Questions about Sega Saturn

Hello community!
First of all congratulate them for this great thread.
I'm also a big fan of the Sega Saturn.
And despite the fact that I have investigated deeply. Through their SDK and official documentation.
I still have doubts, doubts that through emulation, and more specifically yabause, are becoming less.
They could help me with a couple of technical doubts:
1) Why the gourand shading of the SS has less amount of color in its degradation than the PSX? In theory, the SS this effect does with 16bit tables color index, that is, the PSX, to where it is rendered in 16bpp. So why is the PSX gradient of higher quality?
2) Although the VDP1 of the SS, has the ability to make half-transparencies, over several layers, within the process entities process of this unit. That is VDP1 vs. VDP1 but not against VDP2. Because so little was used at the time. Using more of the VDP1 option against VDP2, limiting the ability to do multilayer transparency. From what I read, there were performance issues, but seeing games like Loaded, where semi-transparencies VDP1 vs. VDP1 are used in many elements, and overall game performance and quality is very similar to the PSX version.

Many thanks in advance for helping me to better understand this great machine. Complex but very powerful. :)

Regards!
 
Hello community!
First of all congratulate them for this great thread.
I'm also a big fan of the Sega Saturn.
And despite the fact that I have investigated deeply. Through their SDK and official documentation.
I still have doubts, doubts that through emulation, and more specifically yabause, are becoming less.
They could help me with a couple of technical doubts:
1) Why the gourand shading of the SS has less amount of color in its degradation than the PSX? In theory, the SS this effect does with 16bit tables color index, that is, the PSX, to where it is rendered in 16bpp. So why is the PSX gradient of higher quality?
2) Although the VDP1 of the SS, has the ability to make half-transparencies, over several layers, within the process entities process of this unit. That is VDP1 vs. VDP1 but not against VDP2. Because so little was used at the time. Using more of the VDP1 option against VDP2, limiting the ability to do multilayer transparency. From what I read, there were performance issues, but seeing games like Loaded, where semi-transparencies VDP1 vs. VDP1 are used in many elements, and overall game performance and quality is very similar to the PSX version.

Many thanks in advance for helping me to better understand this great machine. Complex but very powerful. :)

Regards!

Welcome. I'll try to answer these questions.. but it's pretty technical so it might be difficult to explain clearly.

1) This is how gouraud shading works on PS1. All of the points on the triangle have a 24-bit color value (not 15-bit). Each component (red, green, and blue) has an 8-bit value that represents a number between 0 and 2. Those values are averaged across the triangle, then multiplied by the texture color to shade it. The final color is then converted to 5-bits per component either with or without dithering. Most of the time shading doesn't increase intensity, so you can think of these values as being between 0 and 1 or having about 7-bits of precision.

On Saturn a 15-bits is used for each color, or 5-bits per component. They're averaged to represent a number between -0.5 and 0.5, which is then added to the texture color to shade it, and then it's output to the screen without dithering. So, this is inferior to PS1's method because:

a) Addition doesn't perform as correct of lighting model, so the transitions look less evenly stepped.
b) To get lighting that can go from full intensity to zero intensity the texture values have to be biased. But then this cuts off how naturally dark they can be. So you either give up dynamic range of shading or you give up dynamic range of the textures.
c) PS1's colors effectively have fractional input bits, so even when they're thrown out in the end (no dithering) they still give smoother shading transitions as the polygons move around. Similar to why sub-coordinate texel precision is desirable, something both consoles lack.
d) Dithering arguably looks better sometimes (less visible banding), especially when used in higher resolutions than 320x240.
e) I didn't mention this above, but PS1 has a more consistent averaging algorithm for the color values. On Saturn, the sprites are drawn one line at a time, and when it switches from one line to another the averaged color values are rounded down to whole numbers. PS1 on the other hand maintains 12 extra fractional bits for every value it's averaging throughout the triangle (both color components and texture coordinates)

2) The biggest why VDP1 half-transparency is rarely used in 3D (it's used all the time for 2D) is that it causes glitches because some pixels inside the sprite get drawn multiple times, so it has spots blending against itself basically creating sections that have higher intensity than they should. And these spots change a lot as the sprite's deformation changes, making it look very visually jarring unless. These glitches don't happen in 2D so long as the sprites are only scaled and not rotated or otherwise deformed.

The other issue is that VDP1 half-transparency is pretty slow. The mesh patterning feature is free and a lot of TVs at the time were pretty blurry so it looked almost the same.
 
I have a question about quads.
I dont remember the numbers but if I recall the Saturn was reported to be able to do around 300k polygons per second whereas the PS1 was reported at around 500K polygons.
But the Saturn of course used quads. Did Saturn's numbers refer to quads and the PS1 numbers refer to triangles?
A PS1 quad surface would be the equivalent of 2 triangles. The PS1 could be the equivalent of approx 250k quads if this is the case. Perhaps one of the reasons the Saturn was having issues with drawing 3D was the fact that using only quads was forcing the console to have less polygon budget and of course quads were less flexible back then when everything was made of low geometry.
 
I have a question about quads.
I dont remember the numbers but if I recall the Saturn was reported to be able to do around 300k polygons per second whereas the PS1 was reported at around 500K polygons.
But the Saturn of course used quads. Did Saturn's numbers refer to quads and the PS1 numbers refer to triangles?
A PS1 quad surface would be the equivalent of 2 triangles. The PS1 could be the equivalent of approx 250k quads if this is the case. Perhaps one of the reasons the Saturn was having issues with drawing 3D was the fact that using only quads was forcing the console to have less polygon budget and of course quads were less flexible back then when everything was made of low geometry.

I don't know what Sony meant when they originally said 500K polygons, maybe it had something to do with their idea of maximum vertex throughput through the geometry engine. You always have to be careful with company advertised triangle/polygon counts because it's often not clear what they're talking about (or if they're not just making it up entirely). It could refer to transformation rate, counting vertexes that are shared by many polygons, or maybe polygons that are backface or frustum culled before even reaching the rasterizer.

The maximum number of triangles the GPU could actually process is 360K/second. And that is for triangles that are neither textured nor shaded. Maximum rate for shaded and untextured triangles is 180K. Maximum rate for shaded and textured triangles is 90K. These numbers are absolute best cases and may not be attainable in realistic situations. As far as actual real world numbers go I've seen some game scenes with 3000-3500 triangles. I don't know the framerate but I'd assume it was around 15-30 FPS, so maybe 100K polygons/sec, and these games were pretty good about avoiding shading or texturing when they could get away with it.

On PS1 quads are split into two triangles on the GPU. They're useful for saving memory and bandwidth but they don't increase the draw rate.

I don't know about Saturn's actual performance or where the 300K number really comes from but I'd be pretty surprised if any game was remotely close to that in practice. The best looking 3D games on Saturn made very effective use of VDP2 planes, not high polygon count models. PS1 games often benefit a lot from being emulated at higher resolutions while Saturn games not nearly as much (see the NeoGAF thread I linked to earlier) - this is a sign that at least some PS1 games were achieving higher polygon pixel density/polygon counts.
 
The maximum number of triangles the GPU could actually process is 360K/second. And that is for triangles that are neither textured nor shaded. Maximum rate for shaded and untextured triangles is 180K. Maximum rate for shaded and textured triangles is 90K. These numbers are absolute best cases and may not be attainable in realistic situations. As far as actual real world numbers go I've seen some game scenes with 3000-3500 triangles. I don't know the framerate but I'd assume it was around 15-30 FPS, so maybe 100K polygons/sec, and these games were pretty good about avoiding shading or texturing when they could get away with it.
Thats referring to PS1 right?
 
Please, Aku. No need to ride to the defense of your favorite console like a knight champion of old. It's just a hunk of plastic.

Also, enough with the self-martyrization, alright? Saturn ultimately did fail because it was overly expensive to build (and thus sell), whilst underperforming* versus the cheaper PS. There's absolutely no disputing that; demonizing "western media" (lol!) changes nothing, because this is a well-established fact easily backed up by other facts.

*Both in raw hardware power and ease for programmers to tap said power.

Grall apparently you missed the memo that Sega actually cost reduced and made a major mainboard revision which reached retail Japan in March/April 1996 at ¥19,900 therefore your assertions are wrong and misinformed.

This information can be found in Western gamer print magazines of the time...I'll double check and add the precise issue amd page for reference.

Also this major revision of Sega Saturn by Sega was a direct result not just of well over a year of production and engineering advances...but the fact that even by March 1996, the Sega Saturn was outselling the Sony PlayStation in the domestic Japan market...

This also preceded the rather paranoid price cuts made by Sony Computer Entertainment America Vice-president (a common theme ;) ) at E3 1996 of grand standing the $199.99 price as a strategic hype move not just aimed at Sega Saturn but also aimed at putting pressure on Nintendo's then upcoming Nintendo 64.

And yet despite the SCEA VPs grand standing, they're heads rolled when SCEJ got wind of this price loss despite their North American region advantage so please let's stop giving credit to Sony PlayStation without analysis of what was going on back then.

As for "demonizing western gamer media" there is actually a lot of hard evidence forever sealed in print from that time.

Evidence which begs the question of where they got their sources for quotes.

I know I initially only had access to the North American sold and thanks my launch PlayStation having free issues and subscription deal offers of "Next Generation magazine" which even came with inserts for "PlayStation secrets, tips and codes" along with the subscription tag line of the "Ultimate or best number one source for PlayStation" something which is laughably questioned rather suspiciously by readers both before and after the September Sony PlayStation launch...

It's amazing how "marketing" can insult engineering efforts with misinformation.
 
So on the one hand we have people who at least have some understanding of games programming and old console hardware looking at Saturn and saying that it's going to be harder to develop for than PS1. On the other hand, we have you saying they're both easy to develop, with it being really unclear what you're actually basing this on. Because you had both consoles and were comparing the games, apparently?

You keep saying we've been misled by a negative western media when that has nothing to do with that. The difficulty statement is based on things like:

- Having to use the second CPU to get reasonable performance. That means having to worry about concurrency and synchronization on processors that don't even have atomics and locking operations. You also have to manage flushing the cache because they're not cache coherent. These are huge pitfalls for programmers, they're not easy problems. And then trying to actually divide up tasks to get a good work balance on both cores is going to be very hard unless the game has a design that lends itself well to a simplistic partitioning.

- Having to do all the geometry stuff in software, in CPUs that don't have proper divides. The lighting calculations in particular are a lot of work to get right and efficient, which is probably why games don't even bother with it. But even without that the games have to calculate camera transformation, perspective division, frustum culling, backface culling, and depth binning while the GTE on PS1 while either perform these operations or at least compute flags to help you handle them. And god help you if you have to clip.

- Pure quads are just plain harder to mesh decently, making life a lot worse for the artists.

- There's no decompression hardware like the MDEC on PS1 (unless you count the add-on MPEG card), so it has to be done in software. You could use third party libraries like Cinepak, but Cinepak sucked...

- VDP2 is super convoluted. You have to actually set registers that program how the VRAMs are accessed every cycle, and depend on this being consistent with the other settings you've programmed (some games messed this up anyway). There's all sorts of one off features that are not that useful, and some extra design complications that don't really make sense like how the maps are partitioned hierarchically. The rotation layers are not at all intuitive to set up. And the documentation is very confusing, or at least the English documentation is. On PS1 there's less flexibility and raw power with 2D but it's more straightforward to manage, even if you have to turn maps into display lists yourself. You could do 2D games fully on VDP1 but it'd be more limited than PS1 for sure...

- You have to manage the interaction between VDP1 and VDP2, meaning you have to worry about how priority is encoded. And if you want to do alpha blending that isn't broken it has to be on VDP2, but that's very limited and will likely involve having to draw occlusion geometry in VDP1 to make parts of VDP2 layers visible, at least if you want it in moderately complex scenes. Speaking of which, Sonic Jam's use of VDP2 probably involved occlusion geometry as well.

- You need code running on the 68k to babysit the audio hardware, which is really generally not necessary (managing some voice keys is not a demanding task). I'm guessing Sega probably offered libraries for all the audio stuff, though.

- The SCU DSP is really, really hard to code for, which is probably why few games use it for much of anything. You have to think a lot about how operations happen implicitly, in parallel, and in pipeline, and you have very limited memory to do it all in.

Now sure, PS1 has some areas where it's probably harder to program for. You have to manage a small explicit scratchpad in the CPU instead of a transparent unified cache, you have to be careful with how your code is organized to avoid conflict misses in the icache, and you have to be careful with textures to avoid thrashing the texture cache (generally probably means avoiding > 64 wide @ 4bpp textures per triangle if it's in real 3D stuff). But the laundry list for Saturn easily outweighs this.

And it might be pretty easy to program a simple 16-bit style 2D game on Saturn, where you only need to use one CPU, have no FMVs or anything, no reason to use the DSP, and only needing the VDP1 and VDP2 in a really limited straightforward fashion. But is that really what we want to call a baseline for the system?

Your entire post is all about selling the Sony PlayStation as a preferred platform to develop games on.

It's also based on early development tools and to be fair, as I have indicated, the comparison only makes sense from 1994 to 1997 and barely 1998 as those were the years of heavy first party software development on Sega Saturn and Sony PlayStation in comparison and 1997-98 gets derailed by the circumstances of that time.

Making videogames in that era was NOT about "software port parity" so please stop this...it was about selling as many units via proper marketing and killer titles...something that was a massive failure falling square on the responsibility of Sega of America's antics in 1994 and leading into 1995-98.

Sega of America's 32X plays an immense role in derailing not just marketing budgets to sell/promote a platform but also software development funding costs and efforts wasted on a hair-brained add on that was the true source of creating all that "consumer distrust" and "retailer anger" again all forever recorded in contemporary print magazines of the time as news reports, angry letters in letters sections and incompetent interviews where current "heroic firgures" sound like grimy used car salemen.

There's also game developer interviews which do dismiss that notion you sell and repeat from 1995 that Sega Saturn is hard to program for...when you are making a Sega Saturn game in that time your objective is to target that hardware and usually you would be doing this using your experience in assembler code...

Yes the SCU DSP Geometry processor in Sega Saturn did have a rather difficult reputation...but then you had Sonic Team and Team Andromeda and a few others debunking myths on those claims...and in such a short time when you realize that it's barely three years of Sega Saturn in retail life.

As it stands it is rather curious how many 3d polygon Quadrilaterals based distorted sprite graphics games existed as having been shipped in retail in 1995 Japan.

(The question of how many of these titles were later censored or not localized until a whole year later despite being first/second party Sega contract properties...which means they were Sega Saturn exclusive software. That question falls into the incompetent Sega of America management of 1995-96...and rather arrogant management at that.)

You are an expert coder right? You are quite aware that games development isn't a static progress implying no more advancements can be made correct?

It takes game sequels and profits from sales to make this happen...and gamers back then were not really going like "ZOMG, Sega Saturn is hard to program for....I'll just buy a cheaper PlayStation because it's easier" because this is actually marketing and had nothing to do with the actual software.

Despite the Sega internal CS Team being made to waste their time on 32X Virtua Racing Deluxe, that same development team is responsible for porting the rushed Sega Rally Championship 1995...released in December 1995...

Other software was wasted on 32X that could have proven to be helpful IP if they had been funded for Sega Saturn...like Star Wars Arcade...even that crappy Knuckles Chaotix.

But even back then you had Sega of America's SegaSoft making platform agnostic announcements and all these Sega of America development teams NOT making the games they were supposed to be working on...instead we hear tales of wasting time and money on Sonic X-Treme because apparently Sega of America was not only incompetent and greedy...but was still out of control in their focus to support the PC platform with ports or one game instead of making software for the flagship platform...

This is a major difference and polar opposite to Sony PlayStation where SCEA was NOT wasting time and money funding PC ports...because the focus was PlayStation not a rival platform.

I'll just focus on this despite the time but my argument still stands...the Sega Saturn wasn't as hard to make games on as you make it out to be even back then and even acting as a salesman or spokesman for Sony PlayStation marketing selling points.

Sony PlayStation was so damned easy to dev for that it took Namco from 1994 to 1997 to make three Ridge Racer games and three 3d polygon fighting games...with Soul Blade running at 30fps and Tekken 2 being a blocky as hell, lacking animations of character faces and hands...but some really great marketing sauce to mask the sub par gameplay mechanics and technical hyperbole inferiority complex.
 
Even though it was in 2016 postings...I'm getting a guy selling me a "Sony PlayStation as best platform" in a Sega Saturn discussion thread...when I actually purchased the system at launch in 1995.

Talk about marketing hype out of control in a technical discussion.
 
I might be the biggest old school Sega fan here, but it is not Sony's fault that Sega's management totally screwed up its console hardware planning (32X), design (Saturn), marketing (Saturn and Dreamcast) and software planning (no good soccer games on Dreamcast, no good Sonic on Saturn etc.). Sega fucked up all on its own while making some brilliant games. I think tonight I'll play some Panzer Zwei, drink whiskey and cry a bit.
 
Also this major revision of Sega Saturn by Sega was a direct result not just of well over a year of production and engineering advances...but the fact that even by March 1996, the Sega Saturn was outselling the Sony PlayStation in the domestic Japan market...
if I recall Sega was deliberately using numbers sold to retailers instead to consumers
 
I think Saturn actually is more powerful, the tools, such as ready C compiler just wasn't there or wasn't ready at launch.
You had to juggle seven distinct processors and several pools of RAM to get the most out of it. Playstation by comparison had just three and a half (GTE is on-die and is handled as sort of a co-processor) and less RAM but only three pools of it.
 
Last edited:
Back
Top