PS2 vs Dreamcast Graphics

[stu]

This thread has established DC lacked raw power next to PS2 but still was much better in textures and the CPU I think carried more instructions per second than PS2s. Only Gamecube and Xbox beat it from what I remember.

Dreamcast has advantages in texturing and overall IQ thanks to a non borked Anti-aliasing system and supported a 640x480 resolution.

The PS2 Emotion Engine can calculate 6.4 billion floating point operations/sec
Dreamcast SH4 can calculate 1.4 billion floating point operations/sec.

Avantages/disavantages are irrelevant when one company can't finance a console fully thanks to past mistakes.

[Cloofoofoo]

Was hard to know for sure through text. Plus I’ve seen people actually imply stuff like that before.

Besides, weren’t the PS4 and Xbone way more than 3 times as powerful as the Wii U? The RAM on both systems were more than 3 times as much, and the CPU they had were also far more powerful. (Despite being x86 vs Wii U’s Power PC)
It’s really no debate at all that the PS4 and Xbox One were way more powerful than the Wii U, and in Nintendo’s defence, I don’t think the Wii U was supposed to compete on power, but rather recreate the Wii’s success.

I think with the Dreamcast vs PS2 it’s not quite so obvious. Especially since we never saw the Dreamcast truly pushed to it’s limit like we did the PS2.
I mean obviously the PS2 is going to be more powerful (Since the system came out 2 years after and cost 100 dollars more at launch, you’d hope so!), I’m just saying the gap doesn’t seem as big.



Oh? What do you mean?

It's the way dc is arranged , makes things like depth of field , bloom hard to do. Basically Dreamcast access to vram is slow , storing copies of rendered to texture that you plan to work on wouldn't be wise so they usually store in the main ram and either dma or sq copy to the vram. While faster that doesn't mean it's a speed demon. There's other things like fillrate that gets eaten up from multiple renders or bad performance from multiple large overlayed full screen polygons. This is a simplification as there might other issues( framebuffer it self might be an issue since I think you can't use the back buffer directly and have to copy it to main ram to manipulate it) . All this overhead piles up as you can imagine. It's even evident when you look at ps1 games that used things like cloaking , motion blur , radial blur , full screen warping/wave. Even ps1 games that did depth of field ( 2 definitely) . Heck someone I know , that was a hardcore soul calibur 1 player brought to my attention that soul calibur 1 arcade ( suped up ps1 that ran tekken) had a time released refraction mode instead of metal mode. So think about that for a second , the dc being magnitudes faster actually had to settle for a simple metal environment map mode for the costume instead of the framebuffer refraction effect of the ps1-esque the arcade machine did years prior.
It really doesn't seem a question of knowing how to do these effects but the struggle of doing it effectively on hw that doesn't mesh well with them.

Refraction ( alundra 2, threads of fate , soul calibur 1 ( system12), bassrise:






Depth of field ( terracon)

[Cobradile]

Ah I see, that’s really interesting!

Well, either way, you clearly know way more than me on this subject.
I only know the basics, and haven’t really dug into it.
Just curious, though, how does the PS2 square up to the Gamecube and Xbox? I heard those 2 systems absolutely blow the PS2 out of the water.

It's the way dc is arranged , makes things like depth of field , bloom hard to do. Basically Dreamcast access to vram is slow , storing copies of rendered to texture that you plan to work on wouldn't be wise so they usually store in the main ram and either dma or sq copy to the vram. While faster that doesn't mean it's a speed demon. There's other things like fillrate that gets eaten up from multiple renders or bad performance from multiple large overlayed full screen polygons. This is a simplification as there might other issues( framebuffer it self might be an issue since I think you can't use the back buffer directly and have to copy it to main ram to manipulate it) . All this overhead piles up as you can imagine. It's even evident when you look at ps1 games that used things like cloaking , motion blur , radial blur , full screen warping/wave. Even ps1 games that did depth of field ( 2 definitely) . Heck someone I know , that was a hardcore soul calibur 1 player brought to my attention that soul calibur 1 ( super up ps1 that ran tekken) had a time released refraction mode instead of metal mode. So think about that for a second , the dc being magnitudes faster actually had to settle for a simple metal environment map mode for the costume instead of the framebuffer refraction effect of the ps1-esque the arcade machine did years prior.
It really doesn't seem a question of knowing how to do these effects but the struggle of doing it effectively on hw that doesn't mesh well with them

That reminds me of how I heard the Sega Saturn handles transparency. Apparently very badly. It can do them, but doing so is very intensive for the CPU, so most ports settle with a dot style grid or a rainbow effect.

Sorry if I’m going too off topic.

[Leynos]

There is a cool part in Panzer Dragoon II Zwei where you see enemies underwater. The water has a sheen to it as well. It's a really cool effect for the system and era.