An all-out Bleemcast!-type project would be interesting.Quote:
Originally Posted by azonicrider
Printable View
An all-out Bleemcast!-type project would be interesting.Quote:
Originally Posted by azonicrider
I hate fetch questing, but I definitely support the idea of an original effort rather than just porting what was essentially a Playstation version of Rare platformers. If that nutty Sonic fanbase manages to crack Sonic Adventure I will probably play whatever hacks they come up with until I get sick of them. There has to be a third person platforming game with open source by now, even if it is a PC game porting that to the Dreamcast should be relatively simple. Replacing textures and improving models probably isn't that simple though.Quote:
Originally Posted by azonicrider
Also, forget about Bleemcast, although I think all of its various iterations have been leaked and can be tweaked to play any PS1 game. I want a Saturn emulator for the Dreamcast. It should be relatively simple to do as a high level emulator with the SH4's backward compatibility. But, since the PC emulators can't seem to shape up I guess I am just wishing it would be so.
Well the Saturn architecture is very hard to emulate, they had just under 400 processors in the Saturn.
I wanna accomplish what Spyro did for the PS1, with the Dreamcast. Spyro is probably the best looking game on the PS1.
With 64 audio channels, 8 MB VRAM with 5:1 texture compression, the SuperH-4, we could accomplish a game that makes the PS2 look a generation behind. ;)
Saturn was also programmed largely at low-level . . . and those with API use probably still add low-level tweaks. Between the heavy API use on the PSX (both 1st and 3rd party) and the relatively "standard" 3D feature set even at low-level, the PSX is far easier to emulate, same for N64 and pretty much all later consoles save PS2. :pQuote:
Originally Posted by azonicrider
3DO is also easier due to the forced API programming, even though the architecture isn't "typical" 3D. (that and it's still a lot simpler than Saturn, though low-level documentation is tougher to come by in general for the 3do, so low-level emulation would be tougher anyway)
Jaguar is painful to emulate though, not just the relatively complex low-level architecture, but also the smattering of odd bugs and the programming methods many developers used to work around those. (in many cases never actually finding/documenting the bugs/errors in the hardware and, instead just using trial and error to work around problematic code until it "just worked" . . . which means a lot of code that only works properly on real hardware with weird timing errors avoiding register or memory conflicts that SHOULD occur based on actual documenation -and do occur in emulators based on that- ;))
I remember looking at assembly for the SuperH 4, it wasn't that bad. I guess you could go low-level if you really want high performance.
It's not hard, it just takes a LOT of time to add in everything.Quote:
Originally Posted by azonicrider
On analog sticks:
DC's becomes squeaky over time. Have to clean the pad more often than the others of that gen. GC's is pretty "tight", in a good way, but the octagonal slot makes it a pain for some games, and, of course, that C-nipple was idiotic (still better than DC's "nothing", though). The texture grip on the GC analog stick was the best of that gen, though; DC's would've been right up there, but it wears out, becoming too smooth/slippery. PS2's analogs were convex sans a good grip as found on GC and DC, and PS2 had that silly symmetrical layout that Sony should've junked (in addition to the "d-buttons" layout). I think XBox Controller S is the best pad, pound for pound, of that gen. Good rubberized grips on the analog sticks, asymmetrical layout, triggers, etc.
On the idea of 3D platformer game homebrew on DC:
I'd rather a good 2.5D platformer. Maybe have some special stages, ala Sonic, in 3D, but the bulk of the game would be 3D polygonal characters, objects, etc. with gameplay on a 2D plane. Exploration can be handled similarly to how it is in those 2D Sonic games and Metroid for that matter (secret "breakable" hidden areas) and/or like in the Mario and Wonder Boy/Monster World series (going into and out of doors, etc.). Yeah...those would work pretty well in a 2.5D platformer. It would also be easier to implement overall, and not have the issue of how to implement camera control well.
The DC stick's texture also tends to wear down and become slick . . . at least if it's used really heavily/aggressively. (ie similar circumstances that would make it "squeaky" except it's irreparable)
As for texture/grip/feel, I'm rather partial to the Xbox Duke's analog sticks (I think I prefer the more convex right stick too . . . actually I swapped them and I think I like it better as the left stick). Of course, the duke has the problem of less than ideal face buttons (for mashing at least) and being too big for small hands. (OTOH the other controllers have the opposite problem, and aren't nearly as comfortable size-wise . . . except the GC pad oddly enough -one of the few small pads that's still pretty damn confortable, unlike the -technically bigger- Xbox-S or Sony's standard form factor -which I really don't care for, though the old dual analog pad was a short-lived improvement :p )
Well I got alot of time to brainstorm ideas, 2.5D, 3D.Quote:
Originally Posted by spiffyone
I need to accomplish more work on PS Vita shooter before looking into this. Which is going slow, because computer can't support OpenGL 3.
The Dreamcast sure is a weak piece of crap.
Take a look at 5:32 here:
https://www.youtube.com/watch?v=WMlsQELCrmA
3 million is nowhere near 60 million.
Thats like a 1:20 ratio.
No it isn't, 60 million is also ten times the PS2's maximum in game performance.Quote:
Originally Posted by azonicrider
Not sure about actual game examples, but in terms of raw 3D vertex throughput, the PS2 VUs are slightly less than 4x as fast as the SH4 VU, though Sony's quoted polygon rates (not countng the strip/fan-specific ones) are still proportionally higher than that difference though (66-80 million/s) . . . unless you go be the animated 3D model "effects" polygon figures (15-20 million) which are actually less than 4x the DC's quoted 8 million polygons/s.Quote:
Originally Posted by sheath
Then there's the VU-intependent demos using streaming vertex data that show really high polygon rates (showing peak GPU performance) . . . though with both that and the raw VU performance in mind, I'd have thought some demos actually showing things close to the theoretical peak would exist too. And yeah, you've got added overhead for T&L on the VUs on top of geometry (and possibly physics), but same deal on the DC anyway (for CPU burden), and actual DC games still hit better than 1/3 the theoretical peak polygon rates. (which, porportionally speaking, would imply around 12 million polygons/s on the PS2 under similar conditions that DC games pushed 3 million)
Its just more evidence of Sony successfully brainwashing people.
That video will continue to promote the untrue specs.
But it makes me wonder, will we ever figure out how much the PS2 can actually do?
Like okay, Test Drive Le Mans did almost 5 million per second
I hear people say Jak And Daxter does 9 million per second, same with Gran Turismo 4
Or maybe we'll just never find out, because of the PS2's architecture being difficult to harness.
I was actually referring to some facts brought up earlier in this thread. Such as Sony's "How Far Have We Got" document from 2003. Check out page 13, 52,000 polygons per frame at 30FPS at best is 1,560,000 polygons per second. That is the average PS2 game as of 2003. Resident Evil 4 in 2005 was running at 900k polygons per second with none of the lighting effects and texture quality of the Gamecube original.
Also, earlier we reviewed some sources that pretty much spelled it out that the PS2's realistic in game peak was below 5 million polygons per second, although a theoretical model done in another study by EA showed that in a vacuum the system actually could render close to 18 million per second. That number was attained by rendering a high poly model once and taking the milliseconds of that single render and assuming the system could maintain that in a sustained in game environment. No similar approach has surfaced so far for the Dreamcast's performance, and all Dreamcast polygon per second figures are front facing visible on screen only as everything else is culled by the tile based renderer in the PVR2-DC.
The point I got was that was the marketing Sony was using, not that those specs had anything to do with the PS2 actually putting out games with graphics head and shoulders above the DC. :pQuote:
Originally Posted by azonicrider
Albeit the whole argument of technical inferiority is flawed, even if that's the case, or more importantly, perceived technical inferiority. Nintendo had huge marketshare and PR, and the SNES was marketed (and considered) technically superior by much of the early 90s market, but Sega had the advantages in certain software areas, consistently competitive marketing, and pricing. (all of which the DC had too, save consistent marketing . . . plus actual funding to keep pace with the market by 2000 -had the Saturn been closer to SMS level overall success -including financially- by late 90s market standards, they very well could have been in a position against Sony in 2000 not so unlike that against NIntendo in 1991, again at least relative to the market standards of the time . . . except with Sony's advantages, it's a bit more like if you combined the best strengths of Nintendo and NEC in 1991 -except even then Sega didn't have to deal with 2 other major competitors as the DC would have in 2001, though given how much Nintendo and MS struggled that generation, a better off Sega might have done quite well with the DC . . . not to mention the typicla 5-year console turnaround could have meant Sega's successor arriving by 2003 in Japan and 2004 in the west)
Quote:
But it makes me wonder, will we ever figure out how much the PS2 can actually do?
Like okay, Test Drive Le Mans did almost 5 million per second
I hear people say Jak And Daxter does 9 million per second, same with Gran Turismo 4
Or maybe we'll just never find out, because of the PS2's architecture being difficult to harness.
Really, with raw tech specs like that, it's still not wrong, or even lying, if it's actually based on facts. The PS1's 1 million vertices/s figure is totally legit too, it's the peak vertex throughput for the GTE . . . you aren't going to see actual games pushing that, of course, but it's still significant. (as much as ~450,000 verteces/s for an SH2 is accurate)Quote:
Originally Posted by sheath
Actually the big thing I take issue with in that advertized specs comparison was that the DC wasn't specced at 3 million polygons/s, but 8.
On a separate note, the GameCube's GPU is only specced for 60.75 M vertices/s, and the Xbox's 116.5 M . . . but compared to the PS2's theoretical peak, those (and the DC) manage much closer to that in reality. (even if you go by the lowest 15-20 M polygons/s spec)