Could Dreamcast et al handle this/that game/effect? *DC tech retrospective *spawn

No one is saying that it will surpass the PS2 version, lol no!
That's exactly what SonicFan is saying.
After seeing this, his statement was absolutely right! Hell, if they'd done this at the time, i think DC would have looked and performed better than ps2!

Picture quality, color, textures, draw distance, and performance. This is already showing great performance with a much higher draw distance (and actually more geometry) than ps2 at this early stage. It's not even using the cpu yet.

I remember people saying here that it was impossible to run the game on this console....
It'd be worth identifying expected limitations that are being disproven. I'd like to say the streaming wasn't an issue, but I guess this is not streaming from optical drive - they aren't burning a ROM with every build. Or am I wrong on that?
 
I have had to tweak DirectX games' and benchmarks' PowerVR settings specifically to get the Neon 250 to stop dropping polys all over the scene. Sometimes it's transparency, and so enabling Force HW Transparency fixes it. Other times I think it is multitexturing formating, but "disable render until flip" fixes it, as was the case with Soul Reaver 2.

Either way, the "flickering" polys or simply unrendered polys are still in VRAM and still being calculated for. It's just a quirk or the tile based rendered that needs to be accounted for in each game past 1999.
 
Btw, I am pretty sure the game will end up being "below" the PS2 version but it went from impossible to achieve to a playable version on a platform that it was supposed to be incapable of running it is a huge achievement.

I think people need to be reminded of previous discussions.

It has been stated that DC couldn't handle the PS2 version, exactly as it is on PS2 (And you even say you're sure it will end up below the PS2 version)

Everyone has agreed that it could do a 3D version of GTA3, which is what it's doing.

But I wouldn't call it playable, I would call it functional as the game somewhat runs and functions.
 
It could. Let's wait and see
Well I mean those artifacts we are seeing are because there is no front clipping, that is the reason why in my footage once the car is off screen you still see part of it's vertices exploding and being rendering all over the place so once it is implemented the console won't be rendering them and performance should improve, but just like you said let's wait and see....
 
I have had to tweak DirectX games' and benchmarks' PowerVR settings specifically to get the Neon 250 to stop dropping polys all over the scene. Sometimes it's transparency, and so enabling Force HW Transparency fixes it. Other times I think it is multitexturing formating, but "disable render until flip" fixes it, as was the case with Soul Reaver 2.

Either way, the "flickering" polys or simply unrendered polys are still in VRAM and still being calculated for. It's just a quirk or the tile based rendered that needs to be accounted for in each game past 1999.
Answer is even simpler. Here's what the two developers ( skmp and Falco) said . Though that flickering poly thing isn't a quirk, on the dc the tiles seem to have a designated memory amount that can overflow. You could set it higher and it prevents that. Another issue is I think degenerate triangles causing white lines or some crap, small triangle culling fixes. It's just clipping for this port
 

Attachments

  • Screenshot_20240808-105839.png
    Screenshot_20240808-105839.png
    192.7 KB · Views: 15
Lack of front plane clipping maybe doesn't imply something in the realm of "disable render until flip." I'm not entirely sure what this setting does to the renderer either way.

I also just checked the settings and I had zero tweaks enabled for the GTA 3 Neon 250 video. I've mostly only needed them for DX 8 games, before "hardware T&L" became mandatory.

Also, if GTA 3 runs, so will Vice City.
Screenshot_20240808-094226.png
 
Last edited:
I agree with Shifty, we din't know how well GTA 3 was optimised for ps2. I think optimisation was very poor. VC looked way better and SA even better than VC. Of course partialy that was because devs knew PS2 hardware better and better, but I think also because next two games were better optimised. GTA 3 on PS2 also had low fps.
But is someone says what DC was as capable as ps2 or even more capable that is just out of sense. PS2 had 50% higher clocked cpu, VUs with far more power than DC cpu, (~4 times more), 50% higher clocked and more capable gpu, (especially multipass), and also with edram, and had more ram.
 
Of course partialy that was because devs knew PS2 hardware better and better, but I think also because next two games were better optimised.
Two sides of the same coin. With experience and time, games could improve on PS2 utilisation, not just from a developer's own experience but also best-practices over the years. The point of this thread is to try to understand what DC was capable of where it didn't get that accumulated experience. The GTA port looks set to show DC could certainly handle a game it never got. What that doesn't tell us is if DC is overall more powerful than PS2. Which this thread is not about and those sorts of statements and comparisons aren't useful technical conversation.
 
There is a bit of nostalgia and nonsensical old school devotion to fallen giants where people still keep hoping that there was some kind of untapped secret sauce that would have allowed the console to perform at levels, beyond logic.

GTA3 is certainly an interesting case that may actually demonstrate DC' s likely untapped potential since the console died too early. But this is one thing, believing it will surpass it's competitor for retribution or some kind of late catharsis is another.
 
I think this tech demo playing "find my own way" is one I haven't seen before. But it exemplifies the differences/shift in 3D graphics that generation had. When Sony claimed the PS2 could do 30+ million polygons per second, or 75 million, or the gaming press and the majority of online gamers accepted "10X more powerful than any console out there" ad hoc, it made me look. I was used to focusing on gameplay mechanics, specifically unique gameplay. But I shifted to framerate which, on a standard definition TV, meant silky smooth when the screen panned quickly was 60FPS, blurry was 30FPS, and stutter meant below 30PS. I'd note this, then I'd wonder why it mattered if the gameplay wasn't broken by either/or.

When I focused on character models I looked for hard edges, deformed claws instead of fingers on hands, moving mouths and eyeballs, ears that actually had insides and outsides. For other detail, I looked at textures and expected sharp real life detail of rocks, or grass, or sand, or grains of cloth. What I got was something else entirely. I got motion blur instead of fine detail on walls and characters. I got sub 30FPS more often than not, but there were sparklers and splashes, and streaks of light on the screen that simply did not impress me.

Over the decades, gamers moved on and stopped talking about model complexity and texture detail in PS2 games, instead focusing on the particles and "post processing" effects as the true proof of its "polygon monster" status. But that is not what I was looking for when I bought my PS2 for Soul Reaver 2 in 2001. This was not what I was looking for when I bought Kendo Master of Bushido, or Silpheed, or Gun Griffon Blaze, and failed to see even double the detail on screen of "anything else out there." Later on, Tenchu, Shinobi, Virtua Fighter 4, Blood Will Tell and even the popular mainstream games continued to perplex me.

More sparks in Burnout didn't take away the obviously lower resolution, better modeled cars in Gran Turismo 3 A-Spec didn't distract me from the obviously low color/resolution, maybe palatalized, textures especially when compared with F355 Challenge. When I played ports of Dreamcast games I didn't get twice the framerate, or twice the character model detail. I didn't even get more effects, in most cases I got less. So, for the entire generation and beyond I thought it was pretty obvious that whatever was "10X more powerful than anything else out there" was somehow hidden in the background, didn't really affect the graphics. Maybe it was physics, maybe it was lighting or blur effects I didn't care about, but it definitely wasn't the polygons per second/frame promised clearly in the hype and accepted by the mainstream.

I think that, and the near religious belief that the PS2 was the most powerful game hardware of its time, caused a thread like this to go on so long. I don't think anybody believes the 1998 hardware in the Dreamcast could be more powerful all around than the PS2, the problem is the majority opinion for decades is the PS2 was all around more powerful when the games, especially prior to 2004, didn't clearly demonstrate this "fact."

To the point of this topic, based on what I've played on the Neon 250, the Dreamcast could handle any of the games up to and including 2003 titles. There would have been adaptations, concessions, maybe a RAM upgrade would have come along, but the hardware was more than capable of these games. That's a five year life span from the Japanese launch, not too bad for a company that had its mainstay Arcade business swept out from under it while it was designing the hardware.
 
Last edited:
Wouldn’t emulators confirm for example tekken tag tournament on ps2 having 3-5 times the polygon count compared to whatever dreadcast game supposedly comes closest? (Where you can pause and see technical details)
 
Wouldn’t emulators confirm for example tekken tag tournament on ps2 having 3-5 times the polygon count compared to whatever dreadcast game supposedly comes closest? (Where you can pause and see technical details)
Lol Tekken tag does not have 3-5 more polys than the closest Dreamcast equivalent, in this case Dead or Alive 2, actually, at least in terms of polys for characters it is below it, some stages are higher though.
 
I think this tech demo playing "find my own way" is one I haven't seen before. But it exemplifies the differences/shift in 3D graphics that generation had. When Sony claimed the PS2 could do 30+ million polygons per second, or 75 million, or the gaming press and the majority of online gamers accepted "10X more powerful than any console out there" ad hoc, it made me look. I was used to focusing on gameplay mechanics, specifically unique gameplay. But I shifted to framerate which, on a standard definition TV, meant silky smooth when the screen panned quickly was 60FPS, blurry was 30FPS, and stutter meant below 30PS. I'd note this, then I'd wonder why it mattered if the gameplay wasn't broken by either/or.

When I focused on character models I looked for hard edges, deformed claws instead of fingers on hands, moving mouths and eyeballs, ears that actually had insides and outsides. For other detail, I looked at textures and expected sharp real life detail of rocks, or grass, or sand, or grains of cloth. What I got was something else entirely. I got motion blur instead of fine detail on walls and characters. I got sub 30FPS more often than not, but there were sparklers and splashes, and streaks of light on the screen that simply did not impress me.

Over the decades, gamers moved on and stopped talking about model complexity and texture detail in PS2 games, instead focusing on the particles and "post processing" effects as the true proof of its "polygon monster" status. But that is not what I was looking for when I bought my PS2 for Soul Reaver 2 in 2001. This was not what I was looking for when I bought Kendo Master of Bushido, or Silpheed, or Gun Griffon Blaze, and failed to see even double the detail on screen of "anything else out there." Later on, Tenchu, Shinobi, Virtua Fighter 4, Blood Will Tell and even the popular mainstream games continued to perplex me.

More sparks in Burnout didn't take away the obviously lower resolution, better modeled cars in Gran Turismo 3 A-Spec didn't distract me from the obviously low color/resolution, maybe palatalized, textures especially when compared with F355 Challenge. When I played ports of Dreamcast games I didn't get twice the framerate, or twice the character model detail. I didn't even get more effects, in most cases I got less. So, for the entire generation and beyond I thought it was pretty obvious that whatever was "10X more powerful than anything else out there" was somehow hidden in the background, didn't really affect the graphics. Maybe it was physics, maybe it was lighting or blur effects I didn't care about, but it definitely wasn't the polygons per second/frame promised clearly in the hype and accepted by the mainstream.

I think that, and the near religious belief that the PS2 was the most powerful game hardware of its time, caused a thread like this to go on so long. I don't think anybody believes the 1998 hardware in the Dreamcast could be more powerful all around than the PS2, the problem is the majority opinion for decades is the PS2 was all around more powerful when the games, especially prior to 2004, didn't clearly demonstrate this "fact."

To the point of this topic, based on what I've played on the Neon 250, the Dreamcast could handle any of the games up to and including 2003 titles. There would have been adaptations, concessions, maybe a RAM upgrade would have come along, but the hardware was more than capable of these games. That's a five year life span from the Japanese launch, not too bad for a company that had its mainstay Arcade business swept out from under it while it was designing the hardware.

Many of these numbers were a misinterpretation or they represented theoretical maximum polygons outside of actual gameplay conditions or a fully rendered image (i.e just rendering polygons and nothing else).

I believe Jak and Daxter was around the 15 million polygons per second mark or at least that's what the sources suggested, but don't really know.

Then there is the fact that the PS2 was redrawing geometry multiple times just to achieve some effects (multipass). It looks like the PS2 was particularly designed to draw so many polygons for that purpose. So when the theoretical specifications may say X amount of polygons, it doesn't mean the same as looking at X amount of polygons/geometrical complexity on screen even though the calculations of X amount of polygons are actually done to render the scene with all it's effects.

Unsure if 15 million polygons per second on Jak and Daxter is referring to the actual scene or with multipass included btw.

But I remember discussions about the DC's capabilities suggesting around the 4 million mark.

The PS2 was certainly not a straightforward architecture. Games ported to it struggled or were just ok (i.e Max Payne, MDK, Half Life, Unreal Tournament, Quake etc) . Games designed for the PS2 on the other hand performed and looked like nothing else.

For example while PS2 may have run Headhunter at 30fps just like the DC, Metal Gear Solid 2 was running at 60fps while looking 10 times better.

While Test Drive Le Mans on the DC was considered an impressive achievement at 30fps, GT3 was running at 60fps and looked significantly more impressive. Same for Ridge Racer V running at rock solid 60fps and that was a launch game. The most impressive racing game on the DC was Metropolis Street Racer and it run at 30fps. I m sure that if MSR was ported to the PS2 it would not have run at 60fps and might have even received downgrades.
 
Last edited:
10x more powerful on paper never translates to 10x better visuals on your display.

PS2 showed it superiority over DC after only 12 months (GT3, MGS2, DMC) all better looking titles than anything DC had at the time and we're all 60fps too.

They're not 10x better looking than DC games but they're a very noticeable step and up and actually felt 'next gen' then games on DC.

The only 'next gen' wow moment I ever got on DC was playing Code Veronica for the first time.
 
cars in GT3 were around 4000-5000 polygons if i remember correctly.
At the time i was part of the PS2 online beta test, and we tested WRC3, on the forums a dev said their cars were around 30 000 polygons, but of course there was only one car on screen at a time, and i don't remember if it was 30 or 60fps.
 
Back
Top