has dithering, so I'll use the PCE as example. There are games with dithering on the PCE. And you can clearly see it over composite. Such, it's more rare but probably about the same number of titles as those on the Genesis that have vertical bar dithering (which some people seem to be making the exception/line as to where a game should be viewed over composite instead of RGB). How does that factor in? And it's major models don't even have RGB to begin with (nor did the original models even have out of the box RGB). Yet all sites that I've seen show them in raw shot (same with SNES too).

Yeah, they obviously knew it was going to be seen (at least to some extent)... though "meant to be seen" is a bit off as you're "meant" to see the underlying effect of simulated added color for shading/transparency/etc be it blended by poor video or minimally by human perception.

It obviously works better with better algorithms (when doing more than posterized checkerboarding), but in general it works better at higher resolutions.



And there's obviously tons of arcade games to use dithering on sharp RGB monitors... and at pretty low resolutions too.

With the NES, there's plenty of examples, but usually less extreme ones.
An obvious case would be 3D games using 15 colors (so just as limited as virtua racing -for the polygon layer- other than the use of indexed 15-bit RGB -and Star Fox and Vortex seem like they'd cater OK to 9-bit anyway) where dithering is obvious in composite and more so in s-video... though I prefer the lack of distortion or blur in s-video anyway. (SNES composite is pretty clean -though part of it is due to the resolution... the Genesis is pretty close to SNES clarity at H32 depending on model encoder used... though the SNES2 is another story as it has super sharp composite more like the N64)

[j_factor]

You're wrong though: a good RGB display (be it a high end TV or dedicated monitor) WILL LOOK virtually intentical to what emulators show with zero filtering.

I haven't seen that. I've seen a lot of RGB setups, and I have an RGB monitor myself (though I don't yet have the ability to hook my Genesis to it), and I've never seen it look exactly the same as raw emulator shots. Close to it, but not the same. The monitor itself (I assume) softens the image a tad. The only time I've seen real hardware look like an emulator was on an HDTV using converted RGB.

Also, you're screenshots are far more blurry than what I've seen on decent TVs with model 1s.

Definitely. I'm not really sure what you guys are arguing about, though.

[tomaitheous]

I haven't seen that. I've seen a lot of RGB setups, and I have an RGB monitor myself (though I don't yet have the ability to hook my Genesis to it), and I've never seen it look exactly the same as raw emulator shots. Close to it, but not the same. The monitor itself (I assume) softens the image a tad. The only time I've seen real hardware look like an emulator was on an HDTV using converted RGB.

You mean solid without scanline (gaps)? You won't either. HD sets capture the whole image into a frame buffer, then output it to the screen (whatever that may be), thus no need for scanlines. Just solid pixels connecting each other. That's one thing I wish my HD set had... an option for 'scanlines' (it's a CRT but you can't see the scanline gaps at all - even when I put it in 540p mode which you can force anything less than 720p into, in the options menu).

[j_factor]

Yeah, that's what I meant.

[kool kitty89]

I wasn't talking about the gaps or lack thereof, but just the sharpness of the images in general... real CRT monitors (including VGA/SVGA monitors -without hardware line doubling -and if the video card doesn't auto-double the resolution) will tend to show scanline gaps as well at real resolutions... some monitors will do that for higher res than 320x240 too. (we've got a 20" SVGA monitor from an old workstation that has visible gaps in 640x480 too)

So, more accurately: high-end RGB monitors with super crisp, pixel perfect displays (like better ST or Amiga monitors -or any such RGB computer monitor at SDTV sync rates, including early VGA monitors supporting CGA/EGA resolutions) will look rather like emulators with 100% scanlines enabled in 640x480 res... or conversely, displaying an emulator at 320x240 fullscreen on a CRT VGA/SVGA monitor without line doubling or scaling (via the video card or -less often- inside the monitor).
In fact, if you found the right monitor (ie a late 80s VGA/SVGA monitor compatible with SD and higher resolutions), you could even have the Genesis connected to that monitor and directly compare that to a PC using the very same monitor.

Of course, one possible issue is pixel shape (if you're using pure, unscaled, raw mode emulator shots -since Fusion won't display at 256x240 fullscreen -I think ZSNES will though), but you could correct that by manually adjusting the scan width as most PC monitors support externally. (with games in 320x240 H40, it shouldn't matter, but for H32 it would be an issue for emulator vs hardware)


Of course, there may be a buffering/sync stability issue causing the vertical line problem in RGB...

[Team Andromeda]

Again, I'd still take a nice sharp image over a blurry mess in most cases

Not when it comes to FMV on the Mega CD. RGB on the Mega Drive just makes the FMV look really nasty, more so if one using the Model 1 Mega Drive, which output the best RGB signal out of any console, for some reason looks better than the RGB of the Model 2 or Multi Mega console too . The RGB output of the Saturn is mighty too (better than the DC Ect)

So, more accurately: high-end RGB monitors with super crisp, pixel perfect displays

Yes it does, I have a old Philips monitor and it looks bloody awesome, only trouble is the monitor is only 14' and can only take a 50 Hz signal

[kool kitty89]

Not when it comes to FMV on the Mega CD. RGB on the Mega Drive just makes the FMV look really nasty, more so if one using the Model 1 Mega Drive, which output the best RGB signal out of any console, for some reason looks better than the RGB of the Model 2 or Multi Mega console too . The RGB output of the Saturn is mighty too (better than the DC Ect)

Meh, I don't mind unfiltered FMV too much... either in emulators or vial component on real hardware. (got a chance to see that in action a bit over at arnoldthebartender's house on his Sega CD and an HDTV -so it was pretty sharp)
I've played around a good amount with emulation and comparing composite video lien blurring, etc, and more ofthen than not you end up with different artifacts from blending due to the dithering mechanisms used, usually better in some ways but rarely ever ideal.

I end up leaving FMV unfiltered more often than not in emulation (in part due to aspect ratio and scaling issues, but also funky blending artifacts), and interestingly, the games that blend better are often also those that use more pleasing (or sparing) dithering in general like SoulStar or Dragon's Lair. (some are pure posterization, so composite video does no good at all)
If TVs/emulators had vertical blending it would be ideal for the dithering methods often used on MCD FMV, but that's not the case and you always have sharp vertical resolution.

Composite on a decent TV (even in NTSC -especially since most FMV is in low res anyway) still is pretty grainy in composite and not nearly as blended as CVBS emulator filters (on my TVs at least), let alone some youtube uploads and the composite artifacts are still a pain in any case. I think you lose some of that in PAL, but in NTSC there's almost always noticeable artifacts beyond bluring/smearing, especially dot crawl or other luma artifacts (rather heavy "picket fence" edge artifacts are one of the biggest issues on the genesis and related vertical line artifacts).
Then again, you've also got the funky aspect ratio problem in PAL (especially due to the common use of 256 wide on FMV) as well as the 50 Hz skipping issue for non-optimized games. (I'm not sure, but some European games also screwed that up -Novastorm seems to have issues with the intro in NTSC)

Of course, you have examples that look crap in either case... and Rebel Assault is one of those blatantly obvious cases (it's pretty nasty on PC/3DO/Mac even) which is something that someone else (not really tech savvy) noticed right away as well a couple weeks ago while I was at arnoldthebartender's... interestingly he also noticed the obvious improvement with Sewer Shark by comparison and much more so with Loadstar and some other later gen titles. (he actually commented that loadstar looked almost like low quality VHS VCR recordings -mind you this was via component video on an LCD)

[TidalTempest]

I know the conversation seems to have moved on from the original topic, and as interesting as the current conversation is, I'd like to leave my two cents worth on the original topic.

Was the 32X the destruction of the Saturn?

Yes and no.

In the end, it was Sega's internal bickering and mismanagement that ruined the chances of the Saturn becoming successful, and lead to the demise of the company's hardware manufacturing endevours.

But ultimately, at least in the eyes of consumers, and probably developers, Sega releasing the 32X, not supporting it much and then dropping it (and all of their other systems except the Saturn) lost the support and confidence of consumers in the company, and thus, in the Saturn.

People also argue that the Saturn was hard to program for. Next to the Playstation, which was well documented, developer friendly, and programmable in the vastly easier to C, then it was. But developers had been learning how to utilise console hardware through prgramming languages like Assembly in the past, and given no alternative, would have done the same with the Saturn, had the Playstation not made everything too easy.

The 32X was a decent piece of kit, but the time and money spent on the system should have been devoted to the Saturn, and the same goes for the software developed for it.

In 1994, the Mega CD was finally becoming into it's own, not only in terms of quality software, but it was no cheap enough for the average consumer to take a chance on. Developers were already struggling with the mass number of new consoles coming onto the market, and didn't need the short-lived 32X on top of it. Had Sega fully supported the Mega Drive and Mega CD well into 1996, shifting their focus onto the Saturn in late 1995 while using the MD and CD as secondary, cheap consoles, before gradually fading them out by 1998 when the Saturn would have been in it's prime, they would have had the increased cash flow to compete with the Sony juggernaut and consumers and developers would have had more confidence in them.

But then Sega, especially SOJ, were arrogant and unwise with their money, and we all know what happened. But sometimes that's what it takes to be the best - their passion for technology advancements and creating quality software are what attracted us to them and keep us talking about them today.

[kool kitty89]

In the end, it was Sega's internal bickering and mismanagement that ruined the chances of the Saturn becoming successful, and lead to the demise of the company's hardware manufacturing endevours.

There seems to be a lot of vague and unsubstantiated info on those issues, so it's really unclear exactly what the context and impact of said "bickering" really was... but aside form any true infighting, there were certainly conflicts between management as well as conflicts of interest in transitioning from the Genesis with a ton of mistakes made between 1994 and 1998. (still others that came in '98-2001 but everything done from '98 on was exacerbated by the problems prior to that)

But ultimately, at least in the eyes of consumers, and probably developers, Sega releasing the 32X, not supporting it much and then dropping it (and all of their other systems except the Saturn) lost the support and confidence of consumers in the company, and thus, in the Saturn.

In its main active period (late '94 to mid '95), first party software support was relatively strong... but the conflicts with the Saturn and to some extent Sega CD and the Genesis itself hurt it and as it hurt/complicated those... though cutting it like they did only made the situation worse. (phase otu needed to be done more carefully -something they screwed up with Saturn later on -albeit worse as they had no fallback like they did in '95/96)
And the messed up mangement of the Genesis, CD, Saturn, and Game Gear coinciding with all that made it a bigger mess. (some interdependent to the 32x, some unrelated -or tied directly to Saturn or the Arcade problems)

People also argue that the Saturn was hard to program for. Next to the Playstation, which was well documented, developer friendly, and programmable in the vastly easier to C, then it was. But developers had been learning how to utilise console hardware through prgramming languages like Assembly in the past, and given no alternative, would have done the same with the Saturn, had the Playstation not made everything too easy.

C vs assembly is not the only issue... indeed better PSX games also used assembly for optimal performance with the limited CPU resource... but the "difficult to program for" aspects of the Saturn extended far beyond that to the general system configuration and graphics architecture and then the fundumentally inferior technical areas over the PSX. (peak bandwidth, geometry performance, polygon filling/texture mapping performance -and by extension sprite/blitter 2D performance -VDP2's BG generation was the big win for 2D on the Saturn but VDP1 was pretty weak compared to the PSX's GPU in 2D- among other things)
You had difficulties from working with quads (or workarounds to avoid them), dual CPUs (potential performance advantage but a lot of work to actually achieve it, but then losing it again for the many cases where CPU resource is eaten up by operations the PSX can do in hardware -3D calculations, data decompression, etc -the Saturn used the DSP in the SCU for 3D math, but from what I understand that was rarely done -possibly in part due to working with quads making software floating point math being preferred to DSP driven fixed point math or the DSP being harder to work with in general -and even then the DSP in the Saturn is far less powerful than the GTE in the PSX, though it is more flexible for potential coprocessing beyond 3D math -not sure if that was ever used- and I don't think the GTE in the PSX was ever the limiting performance factor -it could crank out vertexes extremely fast, but there's still only so much the GPU can draw -and then there's software z-sorting, AI/game logic, etc)

The 32X was a decent piece of kit, but the time and money spent on the system should have been devoted to the Saturn, and the same goes for the software developed for it.

That wouldn't solve the problems though... but re-investing those funds into other projects (MD/CD/Saturn/GG) would almost certainly have been better all around. (I'm really not sure what SoJ had in mind with the Mars project, but it seems SoA misunderstood the context of the Saturn -or at least did until it was too late to practically turn back or shift things) In context of the MD/CD/Saturn as it was, any additional add-on for the MD would have had to be very cost effective and generally low in price to be realistic at all. (preferably low enough for MCD owners to reasonably adopt and possibly to integrate with a concurrently released duo system -ie the CD-X done right, with a low cost emphasis and lower price point than the MD+CD prices)
A bare bones SVP addon. (probably with some added audio -maybe something like the 32x's PWM) or maybe the more powerful DSP in the Saturn rather than the SVP, or maybe an SH1 or SH2 on a cart. (the latter would be much safer in terms of being developer friendly and generally flexible, but more expensive -though given the price of the 32x, it probably wouldn't be totally unreasonable to think of a~$50-80 stripped down lock-on cart add-on with a single SH2 or SH1 with no VDP and only the 256k SDRAM chip plus custom MD interface+sound chip... but a DSP/SVP based cart would probably more easily manage the $50 mark) If they did use an SH CPU as such, it would at least have the advantage of promoting good coding for that CPU architecture... but not for managing dual CPUs or the many other unique aspects of the Saturn. (and using the Saturn's DSP might have been even more significant in that sense)

In 1994, the Mega CD was finally becoming into it's own, not only in terms of quality software, but it was no cheap enough for the average consumer to take a chance on. Developers were already struggling with the mass number of new consoles coming onto the market, and didn't need the short-lived 32X on top of it. Had Sega fully supported the Mega Drive and Mega CD well into 1996, shifting their focus onto the Saturn in late 1995 while using the MD and CD as secondary, cheap consoles, before gradually fading them out by 1998 when the Saturn would have been in it's prime, they would have had the increased cash flow to compete with the Sony juggernaut and consumers and developers would have had more confidence in them.

Yeah, they'd made some mistakes with the MCD, but it was certainly coming into its own in '94... who knows what might have happened with more focus on it. (it wasn't too late to push for a real cost-optimized duo system as I mentioned above... given the $150+100 CD+Genesis prices, $200 in fall of 1994 for a duo would not have been unrealistic -albeit if it included an SVP/DSP/etc add-on internally, that might make it more like $250 -with other trade-offs of splitting the market too)

Among other things, a lack of SFII in 1993 hurt the CD, but a really good appearance in '94 could have still been extremely significant, or '95 for that matter (especially with SSFII)... or a combo cart+CD game, especially given the amount of space SSFII used (managing that with 512-768 kB loading chunks would be pretty tight -and probably require on the fly compression and such to work well). hat would have been especially nice given the fact that SSFII on the MD pushed very heavily for graphics at the expense of audio, so the companion CD could have replaced the limited sfx and music with arcade quality PCM SFX and music.
If Capcom wasn't especially interested in such, Sega potentially could have licensed their own version like the did for several other Capcom titles (Ghouls n' Ghosts, Final Fight, etc), though in that case it probably would have been unrelated to the Capcom MD SSFII. (still potentially hybrid cart+CD though -but could have been even more optimized for CD and not allowed the cart to be used standalone)
And of course there's SSFII Turbo which never got ported to the MD at all. (but did on the SNES)

But then Sega, especially SOJ, were arrogant and unwise with their money, and we all know what happened. But sometimes that's what it takes to be the best - their passion for technology advancements and creating quality software are what attracted us to them and keep us talking about them today.

I don't think "we all know what happened" and I for one am extremely interested in knowing the real details behind the events from '93-97 if not a bit broader. (like some more details in the 1990-92 period as well)
There's so much speculation and commentary in the info out there and so much else that's questionable or conflicting with other info that it's really not solidified at all.


I'll point again to what's happened with Atari history more recently with tons of "established facts" and "documented events" being totally overturned or heavily revised with the true history being uncovered by some very dedicated historians. (granted, the amount of stuff tied up in Japan certainly limits the possibilities for that with Sega -and the fact that it's still a fully operational company... and that those events were far more recent than the Atari stuff -which took over 30 years to be uncovered and properly documented -and they're still working on that with some additional details coming to light and the actual formal books resulting from said work still to be written and published)