Yes, you are right. I got passes mixed up. Yet I am sure that by the time Xbox hits the streets there will be a NVidia chipset for PCs that does eight textures per pass. Or sixteen, whatever. It seems hardcore gamers will pay anything, so the race isn't gonna slow down any soon.
30 texture passes would give renderman-like quality? hmm... Time to go hit Google.
Mark Peercy of SGI has shown, quite surprisingly, that all Renderman surface
shaders can be decomposed into multi-pass graphics operations if two
extensions are provided over basic OpenGL [...] It may take hundreds or thousands of passes, but it clearly defines an approach with no fundamental limits.
I would appreciate it if you could provide the URL where Carmack says 30 will do. That is only 1.5 generatios away from 4! (well, 3 really;).
I am not confusing megapixels per second with megatris per second. I am comparing both figures, and saying that with decent culling algorithms that minimize overdraw, you dont need more polygons than you have pixels, not really.
An outdoor scene can have a scene complexity of 3.5 - 5 (or average). So at complexity 5, each pixel is overwritten, on average, 5 times
Not to be snotty, but... Where do you get these figures from? I thought VIS (the visibility table compiler for Quake games) took care of those invisble poligons and reduced overdraw.
That is a point. But you are reading/., so you are probably a PC gamer anyway. Average console buyers --the kind that dont have PCs, or use their PCs just for (home)work-- use TVs for their gaming, and they even like that blurry merge-in of the pixels. So you are in a minority there.
I remember another interview with Carmack (won't bother to look up the URL, you will have to trust me on this) where he said that rather than increase resolution, he would up the frame rate and keep the eye candy. But hey, he's only the programmer. I am gilty of going for the 1240x1024 too, sometimes.
I would like to point out that NTSC has a total of 640*480*30fps=9,216,000 pixels per second, and Pal has 720*576*25fps=10,368,000 pixels per second. (Please don't knock me over the specifics, I might be slightly wrong about NTSC -being European, I've never actually worked with it- and I know that Pal has a "square pixel" mode where horizontal resolution is 768, right? Just trying to give you nice thousands here.)
With that in mind, and neglecting overdraw, you don't need more than 12,000,000 polygons/sec anyway. If your rate is steady, that is.
This is why I think Abrash's words are very revealing:
MA: It's impossible to tell what performance developers will get until people are actually programming the hardware. It's also hard to evaluate because the chip is so programmable; how do you compare 125 mtris/s with 1 texture to, say, 12.5 mtris/s with a custom lighting model, along with 4 textures doing reflective bump mapping and a combiner program doing custom shading, plus shadows done on a second pass? It's not a matter of raw polygons anymore, but rather of the impact on image quality of the intersection of many factors: polygons, vertex shading, multitexture, texture lookups, pixel combiners, antialiasing, and multipass.
I am not a graphics überhacker, and don't have the answer on that comparison, but the second option (the way the Xbox design team have taken) sure sounds nicer to programmers. And you don't really need any more triangles anyway. Hmm. It will take Playstation II hackers many headaches to do what will come naturally to the programmers of this simpler-yet-more-complex approach.
On a related note: In a recent interview, John Carmack revealed that the Doom 2000 engine will have eight texture passes per polygon. (I am adding the emphasis). What, the Xbox can only do four? It is clear that id wants us PC gamers to keep our leer on when talking to those lowly conlosers. Hah!
>Unlike their competition, MS has NO EXPERIENCE WHATSOEVER running a giant hardware manufacturing outfit.
It could be a myth, it could be true, but reportedly the Hardware Division at Microsoft is the one that has highest profits per employee.
What's clear is that Microsoft does sell a lot of mice, keyboards and joysticks with their logo on them, so I wouldn't say they are totally unexperienced in the running of a hardware manufacturing outfit. Can you say "farm out"?
Don't pan them just because they are Microsoft. Why dont you wait until they actually screw up?
I would like to point out that NTSC has a total of 640*480*30fps=9,216,000 pixels per second, and Pal has 720*576*25fps=10,368,000 pixels per second. (Please don't knock me over the specifics, I might be slightly wrong about NTSC -being European, I've never actually worked with it- and I know that Pal has a "square pixel" mode where horizontal resolution is 768, right? Just trying to give you nice thousands here.)
With that in mind, and neglecting overdraw, you don't need more than 12,000,000 polygons/sec anyway. If your rate is steady, that is.
This is why I think Abrash's words are very revealing:
MA: It's impossible to tell what performance developers will get until people are actually programming the hardware. It's also hard to evaluate because the chip is so programmable; how do you compare 125 mtris/s with 1 texture to, say, 12.5 mtris/s with a custom lighting model, along with 4 textures doing reflective bump mapping and a combiner program doing custom shading, plus shadows done on a second pass? It's not a matter of raw polygons anymore, but rather of the impact on image quality of the intersection of many factors: polygons, vertex shading, multitexture, texture lookups, pixel combiners, antialiasing, and multipass.
I am not a graphics überhacker, and don't have the answer on that comparison, but the second option (the way tehe Xbox design team have taken) sure sounds nicer to programmers. And you don't really need any more triangles anyway. Hmm. It will take Playstation II hackers many headaches to do what will come naturally to the programmers of this simpler-yet-more-complex approach.
On a related note: In a recent interview, John Carmack revealed that the Doom 2000 engine will have eight texture passes per polygon. (I am adding the emphasis). What, the Xbox only has four? It is clear that id wants us PC gamers to keep our leer on when talking to those lowly conlosers. Hah!
This is one of the (rare) occasions when I wish one could save/. moderation points. Still don't know whether this is more insightful or informative, but if you are reading this at threshold=1 and have a moderation point left, please read the parent comment to this one again and think about it.
Sorry for posting without previewing. This is what I meant to post:
You can <A href="http://barrapunto.com/comments.pl?sid=100/06 /07/0224259&pid=55#95">check out the source code</A> at BarraPunto. <P> IDKVB (I Don't Know Visual Basic;), but it seems that it references Cmos.com when updating the registry. <P> And it definitely writes a binary file at the end of a sub called CopiarCmosAfichero (CopyCmosTofile). To me it seems it is creating Cmos.com so it can write the Cmos to a file, and thus doing what it states, but I would like confirmation on what it is doing.<P>
In 1995 I used to read alt.bio.hacking. It contained a lot of useful debate and info for non-geneticists. I remember reading there that if good genes could be copyrighted, then a couple might get sued for copyright infringement just by having a healthy baby, a meme that has accompanied me since then.
With the advent of the *Earn $$$!!!!* and *How do I hack my BIOS?* crowds, alt.bio.hacking languished and disappeared (at least from the Spanish feeds I read). I have not been able to find an archive (and my own one perished in the Big Disk Drive Meltdown of 1996). If anyone reading this knows where to find old alt.bio.hacking posts, I think they would be a valuable addition to this thread.
Slashdot Mirror
Surely you meant GilDot...
Like Slashdot, but in Portuguese.
Yes, you are right. I got passes mixed up. Yet I am sure that by the time Xbox hits the streets there will be a NVidia chipset for PCs that does eight textures per pass. Or sixteen, whatever. It seems hardcore gamers will pay anything, so the race isn't gonna slow down any soon.
;).
30 texture passes would give renderman-like quality? hmm... Time to go hit Google.
Mark Peercy of SGI has shown, quite surprisingly, that all Renderman surface
shaders can be decomposed into multi-pass graphics operations if two
extensions are provided over basic OpenGL [...] It may take hundreds or thousands of passes, but it clearly defines an approach with no fundamental limits.
I would appreciate it if you could provide the URL where Carmack says 30 will do. That is only 1.5 generatios away from 4! (well, 3 really
I am not confusing megapixels per second with megatris per second. I am comparing both figures, and saying that with decent culling algorithms that minimize overdraw, you dont need more polygons than you have pixels, not really.
An outdoor scene can have a scene complexity of 3.5 - 5 (or average). So at complexity 5, each pixel is overwritten, on average, 5 times
Not to be snotty, but... Where do you get these figures from? I thought VIS (the visibility table compiler for Quake games) took care of those invisble poligons and reduced overdraw.
That is a point. But you are reading /., so you are probably a PC gamer anyway. Average console buyers --the kind that dont have PCs, or use their PCs just for (home)work-- use TVs for their gaming, and they even like that blurry merge-in of the pixels. So you are in a minority there.
I remember another interview with Carmack (won't bother to look up the URL, you will have to trust me on this) where he said that rather than increase resolution, he would up the frame rate and keep the eye candy. But hey, he's only the programmer. I am gilty of going for the 1240x1024 too, sometimes.
This is why I think Abrash's words are very revealing:
I am not a graphics überhacker, and don't have the answer on that comparison, but the second option (the way the Xbox design team have taken) sure sounds nicer to programmers. And you don't really need any more triangles anyway. Hmm. It will take Playstation II hackers many headaches to do what will come naturally to the programmers of this simpler-yet-more-complex approach.On a related note: In a recent interview, John Carmack revealed that the Doom 2000 engine will have eight texture passes per polygon. (I am adding the emphasis). What, the Xbox can only do four? It is clear that id wants us PC gamers to keep our leer on when talking to those lowly conlosers. Hah!
>Unlike their competition, MS has NO EXPERIENCE WHATSOEVER running a giant hardware manufacturing outfit.
It could be a myth, it could be true, but reportedly the Hardware Division at Microsoft is the one that has highest profits per employee.
What's clear is that Microsoft does sell a lot of mice, keyboards and joysticks with their logo on them, so I wouldn't say they are totally unexperienced in the running of a hardware manufacturing outfit. Can you say "farm out"?
Don't pan them just because they are Microsoft. Why dont you wait until they actually screw up?
With that in mind, and neglecting overdraw, you don't need more than 12,000,000 polygons/sec anyway. If your rate is steady, that is.
This is why I think Abrash's words are very revealing:
I am not a graphics überhacker, and don't have the answer on that comparison, but the second option (the way tehe Xbox design team have taken) sure sounds nicer to programmers. And you don't really need any more triangles anyway. Hmm. It will take Playstation II hackers many headaches to do what will come naturally to the programmers of this simpler-yet-more-complex approach.
On a related note: In a recent interview, John Carmack revealed that the Doom 2000 engine will have eight texture passes per polygon. (I am adding the emphasis). What, the Xbox only has four? It is clear that id wants us PC gamers to keep our leer on when talking to those lowly conlosers. Hah!
This is one of the (rare) occasions when I wish one could save /. moderation points. Still don't know whether this is more insightful or informative, but if you are reading this at threshold=1 and have a moderation point left, please read the parent comment to this one again and think about it.
This is embarrasing, I managed to foul up my post -- twice.
/me shoots himself on the foot. Twice.
You can <A href="http://barrapunto.com/comments.pl?sid=100/0
<P>
IDKVB (I Don't Know Visual Basic
<P>
And it definitely writes a binary file at the end of a sub called CopiarCmosAfichero (CopyCmosTofile). To me it seems it is creating Cmos.com so it can write the Cmos to a file, and thus doing what it states, but I would like confirmation on what it is doing.<P>
Javier 'Candyman' Candeira
You can ;), but it seems that it references Cmos.com when updating the registry.
IDKVB (I Don't Know Visual Basic
And it definitely writes a binary file at the end of a sub called CopiarCmosAfichero (CopyCmosTofile).
Ascertain the topology of the network.<BR>
Deduce the IP address of any node.<BR>
Determine the contents of any node.<BR
</i><P>
Interesting notion, and well worded, but can you prove it? <P>
In 1995 I used to read alt.bio.hacking. It contained a lot of useful debate and info for non-geneticists. I remember reading there that if good genes could be copyrighted, then a couple might get sued for copyright infringement just by having a healthy baby, a meme that has accompanied me since then.
With the advent of the *Earn $$$!!!!* and *How do I hack my BIOS?* crowds, alt.bio.hacking languished and disappeared (at least from the Spanish feeds I read). I have not been able to find an archive (and my own one perished in the Big Disk Drive Meltdown of 1996). If anyone reading this knows where to find old alt.bio.hacking posts, I think they would be a valuable addition to this thread.