I think we "choose" to use a 2D interface in the real world for things like paper on a desktop because there is no practical way for it to be 3D. Computers don't necessarily have that limitation. There are a lot of factors in making such an interface practical, but "being 3d" isn't the problem.
The article does address going after the advertisers themselves (those selling the products), but says it is difficult to prove they knew the advertising was via spam, and that this is required under CAN-SPAM in order to find wrongdoing.
One thing that could help, either by setting a precedent in court, or via modification of the law, is to say that unless their advertising contract has a clause saying something like "if we discover that any of this advertising went out via illegal spam, no fee or commission will be paid", then by default they are considered to have potentially opened themselves up to participating in spam advertising. In that case it would be much easier to go after the companies advertising the products, who would then have to shy away from the spammers to keep from getting sued. Of course, they could insert such clauses in their contracts then ignore them (saying "we didn't know they were spamming, and now that we do we'll stop paying them"), but then a lawsuit could perhaps force them to open their books to prove they didn't pay the spammers any fees or commissions.
Since the spamming operations themselves are so slippery, I think going after the original advertisers may be the more productive legislative avenue, in conjunction with the various technical approaches to verifying the source of emails.
Despite having different APIs, 3D cards (and the various APIs) until recently were all basically doing the same thing: applying stored textures (bitmaps) to polygons being rasterized to a 2D frame buffer (usually filtering the textures in the process) - a very repetitive process with little or no decision making involved (so it can run in long pipelines), well-served by the standard pixel-pipeline structure of a 3D chip with its own dedicted memory and high bandwidth that didn't tie up the main CPU and memory with such mundane, highly-repetitive, bandwidth-eating tasks. Until recently, they all more-or-less hard-wired the same basic algorithm.
As has been pointed out, other than possibly neural networks or some other highly-parallel structure, it's not clear what specific algorithm or function would be hard-wired into an AI chip (which is usually the main point of having a special co-processor), or why a standard CPU isn't fairly good at doing AI tasks efficiently (particularly as AI, involving decision-making, probably doesn't benefit so much from long pipelines as it does from generalized branching and quick access to the full range of memory, which is what CPUs are already designed for).
Maybe someday, but as of now, AI processing is neither commonly-used enough nor standardized enough in terms of specific highly-repetitive algorithms for such a chip to be worthwhile.
I think that games for which "dumb" AI can easily be programmed to beat any player betray limitations in the game. The truly deep games tend to be much harder to create good AI for (which doesn't mean the games need to be complicated in an obvious way). Even having perfect knowledge of game state shouldn't make it easy for the AI to beat a player. (A lousy player, yes, but not a good one.)
In a game like Quake, one might think that an AI "bot" that sees through walls (and with a 360 degree view at all times), has perfect aim, instantaneous reflexes, perfect knowledge of its opponent's health and other status, perfect programming and timing for tricky jumps (rocket and otherwise), perfect level knowledge and awareness of item respawn timings, etc. would be unbeatable - but it's not, not by a long shot. The best players should be able to soundly beat even such a bot. Aim and reflexes are just the basic prerequisites for this game, and the rest gets to be so complicated and variable that designing truly great AI for it would be a huge challenge.
CRTs can dispaly far more colors than LCDs can. More shades than a 24-bit graphic card can address. The gamut varies with different technologies and each has its limitations, but LCDs have limitations on color range (not so much in terms of total range, as in how many discrete steps can be displayed), response rate, and viewing angle that are far more restrictive than other common display technologies. (Rear projection CRTs also typically have restricted viewing angles, but are good otherwise.) I am not positive if these restrictions apply so much to LCDoS, though, which is a significantly different technology (reflective, not transmissive).
Even 24-bit color is very frequently inadequate. For much material, it looks fine, but whenever you encounter a gradient of a relatively flat color, it is quite likely to exhibit visible banding. In addition, dark areas almost never show up well. It has just become accepted that the dark areas in many photos, videos, and games are difficult to make out on a computer monitor, even set at 24-bit color, no matter how high the brightness is set. This is because there aren't enough colors to do the job properly. When something gets close enough to black, it just becomes black, instead of a very dark grey. Remember, there are only 256 shades of each of 3 colors at a 24-bit setting. Dithering and gamma correction can cover this up pretty well if they're used properly, but very often they don't seem to be.
(Floating point color representations can address these problems without needing to increase the number of bits used.)
16-bit color is even worse, let alone the more approximately 15 bits most LCDs are capable of. That's only 32 levels of each of 3 colors! Probably the main reason you're not seeing it as a problem is that most sources are similarly limited: most digital TV seems to use severe enough compression that it's not really doing much better, NTSC's color representation sucks (needless to say, though its low resolution and bleed may actually cover up some of these problems. if they even exist in the first place, since it's analog), and if you always use a 16-bit color setting on your computer (which is usually ok since most on-screen graphics are pretty basic color-wise, and desktop color schemes are designed to work well in 16 or even 8-bit color), you're probably not missing much from any of these sources. That just says that the source and the display are all limted pretty much equally, which makes them an OK match in these cases.
But there are better sources out there, including 24-bit graphics cards, good DVDs, maybe good HD, even good analog sources, and for these, LCDs will be more limited.
This is something interesting: when I first got my LCD computer monitor, the motion blur seemed terrible in games (such as QuakeWorld and Unreal Tournament). It was only occasionally really visible, but I think it messed up my performance at first.
But now that I am used to it, I don't notice the motion blur at all... It's funny what the eye can see or not see.
Well the problem is that there isn't an LCD display available that can TRULY display a 60Hz TV signal without any motion blur being added. The problem here is that you are accepting the hype and marketing "specifications" that would lead most people to believe that they can really update that quickly. The reality is quite different.
On the other hand, most people seem highly tolerant of really horrible picture quality in all different ways (slow refresh or framerate, jitter, poor color quality, low resoltion, obvious compression artifacts, etc, which begs the question of why we are even attempting to move to "HD" resolutions, as half the people out there probably wouldn't even notice the difference from a typical viewing distance.)
This is possibly true if your signal always matches the display resolution, but for the majority of LCD and plasma HDTVs, this is NOT the case most of the time, if ever. Most plasmas have 768-line resolution, so ALL images are scaled. Other LCDs and plasmas may be 1024x1024 or some other format. Some even use an interlace refresh technology. Even the ones that are actually 720p (1280x720) will only exactly match a 720p signal; most HDTV signals are 1080i, and the vast majority of TV (including DVDs) isn't HD at all, it's 480 lines, usually interlaced. So nearly everything is scaled on these displays, and the same will be true on the newest (and most expensive) fixed-pixel displays with true 1920x1080 resolution.
The problems with scaling, even when done well, certainly outweigh any inaccuracy found in a good modern CRT.
First of all, the claimed refresh rates of LCDs in ms don't really mean much. Most of them don't truly display images that quickly without motion blur, and most of them aren't even rated better than 20-25ms. Also, TV signals are 60Hz, not 30. There IS discrete motion occurring every 60th of a second (other than when showing movies that were originally 24fps), and even though interlace pretty much sucks, that 60Hz response is important to the smoothness of motion on TV.
The other big problem with LCDs is color depth and accuracy and black levels (ability to display darker images especially). On computers, this is not so much of a problem since the standard video card output has similar problems: even 24-bit color handles dark images poorly, and can have banding on gradients. Either 48/64-bit or going to a floating point format would help a lot with this.
As has been pointed out elsewhere, it may not matter so much for digital TV of various sorts either, as the signals often have limited color ability already. But this is just saying that in some ways, TV picture quality is going backward as we move to digital, in terms of both signal and display technology.
Plasma is better than LCD, and closer to CRT, but still not considered quite as good by those who really pay attention to picture quality.
When it comes to rear projection, CRT is also considered the best (and especially the best bargain), though it is the bulkiest and heaviest, and needs the most setup and adjustment (and is susceptible to burn in, requires a relatively dim environment, has limited viewing angle, etc.) Second is probably LCDoS, which is a REFLECTIVE technology, not the usual transmissive LCD type, followed by DLP, then regular LCD rear projection.
People are buying LCD and plasma displays because they are compact, NOT because they have better image quality (or even just as good): I think roughly half of plasmas sold aren't even HD, so people are willing to spend $3000+ just to get a flat, largish television, when they may not be willing to spend that much to get an actual HD picture. Thus many of them aren't really concerned with other aspects of picture quality. But the ones who are would be making a mistake if they think that the newest and most expensive display technologies necessarily have the best pictures.
I've noticed the color banding a LOT on satellite TV (Dish in this case), on all channels. Since I started noticing it, I now notice it all the time (this is on a Sony direct-view trinitron CRT TV), or else the problem has gotten worse. It's always there, throwing away huge amounts of the image, but it's only blatantly obvious in some scenes. (I suspect a side-by-side comparison would make it obvious in virtually ALL scenes.) So in this case, the signal is already bad so LCD probably wouldn't make it much worse.
At least as much as it does on a computer! You will likely strain to see what is happening in dark parts of a scene, and you may see obvious banding in many places, especially with fog or dark areas. I think most LCDs are physically incapable of displaying more than about 32,000 distinct colors.
On the other hand, if you're watching digital cable or satellite, the digital signal may already be so vastly overcompressed that it has its only obvious banding and looks just as bad on a CRT. (I'm not sure to what extent this problem might also affect DVD or [H]DTV.)
This kind of limitation is already built in on computers and in many digital video signals, even with 24-bit color. Remember that this is only 256 levels of each color, so gradients may be obviously banded, and subtly-colored detail may disappear. This is why John Carmack some years ago called for 48 or 64-bit color, and newer video cards can process at greater than 24-bit precision, at least internally. We've just gotten used to straining to see dark areas in photos, videos, and games on our computers over the years, as well as commonly encountering banded gradients.
(Resulting color inaccuracy is a lesser problem usually when set to 24 bits, as our eyes readily adapt to different gamuts and the variation in monitor calibration is more of a problem.)
I think that any of the engines can easily create stuff that ends up looking not too much different from the game itself. The problem is creating the textures, models, animations, environments, sounds, etc. to make it look like them movie you want, rather than the game, and sometimes having the types of behaviors with the look you want, as opposed to behaving like a game. And once you've done all that, really you probably might as well have worked in a more standard 3D modelling/animation setup, which is more flexible.
If you want to do something quick and dirty in a look that the game already provides, great, but if you want it to look much different from the game itself, you may want to avoid a game engine for movie making.
Slashdot Mirror
I think we "choose" to use a 2D interface in the real world for things like paper on a desktop because there is no practical way for it to be 3D. Computers don't necessarily have that limitation. There are a lot of factors in making such an interface practical, but "being 3d" isn't the problem.
The article does address going after the advertisers themselves (those selling the products), but says it is difficult to prove they knew the advertising was via spam, and that this is required under CAN-SPAM in order to find wrongdoing.
One thing that could help, either by setting a precedent in court, or via modification of the law, is to say that unless their advertising contract has a clause saying something like "if we discover that any of this advertising went out via illegal spam, no fee or commission will be paid", then by default they are considered to have potentially opened themselves up to participating in spam advertising. In that case it would be much easier to go after the companies advertising the products, who would then have to shy away from the spammers to keep from getting sued. Of course, they could insert such clauses in their contracts then ignore them (saying "we didn't know they were spamming, and now that we do we'll stop paying them"), but then a lawsuit could perhaps force them to open their books to prove they didn't pay the spammers any fees or commissions.
Since the spamming operations themselves are so slippery, I think going after the original advertisers may be the more productive legislative avenue, in conjunction with the various technical approaches to verifying the source of emails.
Despite having different APIs, 3D cards (and the various APIs) until recently were all basically doing the same thing: applying stored textures (bitmaps) to polygons being rasterized to a 2D frame buffer (usually filtering the textures in the process) - a very repetitive process with little or no decision making involved (so it can run in long pipelines), well-served by the standard pixel-pipeline structure of a 3D chip with its own dedicted memory and high bandwidth that didn't tie up the main CPU and memory with such mundane, highly-repetitive, bandwidth-eating tasks. Until recently, they all more-or-less hard-wired the same basic algorithm.
As has been pointed out, other than possibly neural networks or some other highly-parallel structure, it's not clear what specific algorithm or function would be hard-wired into an AI chip (which is usually the main point of having a special co-processor), or why a standard CPU isn't fairly good at doing AI tasks efficiently (particularly as AI, involving decision-making, probably doesn't benefit so much from long pipelines as it does from generalized branching and quick access to the full range of memory, which is what CPUs are already designed for). Maybe someday, but as of now, AI processing is neither commonly-used enough nor standardized enough in terms of specific highly-repetitive algorithms for such a chip to be worthwhile.
I think that games for which "dumb" AI can easily be programmed to beat any player betray limitations in the game. The truly deep games tend to be much harder to create good AI for (which doesn't mean the games need to be complicated in an obvious way). Even having perfect knowledge of game state shouldn't make it easy for the AI to beat a player. (A lousy player, yes, but not a good one.)
In a game like Quake, one might think that an AI "bot" that sees through walls (and with a 360 degree view at all times), has perfect aim, instantaneous reflexes, perfect knowledge of its opponent's health and other status, perfect programming and timing for tricky jumps (rocket and otherwise), perfect level knowledge and awareness of item respawn timings, etc. would be unbeatable - but it's not, not by a long shot. The best players should be able to soundly beat even such a bot. Aim and reflexes are just the basic prerequisites for this game, and the rest gets to be so complicated and variable that designing truly great AI for it would be a huge challenge.
CRTs can dispaly far more colors than LCDs can. More shades than a 24-bit graphic card can address. The gamut varies with different technologies and each has its limitations, but LCDs have limitations on color range (not so much in terms of total range, as in how many discrete steps can be displayed), response rate, and viewing angle that are far more restrictive than other common display technologies. (Rear projection CRTs also typically have restricted viewing angles, but are good otherwise.) I am not positive if these restrictions apply so much to LCDoS, though, which is a significantly different technology (reflective, not transmissive).
Even 24-bit color is very frequently inadequate. For much material, it looks fine, but whenever you encounter a gradient of a relatively flat color, it is quite likely to exhibit visible banding. In addition, dark areas almost never show up well. It has just become accepted that the dark areas in many photos, videos, and games are difficult to make out on a computer monitor, even set at 24-bit color, no matter how high the brightness is set. This is because there aren't enough colors to do the job properly. When something gets close enough to black, it just becomes black, instead of a very dark grey. Remember, there are only 256 shades of each of 3 colors at a 24-bit setting. Dithering and gamma correction can cover this up pretty well if they're used properly, but very often they don't seem to be.
(Floating point color representations can address these problems without needing to increase the number of bits used.)
16-bit color is even worse, let alone the more approximately 15 bits most LCDs are capable of. That's only 32 levels of each of 3 colors! Probably the main reason you're not seeing it as a problem is that most sources are similarly limited: most digital TV seems to use severe enough compression that it's not really doing much better, NTSC's color representation sucks (needless to say, though its low resolution and bleed may actually cover up some of these problems. if they even exist in the first place, since it's analog), and if you always use a 16-bit color setting on your computer (which is usually ok since most on-screen graphics are pretty basic color-wise, and desktop color schemes are designed to work well in 16 or even 8-bit color), you're probably not missing much from any of these sources. That just says that the source and the display are all limted pretty much equally, which makes them an OK match in these cases.
But there are better sources out there, including 24-bit graphics cards, good DVDs, maybe good HD, even good analog sources, and for these, LCDs will be more limited.
This is something interesting: when I first got my LCD computer monitor, the motion blur seemed terrible in games (such as QuakeWorld and Unreal Tournament). It was only occasionally really visible, but I think it messed up my performance at first.
But now that I am used to it, I don't notice the motion blur at all... It's funny what the eye can see or not see.
Well the problem is that there isn't an LCD display available that can TRULY display a 60Hz TV signal without any motion blur being added. The problem here is that you are accepting the hype and marketing "specifications" that would lead most people to believe that they can really update that quickly. The reality is quite different.
On the other hand, most people seem highly tolerant of really horrible picture quality in all different ways (slow refresh or framerate, jitter, poor color quality, low resoltion, obvious compression artifacts, etc, which begs the question of why we are even attempting to move to "HD" resolutions, as half the people out there probably wouldn't even notice the difference from a typical viewing distance.)
This is possibly true if your signal always matches the display resolution, but for the majority of LCD and plasma HDTVs, this is NOT the case most of the time, if ever. Most plasmas have 768-line resolution, so ALL images are scaled. Other LCDs and plasmas may be 1024x1024 or some other format. Some even use an interlace refresh technology. Even the ones that are actually 720p (1280x720) will only exactly match a 720p signal; most HDTV signals are 1080i, and the vast majority of TV (including DVDs) isn't HD at all, it's 480 lines, usually interlaced. So nearly everything is scaled on these displays, and the same will be true on the newest (and most expensive) fixed-pixel displays with true 1920x1080 resolution.
The problems with scaling, even when done well, certainly outweigh any inaccuracy found in a good modern CRT.
First of all, the claimed refresh rates of LCDs in ms don't really mean much. Most of them don't truly display images that quickly without motion blur, and most of them aren't even rated better than 20-25ms. Also, TV signals are 60Hz, not 30. There IS discrete motion occurring every 60th of a second (other than when showing movies that were originally 24fps), and even though interlace pretty much sucks, that 60Hz response is important to the smoothness of motion on TV.
The other big problem with LCDs is color depth and accuracy and black levels (ability to display darker images especially). On computers, this is not so much of a problem since the standard video card output has similar problems: even 24-bit color handles dark images poorly, and can have banding on gradients. Either 48/64-bit or going to a floating point format would help a lot with this.
As has been pointed out elsewhere, it may not matter so much for digital TV of various sorts either, as the signals often have limited color ability already. But this is just saying that in some ways, TV picture quality is going backward as we move to digital, in terms of both signal and display technology. Plasma is better than LCD, and closer to CRT, but still not considered quite as good by those who really pay attention to picture quality.
When it comes to rear projection, CRT is also considered the best (and especially the best bargain), though it is the bulkiest and heaviest, and needs the most setup and adjustment (and is susceptible to burn in, requires a relatively dim environment, has limited viewing angle, etc.) Second is probably LCDoS, which is a REFLECTIVE technology, not the usual transmissive LCD type, followed by DLP, then regular LCD rear projection.
People are buying LCD and plasma displays because they are compact, NOT because they have better image quality (or even just as good): I think roughly half of plasmas sold aren't even HD, so people are willing to spend $3000+ just to get a flat, largish television, when they may not be willing to spend that much to get an actual HD picture. Thus many of them aren't really concerned with other aspects of picture quality. But the ones who are would be making a mistake if they think that the newest and most expensive display technologies necessarily have the best pictures.
I've noticed the color banding a LOT on satellite TV (Dish in this case), on all channels. Since I started noticing it, I now notice it all the time (this is on a Sony direct-view trinitron CRT TV), or else the problem has gotten worse. It's always there, throwing away huge amounts of the image, but it's only blatantly obvious in some scenes. (I suspect a side-by-side comparison would make it obvious in virtually ALL scenes.) So in this case, the signal is already bad so LCD probably wouldn't make it much worse.
At least as much as it does on a computer! You will likely strain to see what is happening in dark parts of a scene, and you may see obvious banding in many places, especially with fog or dark areas. I think most LCDs are physically incapable of displaying more than about 32,000 distinct colors. On the other hand, if you're watching digital cable or satellite, the digital signal may already be so vastly overcompressed that it has its only obvious banding and looks just as bad on a CRT. (I'm not sure to what extent this problem might also affect DVD or [H]DTV.)
This kind of limitation is already built in on computers and in many digital video signals, even with 24-bit color. Remember that this is only 256 levels of each color, so gradients may be obviously banded, and subtly-colored detail may disappear. This is why John Carmack some years ago called for 48 or 64-bit color, and newer video cards can process at greater than 24-bit precision, at least internally. We've just gotten used to straining to see dark areas in photos, videos, and games on our computers over the years, as well as commonly encountering banded gradients.
(Resulting color inaccuracy is a lesser problem usually when set to 24 bits, as our eyes readily adapt to different gamuts and the variation in monitor calibration is more of a problem.)
I think that any of the engines can easily create stuff that ends up looking not too much different from the game itself. The problem is creating the textures, models, animations, environments, sounds, etc. to make it look like them movie you want, rather than the game, and sometimes having the types of behaviors with the look you want, as opposed to behaving like a game. And once you've done all that, really you probably might as well have worked in a more standard 3D modelling/animation setup, which is more flexible.
If you want to do something quick and dirty in a look that the game already provides, great, but if you want it to look much different from the game itself, you may want to avoid a game engine for movie making.