It's not the "Rose Bowl Parade", damnit. It's the "Tournament of Roses Parade". The parade was started decades before the football game became a standard part of Pasadena's New Years Day celebration. It would be far more accurate to call the game the Tournament of Roses football game than the other way around.
So what do you define as doing things for the public good? As far as I'm concerned, people vote every day with their dollars. They can use that vote more effectively than voting for any politician.
But people can only vote with their dollars if there's a range of goods and services to choose from. If there's a monopoly or oligopoly, there just aren't enough choices to make voting with ones wallet a reasonable approach. You can view this as an economic counterpart to complaints about the lack of choices in our current two party political system. If major parties (or businesses) can squeeze out most of their competitors, then voters (or consumers) aren't necessarily presented with any acceptable choices.
This is particularly important in an area like broadcast media. Whether or not there are good reasons for it, our current system places a sharp limit on the total number of broadcast outlets in a given area. That makes the broadcast market a ripe target for monopolies or oligopolies to take over. The same government that restricts the number of broadcast outlets must ensure that limited number of outlets aren't concentrated into too few hands, or the public won't be able to vote with their dollars.
Second issue: on camera flash is evil. Only a few compact cameras give you a hotshoe. DSLR's will give way better flash results with their bounce flash/diffuser capability. Almost every flash picture I have yields terrible red-eye. Photoshop Album can generally fix this, but not all the time. Even without red eye, you generally get a sterile, harshly lit result.
An even better thing that DSLR's let you do is to take pictures without needing the flash in the first place. With an ISO speed of 800 or 1600 and a wide aperture lens, you can take pictures in ordinary indoor light without needing a flash. Not all of them will be razor sharp- especially if you're photographing a rampaging toddler or some other rapidly moving thing- but with good technique most of them can be.
Shooting in available light also helps to get the people in your pictures behave more naturally. Most people seem to turn into either hams or wallflowers when they know they're being photographed, and zapping them with a flash is a great way of reminding them that their pictures are being taken. The hams can be fun to photograph, but I find that the best pictures are ones of people who are being themselves rather than thinking about how they'll look in a photograph.
Personally, I've always loathed 50mm lenses. I've much prefered either my 28mm and my 105mm.
Do bear in mind, though, that most of the "affordable" DSLR's these days use sub-35mm sized sensors, which changes the use of most lenses. The Canon D20 and Digital Rebel, for instance, have a 1.6x "focal length mulitplier". That means that your favorite 28mm lens will have a field of view rather close to that of a 50mm lens on your film camera, while the 50mm lens that you dislike will have a field of view much closer to that of an 85mm lens on your film camera. This is wonderful for portrait shooters, since it means that they can use cheap 50/1.8 and 85/1.8 lenses. OTOH, it hurts scenic and street shooters, since it makes their 20mm to 24mm wide angle lenses just barely wide of normal.
4) Responsiveness. If most of your pictures are landscapes or people posing for the camera, it's fine if it takes your camera few seconds to turn on, zoom, or focus, if you have to go through menus to set important controls, of if your camera can only take a few pictures before its buffer is full. If you want to take action pictures like wildlife, sports, or even candid street photography, you need a camera that is ready to take a picture as soon as you can frame it and able to keep taking pictures until the action has stopped.
KenRockwell.com -- lots of good info on Nikon DSLRs and lenses. Be warned though, this guy has a very heavy DSLR bias and thinks you are a chump for getting a "prosumer" camera. If you are a pro, you probably are. If you are a hobbyist, maybe not, depending on your needs and budget. (If you only have about $1000 to spend and you want 10x zoom, you ain't getting a DSLR.)
I'm not sure if I'd go that far. His real point seems to be that the "prosumer" point-and-shoot cameras are a foolish choice when compared to either low-end DSLR's or less expensive point-and-shoot cameras. There really is a qualitative difference between even the cheapest DSLR's and the most expensive point-and-shoot cameras, so there's a compelling reason to take that step up if you demand great picture quality. OTOH, the difference between the "prosumer" point-and-shoots and the ones that cost half as much isn't nearly as big, so it might make more sense to buy a smaller, lighter, cheaper point-and-shoot if you're not super-serious about picture quality.
Now why would I want to get my pay based on seniority rather than performance?
Why do you think that forming a union would require that pay be based on seniority rather than performance? While it's common for unions to adopt that stance, unions that represent highly talented people, such as actors and professional athletes, generally have only a minimum rate and allow their members to negotiate for whatever they can get above that. There's no reason that the gaming industry union couldn't do things the same way.
A paper trail is not a sure thing....particularly a *machine-printed* paper trail.
That depends on how the paper trail is managed. A straightforward system that would make the paper trail reliable is to print out the ballot and give the printed version to the voter. The voter would then be able to check that the printed ballot accurately reflected his choices before finalizing them. After hitting the "Yes, my vote is correct" button, the voter would then take the printed ballot and drop it into a ballot box the same way that he does with a paper ballot today. Manual validation provides a safeguard against fraudulent misprinting of the paper ballot, and the existence of the paper ballot provides a check on computerized fraud.
The question, though, is how to make a simple system that deals with the inherent complexity of the election system. It's not as though an election in the US deals with just one question. The most recent ballot I voted on had no fewer than 27 issues: elections for President, US Senator and Representative, State Senator and Assemblyman, 5 judgeships, and 17 ballot initiatives. The initiatives were simple yes/no questions, but most of the other positions had more than two choices, as well as the possibility of write-in candidates. The complexity of dealing with that many differen elections is inherent to the election process, so you can't just do away with it by fiat. Any proposed voting system has to be able to tally that many elections within a reasonable length of time. That almost certainly means using some kind of electronic help in reading and tallying the votes.
It most clearly says what Novell's board of directors thought they were agreeing to... but is that what they actually got themselves into?
That actually matters less than you realize. SCO's suit is for "Slander of Title", which means that SCO is claiming that Novell maliciously made false statements that caused SCO harm by calling their ownership of System V into question. SCO must prove every part of that claim: that Novell made false statements, that the statements were made maliciously, and that the statements did SCO damage. Novell doesn't have to prove that they actually do own the copyrights (i.e. prove that their claims were true), though this is certainly good evidence of that point. Novell only has to show that their claims were not made maliciously, and an honest, well-founded belief that they still owned the copyrights is sufficient to do so.
The problem with HD storage is that all hard drives have a 100% failure rate.
Sure, but it you look at it hard enough the same thing is true of any medium. Some of them have much longer lifespans than others- good ink on archival quality paper has a much longer lifespan than newsprint, for instance- but nothing is completely immune to the power of entropy. Hard drives at least have the advantage that it's trivial to make perfect copies of their contents.
I'm still not sure that I buy it. Many wild bee species today have lifestyles that seem well suite for riding out disasters. They are only active for a small part of the year- when the plants they specialize in are flowering- and spend the rest of the year dormant. That seems like behavior that would be useful in the event of a nuclear-winter like disaster. They're also well known for storing large quantities of food in the form of honey and pollen, which also seems like an ideal adaptation for surviving.
Rather than manually editing your/etc/yum.conf to point to FC3, it might be better just to download the fedora-release package from FC3, update that using RPM, and then proceed to update yum and then the whole system.
Sadly, Apple has made a mistake by failing to include a card reader. The iPod Photo is stuck in the paradigm of digital music, in which your computer is the center and the iPod is just a way of making the music more portable. For photos, though, I think of my camera as being the central point, not my computer. Being able to download directly from my camera's memory cards to the iPod would massively increase the number of pictures I could take without needing to go back to my desktop or haul around a laptop. Without that, this is just a minor improvement, but with the ability to download straight from memory cards it would be a major step forward.
"Overall," says Dr. Rovner, "we are confident that polygraph tests have a 96% accuracy rate when done properly."
If that is true, then if you have 1 spy and 49 honest people, this polygraph will likely falsely accuse two honest people as being spies.
I can see two problems with this comment. One is that he doesn't state what that "96% accuracy" rate really means. That could mean that it's able to catch 96% of lies, which would be pretty good, or it could mean that it incorrectly calls the truth a lie 4% of the time. If it means catching 96% of all lies, that could mean that it catches 96% of false statements, or that it can figure out when somebody is lying 96% of the time by asking the same question multiple times and carefully comparing the results. How useful it is depends a great deal on which definition he's using; a system that can flag 96% of false statements is obviously much more useful than one that calls 4% of honest people liars.
Another possible problem is whether they've done tests on enough different populations to know whether those results are generally useful. It's entirely possible that the accuracy rate when testing college students is 96%, but that the rate when testing spies, con-men, and similar trained liars is much lower. If they've validated their method on a different population than the one they're using it on, there's a very good chance that their expected accuracy rate, however defined, is grossly misleading.
Don't you know anything? You have to make your own tinfoil hat. The commercial tinfoil hat makers have been suborned to make theirs defective. Commercial models act as antennas for, rather than deflectors of, the CIA's mind control beams. Trust no one!
If you read the article closely, you'll see that it doesn't say that this is the first one with a full 35mm sensor; it's the first one that can "effectively reproduce the image quality of 35mm film". Of course that isn't necessarily true, either. There's considerable argument about film vs. digital quality. The 1-Ds Mark II still may not be able to match the resolution of the best film, but DSLR's absolutely crush film for signal to noise. Depending on which of those matters more to you, digital may have surpassed film some time ago, or it may not have done so yet. Of course that assumes that you're restricting yourself to 35mm format digital cameras. There have been digital backs for medium and even large format cameras that have resolution far surpassing 35mm film for a fair while.
By the way, given the rapidly falling prices on 512 MB flash memory cards in Compact Flash, SD and xD formats, I wouldn't be surprised that higher-end point and shoot digital still cameras will offer DNG support (it's obvious that digital SLR's will get DNG support).
Most of the higher end point and shoot digitals, and even some of the less expensive ones, already offer raw formats, so it's no stretch at all to predict that they'll switch to DNG.
If you're looking for a command-line tool, jhead might read Exif tags.
Jhead does read EXIF tags. This can be very useful if you want to analyze something about your photo usage. As an example, I've been considering getting a fast prime lens for available light photography in a range that I currently have covered by a slower zoom lens, but I wasn't sure whether to get 20 mm, 24 mm, or 28 mm. I used jhead to extract the focal length for every picture I've taken and found that I use 24 mm a lot more than either 20 mm or 28 mm. Now I know which lens to buy.
Raw files are reasonably practical on many DSLRs. With a fast memory card, my Nikon D70 can shoot raw images at 3 fps for the first 4-5 shots and 1 fps after that until the card is full. The compression is fast enough that write speed is the limiting factor, so uncompressed files aren't even an option. The professional grade Canon 1D Mark II, which is aimed at sports and wildlife photographers, has a large buffer that lets it shoot up to 20 files in raw format at 8 fps, though the sustained speed with the buffer full is only about one frame every 2 seconds.
dcraw has provided some code I imagine for deconstructing the file formats - not for creating the "camera profiles" needed to develop linear image data into beautiful non-linear color images that meet users expectations.
Sure, but what does that have to do with DNG? I haven't seen any indication that DNG is going to add camera profile information into the raw data format. It's just going to replace the existing multitude of raw formats with a single unified format. I certainly doubt that Adobe's raw to DNG converter is going to include their carefully created camera profiles; that would be giving away the store. Lack of those profiles may mean that conversions for new models won't be quite as impressive as for ones that Adobe has been able to profile, but it will still be a big improvement over not being able to convert them at all.
Adobe can save money with the adoption of DNG - but more money will be saved by smaller developers who cannot do the ongoing reverse engineering that Adobe does to support new RAW formats.
It's unclear to me how much effort Adobe is actually spending on the problem. My understanding is that just about everyone uses Dave Coffin's dcraw program to one extent or another. (Mr. Coffin reports Adobe as a company that uses at least part of his program. I don't know if they use his interpolation, but I'd guess that they use his decoding work.) Of course that program is available under a very liberal license, so all the little guys can make use of it, too. A bigger issue, IMO, is that if a camera uses DNG then it will be supported as soon as it's released, rather than requiring a software update before people can read its raw files.
It depends. A key thing to remember is that most cameras have quite limited processing power, so they can't use the most sophisticated raw conversion algorithms. That's not much of an issue when processing at home, but even on a quite fast computer the better algorithms take a few seconds per picture. A bigger issue is that raw conversion requires you to stop and think about things like white balance, contrast vs. dynamic range, color saturation, etc. Frequently you'll want to try tweaking some or all of those, which can add quite a bit of time to photo processing, and then the picture will need some additional tweaking, like sharpening. Of course many photographers would be doing quite a bit of tweaking of their photographs anyway, so much of that effort isn't unique to the raw files.
My personal feeling (and one that I know that many others agree with) is that most pictures aren't worth it. I know when I start taking pictures that only a small fraction are going to be good enough to make my personal portfolio. The ones that don't make it usually suffer from things like camera shake, subject motion, poor light quality, bad framing, or (most frequently) poor choice of subject, which can't be fixed using raw processing. That's why many cameras now have a "RAW+JPEG" mode. There's a JPEG that lets you examine the pictures without having to expend time and energy converting the raw files for the majority of pictures that don't deserve the effort, and a raw file for the few that do.
Of course, since the nature of the data will be very different depending on what kind of image sensor you have
Which shouldn't be a worry for the vast majority of cameras that use square pixels in a RGB Bayer array. I'd expect that it would be pretty easy to deal with other square pixel Bayer arrays, like Sony's RGBE and the occasional CMY. Non-square arrays, like the ones used in some Nikon pro-series SLRs, and non-Bayer setups, like Foveon's X3 system, might be harder, but I'd expect that they could be accomodated without great difficulty.
Slashdot Mirror
It's not the "Rose Bowl Parade", damnit. It's the "Tournament of Roses Parade". The parade was started decades before the football game became a standard part of Pasadena's New Years Day celebration. It would be far more accurate to call the game the Tournament of Roses football game than the other way around.
But people can only vote with their dollars if there's a range of goods and services to choose from. If there's a monopoly or oligopoly, there just aren't enough choices to make voting with ones wallet a reasonable approach. You can view this as an economic counterpart to complaints about the lack of choices in our current two party political system. If major parties (or businesses) can squeeze out most of their competitors, then voters (or consumers) aren't necessarily presented with any acceptable choices.
This is particularly important in an area like broadcast media. Whether or not there are good reasons for it, our current system places a sharp limit on the total number of broadcast outlets in a given area. That makes the broadcast market a ripe target for monopolies or oligopolies to take over. The same government that restricts the number of broadcast outlets must ensure that limited number of outlets aren't concentrated into too few hands, or the public won't be able to vote with their dollars.
An even better thing that DSLR's let you do is to take pictures without needing the flash in the first place. With an ISO speed of 800 or 1600 and a wide aperture lens, you can take pictures in ordinary indoor light without needing a flash. Not all of them will be razor sharp- especially if you're photographing a rampaging toddler or some other rapidly moving thing- but with good technique most of them can be.
Shooting in available light also helps to get the people in your pictures behave more naturally. Most people seem to turn into either hams or wallflowers when they know they're being photographed, and zapping them with a flash is a great way of reminding them that their pictures are being taken. The hams can be fun to photograph, but I find that the best pictures are ones of people who are being themselves rather than thinking about how they'll look in a photograph.
You obviously must be one of those "standard" camera guys. DSLR people know that available light photography rules. Long live 800 ISO at f/2!
Do bear in mind, though, that most of the "affordable" DSLR's these days use sub-35mm sized sensors, which changes the use of most lenses. The Canon D20 and Digital Rebel, for instance, have a 1.6x "focal length mulitplier". That means that your favorite 28mm lens will have a field of view rather close to that of a 50mm lens on your film camera, while the 50mm lens that you dislike will have a field of view much closer to that of an 85mm lens on your film camera. This is wonderful for portrait shooters, since it means that they can use cheap 50/1.8 and 85/1.8 lenses. OTOH, it hurts scenic and street shooters, since it makes their 20mm to 24mm wide angle lenses just barely wide of normal.
Another important point:
4) Responsiveness. If most of your pictures are landscapes or people posing for the camera, it's fine if it takes your camera few seconds to turn on, zoom, or focus, if you have to go through menus to set important controls, of if your camera can only take a few pictures before its buffer is full. If you want to take action pictures like wildlife, sports, or even candid street photography, you need a camera that is ready to take a picture as soon as you can frame it and able to keep taking pictures until the action has stopped.
I'm not sure if I'd go that far. His real point seems to be that the "prosumer" point-and-shoot cameras are a foolish choice when compared to either low-end DSLR's or less expensive point-and-shoot cameras. There really is a qualitative difference between even the cheapest DSLR's and the most expensive point-and-shoot cameras, so there's a compelling reason to take that step up if you demand great picture quality. OTOH, the difference between the "prosumer" point-and-shoots and the ones that cost half as much isn't nearly as big, so it might make more sense to buy a smaller, lighter, cheaper point-and-shoot if you're not super-serious about picture quality.
Why do you think that forming a union would require that pay be based on seniority rather than performance? While it's common for unions to adopt that stance, unions that represent highly talented people, such as actors and professional athletes, generally have only a minimum rate and allow their members to negotiate for whatever they can get above that. There's no reason that the gaming industry union couldn't do things the same way.
That depends on how the paper trail is managed. A straightforward system that would make the paper trail reliable is to print out the ballot and give the printed version to the voter. The voter would then be able to check that the printed ballot accurately reflected his choices before finalizing them. After hitting the "Yes, my vote is correct" button, the voter would then take the printed ballot and drop it into a ballot box the same way that he does with a paper ballot today. Manual validation provides a safeguard against fraudulent misprinting of the paper ballot, and the existence of the paper ballot provides a check on computerized fraud.
The question, though, is how to make a simple system that deals with the inherent complexity of the election system. It's not as though an election in the US deals with just one question. The most recent ballot I voted on had no fewer than 27 issues: elections for President, US Senator and Representative, State Senator and Assemblyman, 5 judgeships, and 17 ballot initiatives. The initiatives were simple yes/no questions, but most of the other positions had more than two choices, as well as the possibility of write-in candidates. The complexity of dealing with that many differen elections is inherent to the election process, so you can't just do away with it by fiat. Any proposed voting system has to be able to tally that many elections within a reasonable length of time. That almost certainly means using some kind of electronic help in reading and tallying the votes.
That actually matters less than you realize. SCO's suit is for "Slander of Title", which means that SCO is claiming that Novell maliciously made false statements that caused SCO harm by calling their ownership of System V into question. SCO must prove every part of that claim: that Novell made false statements, that the statements were made maliciously, and that the statements did SCO damage. Novell doesn't have to prove that they actually do own the copyrights (i.e. prove that their claims were true), though this is certainly good evidence of that point. Novell only has to show that their claims were not made maliciously, and an honest, well-founded belief that they still owned the copyrights is sufficient to do so.
Sure, but it you look at it hard enough the same thing is true of any medium. Some of them have much longer lifespans than others- good ink on archival quality paper has a much longer lifespan than newsprint, for instance- but nothing is completely immune to the power of entropy. Hard drives at least have the advantage that it's trivial to make perfect copies of their contents.
I'm still not sure that I buy it. Many wild bee species today have lifestyles that seem well suite for riding out disasters. They are only active for a small part of the year- when the plants they specialize in are flowering- and spend the rest of the year dormant. That seems like behavior that would be useful in the event of a nuclear-winter like disaster. They're also well known for storing large quantities of food in the form of honey and pollen, which also seems like an ideal adaptation for surviving.
Rather than manually editing your /etc/yum.conf to point to FC3, it might be better just to download the fedora-release package from FC3, update that using RPM, and then proceed to update yum and then the whole system.
Sadly, Apple has made a mistake by failing to include a card reader. The iPod Photo is stuck in the paradigm of digital music, in which your computer is the center and the iPod is just a way of making the music more portable. For photos, though, I think of my camera as being the central point, not my computer. Being able to download directly from my camera's memory cards to the iPod would massively increase the number of pictures I could take without needing to go back to my desktop or haul around a laptop. Without that, this is just a minor improvement, but with the ability to download straight from memory cards it would be a major step forward.
I can see two problems with this comment. One is that he doesn't state what that "96% accuracy" rate really means. That could mean that it's able to catch 96% of lies, which would be pretty good, or it could mean that it incorrectly calls the truth a lie 4% of the time. If it means catching 96% of all lies, that could mean that it catches 96% of false statements, or that it can figure out when somebody is lying 96% of the time by asking the same question multiple times and carefully comparing the results. How useful it is depends a great deal on which definition he's using; a system that can flag 96% of false statements is obviously much more useful than one that calls 4% of honest people liars.
Another possible problem is whether they've done tests on enough different populations to know whether those results are generally useful. It's entirely possible that the accuracy rate when testing college students is 96%, but that the rate when testing spies, con-men, and similar trained liars is much lower. If they've validated their method on a different population than the one they're using it on, there's a very good chance that their expected accuracy rate, however defined, is grossly misleading.
Don't you know anything? You have to make your own tinfoil hat. The commercial tinfoil hat makers have been suborned to make theirs defective. Commercial models act as antennas for, rather than deflectors of, the CIA's mind control beams. Trust no one!
If you read the article closely, you'll see that it doesn't say that this is the first one with a full 35mm sensor; it's the first one that can "effectively reproduce the image quality of 35mm film". Of course that isn't necessarily true, either. There's considerable argument about film vs. digital quality. The 1-Ds Mark II still may not be able to match the resolution of the best film, but DSLR's absolutely crush film for signal to noise. Depending on which of those matters more to you, digital may have surpassed film some time ago, or it may not have done so yet. Of course that assumes that you're restricting yourself to 35mm format digital cameras. There have been digital backs for medium and even large format cameras that have resolution far surpassing 35mm film for a fair while.
Most of the higher end point and shoot digitals, and even some of the less expensive ones, already offer raw formats, so it's no stretch at all to predict that they'll switch to DNG.
Jhead does read EXIF tags. This can be very useful if you want to analyze something about your photo usage. As an example, I've been considering getting a fast prime lens for available light photography in a range that I currently have covered by a slower zoom lens, but I wasn't sure whether to get 20 mm, 24 mm, or 28 mm. I used jhead to extract the focal length for every picture I've taken and found that I use 24 mm a lot more than either 20 mm or 28 mm. Now I know which lens to buy.
Raw files are reasonably practical on many DSLRs. With a fast memory card, my Nikon D70 can shoot raw images at 3 fps for the first 4-5 shots and 1 fps after that until the card is full. The compression is fast enough that write speed is the limiting factor, so uncompressed files aren't even an option. The professional grade Canon 1D Mark II, which is aimed at sports and wildlife photographers, has a large buffer that lets it shoot up to 20 files in raw format at 8 fps, though the sustained speed with the buffer full is only about one frame every 2 seconds.
Sure, but what does that have to do with DNG? I haven't seen any indication that DNG is going to add camera profile information into the raw data format. It's just going to replace the existing multitude of raw formats with a single unified format. I certainly doubt that Adobe's raw to DNG converter is going to include their carefully created camera profiles; that would be giving away the store. Lack of those profiles may mean that conversions for new models won't be quite as impressive as for ones that Adobe has been able to profile, but it will still be a big improvement over not being able to convert them at all.
It's unclear to me how much effort Adobe is actually spending on the problem. My understanding is that just about everyone uses Dave Coffin's dcraw program to one extent or another. (Mr. Coffin reports Adobe as a company that uses at least part of his program. I don't know if they use his interpolation, but I'd guess that they use his decoding work.) Of course that program is available under a very liberal license, so all the little guys can make use of it, too. A bigger issue, IMO, is that if a camera uses DNG then it will be supported as soon as it's released, rather than requiring a software update before people can read its raw files.
It depends. A key thing to remember is that most cameras have quite limited processing power, so they can't use the most sophisticated raw conversion algorithms. That's not much of an issue when processing at home, but even on a quite fast computer the better algorithms take a few seconds per picture. A bigger issue is that raw conversion requires you to stop and think about things like white balance, contrast vs. dynamic range, color saturation, etc. Frequently you'll want to try tweaking some or all of those, which can add quite a bit of time to photo processing, and then the picture will need some additional tweaking, like sharpening. Of course many photographers would be doing quite a bit of tweaking of their photographs anyway, so much of that effort isn't unique to the raw files.
My personal feeling (and one that I know that many others agree with) is that most pictures aren't worth it. I know when I start taking pictures that only a small fraction are going to be good enough to make my personal portfolio. The ones that don't make it usually suffer from things like camera shake, subject motion, poor light quality, bad framing, or (most frequently) poor choice of subject, which can't be fixed using raw processing. That's why many cameras now have a "RAW+JPEG" mode. There's a JPEG that lets you examine the pictures without having to expend time and energy converting the raw files for the majority of pictures that don't deserve the effort, and a raw file for the few that do.
Which shouldn't be a worry for the vast majority of cameras that use square pixels in a RGB Bayer array. I'd expect that it would be pretty easy to deal with other square pixel Bayer arrays, like Sony's RGBE and the occasional CMY. Non-square arrays, like the ones used in some Nikon pro-series SLRs, and non-Bayer setups, like Foveon's X3 system, might be harder, but I'd expect that they could be accomodated without great difficulty.