Pixar used some raytracing for Cars and later described it as a huge mistake. Certain shots took over 200 hours per frame. In terms of performance vs. quality, even in movies, they prefer to go scanline. You won't see games going to raytracing any time soon.
In Transformers, they used cube-maps because raytracing was too slow. Is anyone here seriously going to make the case that Transformers looked bad because the reflections weren't perfect?
I work in video games, and ultimately the graphics cards render polygons. There are definitely tools that can down res, but have you ever seen what a 40k poly mesh looks like when you use an auto-down-res to 20k? It gives you a good start, but you still have to remesh it and tweak it by hand.
As for CSG, it's ok for CAD, but not practical at all for games. The realtime stuff is useful when comples models need to be edited in realtime (like for designing car engines), but the approaches like OpenCSG are too slow because each primitive needs to be drawn a bunch of times.
As an example, if you have the expression (A+B+C)*(D+E+F), you have to break the expression into sum-of-terms form which is AD+AE+AF+BC+BE+BF+CD+CE+CF, so that's at least 18 draw calls (acutally more). Alternatively, you could just precompute the CSG model into a triangle mesh. There are other techniques (like Blister) but they're still too slow.
CSG is used in moderation in games for tasks that make sense. For most applications, it doesn't. Imagine trying to model Wolverine with CSG. But even if you use CSG, at the end of the day, you still need an artist to tweak the vertices (which artists can do very quickly).
Not to mention, you also need to be able to lay out UVs.
If you really want your model to look good at higher resolutions, you can always subdivide your triangle meshes, which is robust (unlike triangle reduction). Up-resing meshes is easy. Down-resing takes artist work.
Yeah, the FUD about MS is a bit out of control on slashdot. The bottom line is that every time a new feature comes out on a graphics card, every previous card is technically "obsolete". It doesn't matter if you are using DirectX or OpenGL, with cards doubling in speed every year, you need some way to accomodate new features. In the past, this was a nightmare for game developers, because every feature had to be tested individually (on both DX and OpenGL).
So if you are coding, you have to check for double sided stencil, seperate alpha blending ops, fp16 blending, multiple render targets, etc. and then have code paths for all those combinations. Now, by making DirectX 10.1, all that microsoft is doing is combining upgrades into groups. So instead of having a zillion code paths for a game to take, the developer just has to check if the card supports DX10 or DX10.1, and then you're golden.
The thing to keep in mind is DX10 is in its infancy. DX9 came out in what, 2003? Keep in mind, that most games will support DX9 for some time, mainly because the PS3 and Xbox360 are basically DX9 cards (the Xenos is more like DX9.5). For games which target the consoles, there will be DX9 pc ports, and for developers that need top of the line visuals (see: Valve & Crytek), they will support a DX10 port too.
Dude, no kidding. I do cross platform work on the 360/PS3. You missed a few important points.
That fan on the PS3 kits is no joke, especially the early CEB kits. I turned it on one time, and then my proDG license expired, so I couldn't log in to turn it off!!! Didn't want to hard-reset it either...so yeah, my coworkers were a little ticked.
And of course, there is the whole libgcm situation. Sony's opengl (PSGL) is crap...no one uses it. Libcm is a little more than direct hardware access, but not much.
On the Xbox 360 side, PIX is the coolest thing ever!!!! My only complaint for the Xbox 360 is not enough EDRAM...predicated tiling is such a buzz-kill.
Do you ever wonder what really drives guys to spend the time working on open source software? According to #7:
Both crime and genius are expressions of young men's competitive desires, whose ultimate function in the ancestral environment would have been to increase reproductive success....
Men have had to conquer foreign lands, win battles and wars, compose symphonies, author books, write sonnets, paint cathedral ceilings, make scientific discoveries, play in rock bands, and write new computer software in order to impress women so that they will agree to have sex with them.
The author is missing the point of the uncanny valley. The point of the uncanny valley is the dip. If you have a non-realistic pixar-like character (The Incredibles), you empathise with them. If you have a more realistic, but not good enough character, the character gets creepier (Polar Express). Then, when it gets really good, you can empathise with them again (Hugo Weaving, i.e. Agent Smith, in the Matrix Sequels).
The author basically says "AI is hard". But he doesn't make any real argument as to there being some "valley" where as the characters get less realistic they act more believably, and as they get more realistic, they also act more believably. A much more accurate title would be "AI is a steep hill".
Total BS. The staff on most games is about 40% engineering. There are more artists per title than programmers, but the programmers make more money. I know because I work as a Rendering Engineer for a game company.
Programmers are needed to do everything from writing the code to optimize a mesh's index list for the GPU's post-transform cache, to writing blinn shaders, to making sure that the sliders work in the game so that the artists can work. The visual quality of a game depends on both artists and engineers. You need both, and even if you have an engine like Unreal, there is always more work to do.
You can have mine. I have some ketchup for you if you'd like.
Microsoft has one of the best graphics research groups in the world, probably the best. The gold standard for publishing graphics research is Siggraph, and last year they had authors on 18 of the 98 papers. In contrast, MIT faculty had 5, Intel had 2, and AMD/nVidia/ATI had 0. In the world of graphics, MS Research is a powerhouse. You can see the official list of papers here:
http://www.siggraph.org/s2006/main.php?f=conferenc e&p=papers
or the entire list on one page:
http://www.cs.brown.edu/~tor/sig2006.html
I certainly don't love MS either, but they have a lot of exceptional graphics people as MS Research.
Price discrimination is definitely legal. The airlines openly admit to it. When I was in college economics class a few years ago, they brought in a senior Delta employee who explained who the system worked. They use tricks like if you are staying for a weekend, then you are probably a personal traveler, and would not pay as much, whereas if you are not staying a weekend, then you are more than likely a business traveler who is expensing it. Another technique is to change price based on how you buy it (i.e. in person, from their site, from a 3rd party site like Travelocity.com).
Where Best Buy gets in trouble is that the store managers don't know the policy, or do know and don't enforce it. An example is PriceMatch, where in Canada, they will match any advertised price based on the prices in their computer system. The catch is that they only have three of their competitors in their computer system. So a friend of mine came in with an ad from a company that wasn't on the list, and had to fight for over an hour to get the right price.
These practices are shady, and need tough penalties because if a company gets caught once, they've probably done it literally a hundered times.
I think it's worth noting that because of this issue, some companies are changing their tune. I saw the first half of Season 2 of Battlestar Galactica in the US, but then moved to Canada, where the whole series was delayed by several months (on Space Network). But, now in season three, the Canadian episodes are not delayed. So, companies are adjusting.
I feel like this whole discussion is missing the point. Any individual who makes a lot of money selling eBay has to report their gross income to the IRS. The question is, can the IRS force eBay to give them the information that would allow tham to actually figure which sellers aren't paying.
The strategey is pretty interesting. Essentially, the old business model with DRM specifies that when you are buying a song, you are paying for access to listening to that song. With the new DRM-free model, the RIAA is acknowledging that anyone can steal music online. What you are fundamentally paying for is the convenience of not having to go through the hassle of finding a torrent online/waiting for it to come on the radio/ripping it from a friend's CD.
You're totally right -- this is journalistic sensationalism.
Generally, you don't submit a paper to TOG unless you get rejected from Siggraph first. It seems like for the last few years, Siggraph has accepted 4 fluids papers (i.e. 1 session) per year. Those 4 are the ones that the brightest minds in computer graphics were the most impressed by.
The only two reasons that really matter are 4 and 5.
There really is virtually no R&D involved -- there are numerous companies that have off-the-shelf solutions for motion sensing. Yes, you woud need to design a controller around it, but the core technology that does the motion detection has been available for years.
In terms of copycatting, there is no way that MS or Sony would create a Wii-mote clone. But it wouldn't be shocking to see a motion-controlled remote with an analog stick directly on the remote. There are numerous ways to create a motion sensing remote that is completely different from the wii-mote. An example would be a motion-sensing remote with an anolog stick on one side and 4 buttons on the other.
I can't believe he the article doesn't mention "Independance Day" where Jeff Goldbloom takes a virus on his mac, and uploads it to the alien mothership, which takes down the shields of all the little fighters.
I also liked how the aliens used earth's satelites to send a signal to co-ordinate the time that they would strike. Naturally, an alien race which has mastered faster-than-light travel and can take over our complex satelite system still hasn't figured out how to synchronize its watches.
Actually, they both 512 megs of ram. The difference is that on the Xbox 360, the GPU and the CPU share the same bank of 512 megs (plus an additional 10 megs of EDRAM for the framebuffer) whereas on the PS3, 256 megs are dedicated to the CPU and the other 256 megs are dedicated to the GPU.
They should at least get their facts straight. The article continually states that the Cell has 1 PPU and 8 SPUs, which is WRONG. It has 1 PPU and 7 SPUs because one of the SPUS is disabled to increase yields.
What's facinating is that all the hype about the strength about the console is based on the Cell. But if you read the article that has the cost breakdown, the Cell is $89, and the RSX is $129. The PS3 isn't expensive because Sony is arrogant, or they tried to make it too powerful. It's expensive because of Blue-ray.
In terms of the success of the console, it Blue-Ray becomes the standard, then the PS3 is a success. Case closed. If they win that battle, and every High-Def DVD player made over the next 10 years sends $10 Sony's way, then the PS3 is a success, even if it only sells 10 million consoles.
Then again, if HD-DVD wins, Sony will be in a pretty sticky situation. Sony looks to me like they've put all thier chips on the table and are betting the company on Blue-Ray.
It was announced at GDC that the ps3 supports both PSGL which is the high-level language, and libgcm which is the low level language. The console is a closed system that you more that likely have to optimize the hell out of (just ask anyone who programmed for the ps2). It's a good bet that most people will go libgcm.
Generally, the more units the game sells, the more platforms that will be supported. So even if only 1% of WoW's users use Win2k, it still makes sense to support it. So the huge cash-cow games like WoW will support everything.
The question is what will the smaller games do?
Very few people use opengl for a variety of reasons. The main reason is that in order to do anything useful, you have to use a ton of extensions. That, and opengl is still getting it's act together. Microsoft finished their API for next-gen graphics (DX10) as well as providing a reference rasterizer last December. Right now, Kronos group just has some proposals for things like geometry shaders and stream out, but nothing definite. So if you're developing a Vista-only game, the choice is pretty obvious.
In terms of cross-platform compatibility, the number of units that a typical game would expect to sell for Linux or Mac is so small that it's not worth the cost. What most games are more concerned about is PS3, Xbox 360, and Wii, generally in that order.
Slashdot Mirror
Definitely.
Pixar used some raytracing for Cars and later described it as a huge mistake. Certain shots took over 200 hours per frame. In terms of performance vs. quality, even in movies, they prefer to go scanline. You won't see games going to raytracing any time soon.
In Transformers, they used cube-maps because raytracing was too slow. Is anyone here seriously going to make the case that Transformers looked bad because the reflections weren't perfect?
Not exactly.
I work in video games, and ultimately the graphics cards render polygons. There are definitely tools that can down res, but have you ever seen what a 40k poly mesh looks like when you use an auto-down-res to 20k? It gives you a good start, but you still have to remesh it and tweak it by hand.
As for CSG, it's ok for CAD, but not practical at all for games. The realtime stuff is useful when comples models need to be edited in realtime (like for designing car engines), but the approaches like OpenCSG are too slow because each primitive needs to be drawn a bunch of times.
As an example, if you have the expression (A+B+C)*(D+E+F), you have to break the expression into sum-of-terms form which is AD+AE+AF+BC+BE+BF+CD+CE+CF, so that's at least 18 draw calls (acutally more). Alternatively, you could just precompute the CSG model into a triangle mesh. There are other techniques (like Blister) but they're still too slow.
CSG is used in moderation in games for tasks that make sense. For most applications, it doesn't. Imagine trying to model Wolverine with CSG. But even if you use CSG, at the end of the day, you still need an artist to tweak the vertices (which artists can do very quickly).
Not to mention, you also need to be able to lay out UVs.
If you really want your model to look good at higher resolutions, you can always subdivide your triangle meshes, which is robust (unlike triangle reduction). Up-resing meshes is easy. Down-resing takes artist work.
Yeah, the FUD about MS is a bit out of control on slashdot. The bottom line is that every time a new feature comes out on a graphics card, every previous card is technically "obsolete". It doesn't matter if you are using DirectX or OpenGL, with cards doubling in speed every year, you need some way to accomodate new features. In the past, this was a nightmare for game developers, because every feature had to be tested individually (on both DX and OpenGL).
So if you are coding, you have to check for double sided stencil, seperate alpha blending ops, fp16 blending, multiple render targets, etc. and then have code paths for all those combinations. Now, by making DirectX 10.1, all that microsoft is doing is combining upgrades into groups. So instead of having a zillion code paths for a game to take, the developer just has to check if the card supports DX10 or DX10.1, and then you're golden.
The thing to keep in mind is DX10 is in its infancy. DX9 came out in what, 2003? Keep in mind, that most games will support DX9 for some time, mainly because the PS3 and Xbox360 are basically DX9 cards (the Xenos is more like DX9.5). For games which target the consoles, there will be DX9 pc ports, and for developers that need top of the line visuals (see: Valve & Crytek), they will support a DX10 port too.
Dude, no kidding. I do cross platform work on the 360/PS3. You missed a few important points.
That fan on the PS3 kits is no joke, especially the early CEB kits. I turned it on one time, and then my proDG license expired, so I couldn't log in to turn it off!!! Didn't want to hard-reset it either...so yeah, my coworkers were a little ticked.
And of course, there is the whole libgcm situation. Sony's opengl (PSGL) is crap...no one uses it. Libcm is a little more than direct hardware access, but not much.
On the Xbox 360 side, PIX is the coolest thing ever!!!! My only complaint for the Xbox 360 is not enough EDRAM...predicated tiling is such a buzz-kill.
Do you ever wonder what really drives guys to spend the time working on open source software? According to #7:
Both crime and genius are expressions of young men's competitive desires, whose ultimate function in the ancestral environment would have been to increase reproductive success.
Men have had to conquer foreign lands, win battles and wars, compose symphonies, author books, write sonnets, paint cathedral ceilings, make scientific discoveries, play in rock bands, and write new computer software in order to impress women so that they will agree to have sex with them.
The author is missing the point of the uncanny valley. The point of the uncanny valley is the dip. If you have a non-realistic pixar-like character (The Incredibles), you empathise with them. If you have a more realistic, but not good enough character, the character gets creepier (Polar Express). Then, when it gets really good, you can empathise with them again (Hugo Weaving, i.e. Agent Smith, in the Matrix Sequels).
The author basically says "AI is hard". But he doesn't make any real argument as to there being some "valley" where as the characters get less realistic they act more believably, and as they get more realistic, they also act more believably. A much more accurate title would be "AI is a steep hill".
Total BS. The staff on most games is about 40% engineering. There are more artists per title than programmers, but the programmers make more money. I know because I work as a Rendering Engineer for a game company.
Programmers are needed to do everything from writing the code to optimize a mesh's index list for the GPU's post-transform cache, to writing blinn shaders, to making sure that the sliders work in the game so that the artists can work. The visual quality of a game depends on both artists and engineers. You need both, and even if you have an engine like Unreal, there is always more work to do.
You can have mine. I have some ketchup for you if you'd like. Microsoft has one of the best graphics research groups in the world, probably the best. The gold standard for publishing graphics research is Siggraph, and last year they had authors on 18 of the 98 papers. In contrast, MIT faculty had 5, Intel had 2, and AMD/nVidia/ATI had 0. In the world of graphics, MS Research is a powerhouse. You can see the official list of papers here: http://www.siggraph.org/s2006/main.php?f=conferenc e&p=papers
or the entire list on one page:
http://www.cs.brown.edu/~tor/sig2006.html
I certainly don't love MS either, but they have a lot of exceptional graphics people as MS Research.
Price discrimination is definitely legal. The airlines openly admit to it. When I was in college economics class a few years ago, they brought in a senior Delta employee who explained who the system worked. They use tricks like if you are staying for a weekend, then you are probably a personal traveler, and would not pay as much, whereas if you are not staying a weekend, then you are more than likely a business traveler who is expensing it. Another technique is to change price based on how you buy it (i.e. in person, from their site, from a 3rd party site like Travelocity.com).
Where Best Buy gets in trouble is that the store managers don't know the policy, or do know and don't enforce it. An example is PriceMatch, where in Canada, they will match any advertised price based on the prices in their computer system. The catch is that they only have three of their competitors in their computer system. So a friend of mine came in with an ad from a company that wasn't on the list, and had to fight for over an hour to get the right price.
These practices are shady, and need tough penalties because if a company gets caught once, they've probably done it literally a hundered times.
I think it's worth noting that because of this issue, some companies are changing their tune. I saw the first half of Season 2 of Battlestar Galactica in the US, but then moved to Canada, where the whole series was delayed by several months (on Space Network). But, now in season three, the Canadian episodes are not delayed. So, companies are adjusting.
I feel like this whole discussion is missing the point. Any individual who makes a lot of money selling eBay has to report their gross income to the IRS. The question is, can the IRS force eBay to give them the information that would allow tham to actually figure which sellers aren't paying.
The strategey is pretty interesting. Essentially, the old business model with DRM specifies that when you are buying a song, you are paying for access to listening to that song. With the new DRM-free model, the RIAA is acknowledging that anyone can steal music online. What you are fundamentally paying for is the convenience of not having to go through the hassle of finding a torrent online/waiting for it to come on the radio/ripping it from a friend's CD.
You're totally right -- this is journalistic sensationalism. Generally, you don't submit a paper to TOG unless you get rejected from Siggraph first. It seems like for the last few years, Siggraph has accepted 4 fluids papers (i.e. 1 session) per year. Those 4 are the ones that the brightest minds in computer graphics were the most impressed by.
The only two reasons that really matter are 4 and 5.
There really is virtually no R&D involved -- there are numerous companies that have off-the-shelf solutions for motion sensing. Yes, you woud need to design a controller around it, but the core technology that does the motion detection has been available for years.
In terms of copycatting, there is no way that MS or Sony would create a Wii-mote clone. But it wouldn't be shocking to see a motion-controlled remote with an analog stick directly on the remote. There are numerous ways to create a motion sensing remote that is completely different from the wii-mote. An example would be a motion-sensing remote with an anolog stick on one side and 4 buttons on the other.
I can't believe he the article doesn't mention "Independance Day" where Jeff Goldbloom takes a virus on his mac, and uploads it to the alien mothership, which takes down the shields of all the little fighters.
I also liked how the aliens used earth's satelites to send a signal to co-ordinate the time that they would strike. Naturally, an alien race which has mastered faster-than-light travel and can take over our complex satelite system still hasn't figured out how to synchronize its watches.
What, you mean you aren't completely sold the Sony strategy? LOOK!!! IT COST'S $600!! IT MUST BE AWESOME!!!!!
Actually, they both 512 megs of ram. The difference is that on the Xbox 360, the GPU and the CPU share the same bank of 512 megs (plus an additional 10 megs of EDRAM for the framebuffer) whereas on the PS3, 256 megs are dedicated to the CPU and the other 256 megs are dedicated to the GPU.
They should at least get their facts straight. The article continually states that the Cell has 1 PPU and 8 SPUs, which is WRONG. It has 1 PPU and 7 SPUs because one of the SPUS is disabled to increase yields.
What's facinating is that all the hype about the strength about the console is based on the Cell. But if you read the article that has the cost breakdown, the Cell is $89, and the RSX is $129. The PS3 isn't expensive because Sony is arrogant, or they tried to make it too powerful. It's expensive because of Blue-ray.
In terms of the success of the console, it Blue-Ray becomes the standard, then the PS3 is a success. Case closed. If they win that battle, and every High-Def DVD player made over the next 10 years sends $10 Sony's way, then the PS3 is a success, even if it only sells 10 million consoles.
Then again, if HD-DVD wins, Sony will be in a pretty sticky situation. Sony looks to me like they've put all thier chips on the table and are betting the company on Blue-Ray.
It was announced at GDC that the ps3 supports both PSGL which is the high-level language, and libgcm which is the low level language. The console is a closed system that you more that likely have to optimize the hell out of (just ask anyone who programmed for the ps2). It's a good bet that most people will go libgcm.
p s2-info-from-gdc--rsx-based-on-nv47.html
http://www.motherboardpoint.com/t87377-technical-
Generally, the more units the game sells, the more platforms that will be supported. So even if only 1% of WoW's users use Win2k, it still makes sense to support it. So the huge cash-cow games like WoW will support everything. The question is what will the smaller games do?
Very few people use opengl for a variety of reasons. The main reason is that in order to do anything useful, you have to use a ton of extensions. That, and opengl is still getting it's act together. Microsoft finished their API for next-gen graphics (DX10) as well as providing a reference rasterizer last December. Right now, Kronos group just has some proposals for things like geometry shaders and stream out, but nothing definite. So if you're developing a Vista-only game, the choice is pretty obvious. In terms of cross-platform compatibility, the number of units that a typical game would expect to sell for Linux or Mac is so small that it's not worth the cost. What most games are more concerned about is PS3, Xbox 360, and Wii, generally in that order.