lethal_illuminati
Ban Hammered
mayne, my thread done fell off and went off topic, All i ask was if they was any games that work like that dc game i saw...found it too nephews, it's called dodonpachi and it's aight.
Ikaruga
- Thread starter lethal_illuminati
- Start date
Well, both. The slowdown doesn't really hinder the gameplay that much though, in fact it makes the game easier. The arcade version also has slowdowns, just not quite to the same extent as the Saturn version. I hear the PS version is slightly closer to the arcade as well, although I'm not sure if it supports tate or not.
No... it isn't. But the effect in question, the motion blur, is a framebuffer effect and PC based graphics chips are notorious for really, really sucking at those. Xbox's UMA doesn't help, either.
For reference, the PS2 version not only never slows down, it even runs too fast most of the time (frame rate, not gameplay wise), sometimes flying into triple-digit FPS.
I may be a little late, but I think I may know why the Saturn uses quads instead of triangles. I believe VDP 1 is the descendent of the graphics co-processor in the Sega CD. The Sega CD graphics co-processor was designed to implement the kind of rotation and scaling abilities granted to the SNES on the Genesis; however, since there is no way for the co-processor to interface directly with the VDP, it renders its output to RAM, much like a modern GPU (with the exception that it renders things in 4-bit per pixel tiles). One of its functions was handling sprite rotation/scaling/warping and if you're trying to render sprites (which tend to be more square than triangular) and not render 3D environments it would be entirely logical to render quads.
I haven't looked into the functioning of the sprite mode yet, but I would not be surprised if it shared some similarities with VDP1 from the Saturn.
I haven't looked into the functioning of the sprite mode yet, but I would not be surprised if it shared some similarities with VDP1 from the Saturn.
AFAICT, it's a semicustom predecessor of the Hitachi Q2SD family. I can find no Sega patents relating to VDP1's design (though I did find a rather detailed patent for VDP1-VDP2 interaction as Pat. No. 5,848,201). I may have just missed it, though.
And in other news, the patent for the "upgrade" is apparently 5,946,004 if anyone wants to wade through the patent-ese to see what it would have been like.
not really, I find it hard to believe that the PS2 has made that big of a jump in graphics, but i don't see it as impossible, as how much the Saturn jumped in power in it's last moments.
well i disagree with you and your opinion on TechTV, thats your opinion, and no they weren't the ones who said GT4 maxes out the PS2 that was from Polyphony Digital directly, and they said thats the reason the next installment of the GT series won't be on the PS2, they said that they have pushed the system as far as possible and it's time to move on after GT4!
umm... i never said this myself the way you make it sound, the PS2 version is not arcade perfect, look more carefully, there are graphical flaws in the translation, and i never said VF4 arcade pushes 10 mil, thats just the max for NAOMI 2, VF 4 pushes 8+ mil.
I find the XBOX number hard to believe, i can see maybe going the extra mile maybe to 25 mil, but i'd have to see it to believe 30 mil, and yes i have seen silent hill 3 and no doubt it's nice, but i stand firm that the GC/XBOX kick PS2's ass in the graphics department, all of the nicer looking PS2 games have a sort of grainy look to them.
just zip it and post it, i bet i'm not the only one who'd like to see it!
Sony's official statement is 66mil, reduced from 75mil, and thats raw graphics with no effects or textures, SEGA's is about 12mil raw. However the 7mil i posted is SEGA's number for a working game not a tech demo and not raw graphics!
the DC can handle textures both more easily and more effciently than a PS2, although i can certainly see that some PS2 games can never be done perfectly on a DC like the Gran Turismos (also note that the PS2 suffers from anti-alising issue even in this amazing looking game), most i can see being ported without much hassle to the DC, and yes there are some DC games that would never be able to run on a PS2 because the texture demand from the game is too high, like Sonic Adventure 2, Phantasy Star Online, Shenmue II, Ikaruga, Border Down, etc.
Also something i'd like to address on a separate note, not directed at you is why so many 3rd party DC games look like crap... an official statement from SEGA, "less than 10% of all developers pushed the Dreamcast passed 1 million polygons" I don't believe the tech demo tapped into DC's full power and I don't know if we'll ever know what the DC can really do!
MGS2 = yes, SH3 = i'd have to get a better look at it, GT4 = no
and as for XBOX, the fault of MGS2 not running effciently is the developers fault, the only difficult thing about programming on the XBOX is, like the Saturn, Transparency's
Yes, have a nice day...
Oops missed that, still learning about the XBOX. but the transparecy issue is true according to AM2 on the port of Shenmue II to the XBOX.
The transparency problem AM2 complained about was due the fact the PowerVR2 uses per-pixel sorting for semi-transparent objects in the hardware. That allows it to properly interesect alpha-blended polygons.
If someone here is not aware of it, the Z-buffer principle is incompatible with semi-transparency. So in the PS2, GC, Xbox and every HW-accelerated PC 3D game, the engines usually draw the opaque stuff first, as usual (read/write the z-buffer), and then sort the semi-transparent triangles (or objects, depends on the implementation), back to front, in the CPU, and then tell the videocard to draw them, checking the z-buffer (so they'll intersect the opaque stuff), but not writing to it (to avoid parts of a transparent triangle from blocking another one).
So, in every (modern) system with exception of the DC, a semi-transparent polygon will always be drawn fully above or fully below another: they'll never intersect properly, because they aren't writing themselves to the z-buffer.
The PowerVR2 uses some different concepts. I don't remember the exact process, but the PVR2 handles the sorting by itself, automatically drawing the opaque stuff first, and it sorts the transparent triangles back to front per pixel, so a triangle can have part of it above, and part of it below another triangle, and the intersection will be flawlessly.
(Of course, this was one of the reasons why the DC could be brought to it's knees if you have tons of semi-transparent polygons on top of each other - Quake3 Arena's rockets)
The X-box video chip doesn't have such commodity, so AM2 had to modify their rendering code a bit more to handle the transparency, and they probably had to tesselate/split the trees'/bushes' faces to avoid interpenetrations.
If someone here is not aware of it, the Z-buffer principle is incompatible with semi-transparency. So in the PS2, GC, Xbox and every HW-accelerated PC 3D game, the engines usually draw the opaque stuff first, as usual (read/write the z-buffer), and then sort the semi-transparent triangles (or objects, depends on the implementation), back to front, in the CPU, and then tell the videocard to draw them, checking the z-buffer (so they'll intersect the opaque stuff), but not writing to it (to avoid parts of a transparent triangle from blocking another one).
So, in every (modern) system with exception of the DC, a semi-transparent polygon will always be drawn fully above or fully below another: they'll never intersect properly, because they aren't writing themselves to the z-buffer.
The PowerVR2 uses some different concepts. I don't remember the exact process, but the PVR2 handles the sorting by itself, automatically drawing the opaque stuff first, and it sorts the transparent triangles back to front per pixel, so a triangle can have part of it above, and part of it below another triangle, and the intersection will be flawlessly.
(Of course, this was one of the reasons why the DC could be brought to it's knees if you have tons of semi-transparent polygons on top of each other - Quake3 Arena's rockets)
The X-box video chip doesn't have such commodity, so AM2 had to modify their rendering code a bit more to handle the transparency, and they probably had to tesselate/split the trees'/bushes' faces to avoid interpenetrations.
Why do you find it hard to believe that a cluster of THREE processors, TWO OF WHICH are dedicated VECTOR PROCESSORS, running at 300MHz, can outpace a single, less efficient (if more developer friendly) 200MHz chip? *bows*
Keep in mind, optimising for multiple processors, as you have said, isn't exactly the easiest thing in the world, and PS2 shipped with one impressively shitty development environment. It took Sony more than a YEAR to get their act together and ship some libraries worth shit.
TechTV also sang massive praises of how the GeForce FX 5800 Ultra was the best graphics card ever created. Two months and a single production run later, it was silently cancelled. Because it sucked. A lot.
Polyphony Digital said they were satisfied with what they had achieved with GT4, and that in order to achieve their vision of GT5 it had to be PS3. Would you like to dig up the link or should I?
OK... so um... ok? I dunno, this review of VF4 doesn't complain about any losses from the arcade version, with the sole exception of pixillation which I'll cover later.
And regardless whether VF4 is 8+ Million or 10 million, the fact that it's at least 99% arcade perfect, AND a very mediocre game on the programming side (it was released pretty early, a LOT of new tricks and techniques exist now for PS2 development that are even publicly known).
Regardless whether it's 25 million or 30 million, it still overpowers your blind guesstimates.
Again, I'll cover the "grainy look" shortly.
The video is 27MB. I'll attach it to my next post, immediately following this one.
Couple things.
1. You said 8 mil before. Where'd that 1M go?
2. Do you really think ANY architecture could be 66% efficient with full effects? As in, go from zero lights and all same-sized triangles to variable sizes, irregular depth values, multiple texture layers, multiple light sources...
3. Once again. Tech demos are IN NO WAY indicative of actual game performance possibilities. Most tech demos run little to no AI, physics, collision detection... all of which drain CPU power tremendously in an actual game environment.
4. I don't remember ever reading 12 million raw anywhere... link please? I have never seen such a statistic.
I'll just be blunt here. There's no reason most of those wouldn't be possible on PS2.
PS2 DOES INDEED support texture compression, and it has that 1.2GB/sec very-low-latency bus from the Emotion Engine to the GS. One of Naughty Dog's programmers joked that they used to think that the GS's onboard 4MB was a horribly limited space for framebuffer and textures, now they think that filling 32MB (system RAM) is almost as hard! And just FYI, Dreamcast can only hold up to 8MB textures (before you create a frame buffer).
The Gran Turismos couldn't possibly be ported to DC. The OFFICIAL number for GT3's poly count is around 12M/sec... that's already right on the edge of YOUR OWN "DC theoretical max" count.
What meaning does that have to this discussion? Seriously. Does that have any relevance AT ALL? NO! We're discussing the top-runners here. There are also examples of PS2 games that push incredibly sorry excuses for "polygon counts" - Disgaea: Hour of Darkness, for one.
What makes you think MGS2 would be possible on DC as-is? The game keeps track of a lot in RAM, I would be surprised if compromises wouldn't be necessary to get the game engine to run in 16MB RAM.
Graphics wise, though, it is somewhat possible, though the motion blur during cinematics would have to go.
MGS2 uses very few transparencies to begin with! You obviously haven't played the game much! And Xbox has very few problems with transparency, the problem Shenmue II had was that DC is actually quite exceptional at handling transparencies, as already covered, although it has so little fill rate that you can't really get away with many layers before your frame rate is screwed.
Now, I'm going to explain to you why you have now forfeited the right to use the term "Anti-aliasing" and officially don't know what you're talking about when you mention it.
Most Dreamcast, PS2, GameCube, and Xbox games do not use anti-aliasing at all. For the most part, they are rendering to a perfectly normal 640x480 frame.
If you want to see the only game console thus far that uses AA consistently, play some N64 games on a real N64.
AA is a kind of "blurring" effect between pixels, a way of smoothing out imperfections due to 640x480 being a pretty... thin sample range. This is most visible on polygon edges, such as stairs (look at Quake3 for Dreamcast for a good example of this). Viewed at a slight angle, there will be a hard 'stairstepping' effect along the boundary between polygons.
Anti-aliasing would result in that 'stairstepping' being smoothed out. Instead of the edge going from being a nice line, to suddenly jumping to the next pixel over, and then resuming its straight-lineyness, it will go from being a solid line, to being a line blended with the background colour by one step, before disappearing. It makes edges more 'continuous'.
This is easily visible in a very huge number of N64 games - the N64 features an extremely efficient - if not exactly perfect - algorithm that is VERY fast, but not the most accurate thing in the world. In a given scene, often many edges can be seen that aren't anti-aliased at all.
Anyway, what you're mixing up is filter quality. Most TV's as I'm sure we're all aware output an interlaced signal, whereby only even or odd scanlines of a frame are visible in a given 60th of a second. The console is rendering a full frame in that interval (usually), so in order to show a 'correct' image, the output is filtered before going to the TV. You can see a few examples of shit filtering in hi-res Saturn games - flicker mania (good example of this is Virtua Fighter 2).
Nearly everything in PS2 is highly programmable. This includes the flicker filter, amusingly enough. Which means that, as with nearly everything else in the console, the fliter quality is directly related to the effort the developer is willing to put in. Early on, thanks to the incredibly bad development environment and mediocre-at-best documentation, the flicker filtering was REALLY, REALLY BAD (see also just about any game before around FFX's release).
PS2 has an additional problem that mauls the possible flicker filter quality... and that's the limited memory on the graphics chip. A lot of early games cut a huge corner in order to make room for textures... instead of rendering the normal, expected full 640x480 front and back buffers, they would render one (or in some particularly dreadful cases, BOTH) buffers half-sized, at 640x240. Visually, since you're only displaying that much anyway, there isn't much difference, but... ouch. Flicker festival.
THAT is the problem older PS2 games have. Newer ones, as far as I can see, just have mediocre flicker filters. There are some standouts, though... Guilty Gear X2's flicker filter is just amazing (comparing the misnamed "Anti alias" option ON versus OFF), and Silent Hill 3 has multiple selectable filters, depending on how you want to see the game. Apparently Ridge Racer 6 (aka R:Racing Evolution, but it's RR6 as far as I'm concerned) has three selectable filters, that they also refer to as "Anti aliasing" but screenshots make it obvious that it isn't AA.
