Canon does make some awesome lenses, but even some of their L-lenses look somewhat lacking when used on their high-resolution sensors.
My father has a 70-200 f/2.8L. It actually shows pretty low contrast and a "hazy" look until you stop it down to f/4 or more, especially at the long end. The new 70-200 mk2 is much better.
What's improving is in part processing technology. At some point you run into the brick wall of Planck's constant: the fluctuations in a counting experiment counting N photons can be no less than sqrt(N) due to shot noise.
Not quite. It's the same PER PIXEL: if you crop a 10MP area out of this thing then it'll be roughly as noisy as a 10MP 1/1.8" sensor. (Did you use crop factor 1.6 or 1.3 in your math? This sensor is APS-H, 1.3x, not the commonly-used 1.6x APS-C.) But if you print at any given output size, the pixels from the higher-resolution sensor will be smaller and thus whatever noise is present at a pixel level will be less intrusive.
For the math geeks: the real thing you should look at is the signal-to-noise ratio at any given spatial frequency (in cycles per image height).
If a difference of a few hundred grams in weight is such a big deal then you should work out more.
I have one of the heaviest netbooks out there (Asus eeepc 1000HE), and it's plenty light, and I'm a small and relatively weak guy. Any lighter and I'd break it.
A netbook can do anything an ipad can do. It's cheaper. It's just about as portable.
And it is YOURS. You have root on it and can do whatever you bloody well please on it. It's a complete computer, with a modern multitasking O/S and the ability to do anything your desktop computer can do -- except slower.
The decision says that simply making the recording is not a tort or crime per se, but if you intend to use the recording to commit a tort or crime, then making the recording is itself prohibited.
i.e. I can record you admitting that you're having an affair and send the recording to your spouse, but if I intend to use the recording to blackmail you, then the recording is itself a crime.
If you replace "educated guess" with "first-order approximation", it sounds a lot better -- and, in fact, this happens in the sciences all the time. But that's just what a first-order approximation is: it's a guess (based on the first term of a series) that is educated (based on some belief that the subsequent terms are smaller).
The lattice QCD people, at least, are porting their code to CUDA just as fast as they can. The bottleneck right now is that there's no good way to get multiple GPU's to communicate (quickly). So, for the largest problems (simulating a 64x64x64x192 lattice), you still need a conventional supercomputer (like Ranger, the one in the article here), because it's just too huge to put on a single GPU and multi-GPU doesn't scale well.
But for smaller problems (like a 24x24x24x64 lattice), GPU's will be great, and people are developing this capability as fast as they can.
Whether someone wants to charge money for something has no bearing on whether someone else is allowed to give it away for free. If I make a business of selling prints of the Mona Lisa outside the Louvre, I can't complain when someone else starts making them and handing them to tourists for free.
This guy's actions might have been technically illegal based on other factors, but what Blizzard charges money for isn't the issue. I can violate the Ubuntu Foundation's copyright just as easily and in the same ways as I can violate the RIAA's -- copy their stuff without a license to do so (or, in the Ubuntu case, without following the terms of the license distributed with the software).
The US doesn't have that capability -- after all, have they gotten bin Laden yet? It took us two years to capture Saddam Hussein, a notorious man in a country we occupied with a hundred thousand soldiers.
It's probably also a lot lower since much of the price of a supercomputer is the expensive Infiniband interconnect architecture. I imagine the way they're doing this is heavily compute-limited, not communication-limited, so you could do it on commodity desktops in spare time.
Actually, you could probably do it on GPU's, come to think about it.
Ah, yes, you're correct: there are challenges where a bunch of compute time would be welcome, but that time is available, as you point out. (If MILC can get a billion hours over eight years to simulate quarks, someone can get a few million to do cancer research).
There's a movement in health research now geared at extending what they call "healthspan" rather than just "lifespan" -- not "how long does this dude keep breathing", but "how long can we keep this dude active and happy"?
Turns out that many of the things that make people live longer also make their late years healthier. My grandfather is 94 and still travels the world with his girlfriend (a spry young 75, but he'll never see her again now that she's taken up Farmville). He got prostate cancer a few years ago (and colon cancer a few decades ago), received aggressive treatment for it, and is now cancer-free and healthy.
Old does not *have* to mean feeble. Sometimes it does, of course, and that's bad; this is why we should look at healthspan rather than lifespan.
He didn't need compute time; he was giving a talk to a bunch of physicists about something that a) they don't know much about, but would find interesting, and b) they could go work on if they get bored with quarks, and would probably bring a fresh perspective to since they have different skills.
(And, in a sense, he *did* talk to the TeraGrid folks, since some of the largest TeraGrid users were there.)
35 years is about 300k core-hours, a standard measure of computing resources. This is a big pile of computer time, but is not unreasonable.
So how much does this cost?
A typical supercomputer, Ranger, cost $59 million to build and operate for four years. It's got about 60k cores, so $59 million delivers 240k core-years; they used 35 core-years to do this computation. Doing the division, you get $9000 of computer time -- not all that bad. Plugging in the cost numbers for another production supercomputer, Kraken, gives a slightly lower cost.
Actually, I was at a high-performance physics computing conference this summer in which a genetic oncologist talked about some of the computational challenges in cancer genomics and said, basically, "There's lots of room over here if you physics folks want something else to chew on." It won't be cured by brute-force computing alone, but there are certainly computational challenges where a few million core-hours would be welcome.
If you insist that it's your body and your choice, then, well, make that choice. Saying that organ donation is the default is a whole lot different than mandating it.
The current law states that if you die without a will, your family gets your stuff. Is this a violation of your rights? Of course not -- it's people trying to put your stuff to best use once you're dead. Don't like it, write a will.
Don't like the fact that, if you die, some sick person will get your kidneys? Write a living will.
As I understand it, because of the way film grain works, film has very good definition for fine high-contrast detail (which is somewhat above the Nyquist frequency of even the best FX sensors), but when you start talking about lower-contrast detail, the grain structure of the film obscures the detail.
So if I'm trying to read black type on white paper at a very long distance, I want film, but if I'm trying to make out low-contrast feather detail on a distant bird, top-end digital is better (if more expensive).
Slashdot Mirror
Canon does make some awesome lenses, but even some of their L-lenses look somewhat lacking when used on their high-resolution sensors.
My father has a 70-200 f/2.8L. It actually shows pretty low contrast and a "hazy" look until you stop it down to f/4 or more, especially at the long end. The new 70-200 mk2 is much better.
What's improving is in part processing technology. At some point you run into the brick wall of Planck's constant: the fluctuations in a counting experiment counting N photons can be no less than sqrt(N) due to shot noise.
Not quite. It's the same PER PIXEL: if you crop a 10MP area out of this thing then it'll be roughly as noisy as a 10MP 1/1.8" sensor. (Did you use crop factor 1.6 or 1.3 in your math? This sensor is APS-H, 1.3x, not the commonly-used 1.6x APS-C.) But if you print at any given output size, the pixels from the higher-resolution sensor will be smaller and thus whatever noise is present at a pixel level will be less intrusive.
For the math geeks: the real thing you should look at is the signal-to-noise ratio at any given spatial frequency (in cycles per image height).
Explained in a picture:
http://imgs.xkcd.com/comics/interblag.png
Then get a netbook and a cellular modem. Same thing as an iPad, except you can do other stuff with it.
If a difference of a few hundred grams in weight is such a big deal then you should work out more.
I have one of the heaviest netbooks out there (Asus eeepc 1000HE), and it's plenty light, and I'm a small and relatively weak guy. Any lighter and I'd break it.
This.
A netbook can do anything an ipad can do. It's cheaper. It's just about as portable.
And it is YOURS. You have root on it and can do whatever you bloody well please on it. It's a complete computer, with a modern multitasking O/S and the ability to do anything your desktop computer can do -- except slower.
In Illinois, this just means you don't vote /often/ enough.
That's basically what I was saying, but I didn't want to trot out the scary Latin term. :)
The decision says that simply making the recording is not a tort or crime per se, but if you intend to use the recording to commit a tort or crime, then making the recording is itself prohibited.
i.e. I can record you admitting that you're having an affair and send the recording to your spouse, but if I intend to use the recording to blackmail you, then the recording is itself a crime.
If you replace "educated guess" with "first-order approximation", it sounds a lot better -- and, in fact, this happens in the sciences all the time. But that's just what a first-order approximation is: it's a guess (based on the first term of a series) that is educated (based on some belief that the subsequent terms are smaller).
The lattice QCD people, at least, are porting their code to CUDA just as fast as they can. The bottleneck right now is that there's no good way to get multiple GPU's to communicate (quickly). So, for the largest problems (simulating a 64x64x64x192 lattice), you still need a conventional supercomputer (like Ranger, the one in the article here), because it's just too huge to put on a single GPU and multi-GPU doesn't scale well.
But for smaller problems (like a 24x24x24x64 lattice), GPU's will be great, and people are developing this capability as fast as they can.
... Terran-shaped pylons! That way you can disguise one in their base until you're ready and then bam! warp in dudes.
People distribute things all the time that rely on the recipient already having some piece of copyrighted data. RiffTrax, or Windows software.
Whether someone wants to charge money for something has no bearing on whether someone else is allowed to give it away for free. If I make a business of selling prints of the Mona Lisa outside the Louvre, I can't complain when someone else starts making them and handing them to tourists for free.
This guy's actions might have been technically illegal based on other factors, but what Blizzard charges money for isn't the issue. I can violate the Ubuntu Foundation's copyright just as easily and in the same ways as I can violate the RIAA's -- copy their stuff without a license to do so (or, in the Ubuntu case, without following the terms of the license distributed with the software).
Ah, now I understand. It took a car analogy; thanks.
The US doesn't have that capability -- after all, have they gotten bin Laden yet? It took us two years to capture Saddam Hussein, a notorious man in a country we occupied with a hundred thousand soldiers.
It's probably also a lot lower since much of the price of a supercomputer is the expensive Infiniband interconnect architecture. I imagine the way they're doing this is heavily compute-limited, not communication-limited, so you could do it on commodity desktops in spare time.
Actually, you could probably do it on GPU's, come to think about it.
Ah, yes, you're correct: there are challenges where a bunch of compute time would be welcome, but that time is available, as you point out. (If MILC can get a billion hours over eight years to simulate quarks, someone can get a few million to do cancer research).
There's a movement in health research now geared at extending what they call "healthspan" rather than just "lifespan" -- not "how long does this dude keep breathing", but "how long can we keep this dude active and happy"?
Turns out that many of the things that make people live longer also make their late years healthier. My grandfather is 94 and still travels the world with his girlfriend (a spry young 75, but he'll never see her again now that she's taken up Farmville). He got prostate cancer a few years ago (and colon cancer a few decades ago), received aggressive treatment for it, and is now cancer-free and healthy.
Old does not *have* to mean feeble. Sometimes it does, of course, and that's bad; this is why we should look at healthspan rather than lifespan.
He didn't need compute time; he was giving a talk to a bunch of physicists about something that a) they don't know much about, but would find interesting, and b) they could go work on if they get bored with quarks, and would probably bring a fresh perspective to since they have different skills.
(And, in a sense, he *did* talk to the TeraGrid folks, since some of the largest TeraGrid users were there.)
35 years is about 300k core-hours, a standard measure of computing resources. This is a big pile of computer time, but is not unreasonable.
So how much does this cost?
A typical supercomputer, Ranger, cost $59 million to build and operate for four years. It's got about 60k cores, so $59 million delivers 240k core-years; they used 35 core-years to do this computation. Doing the division, you get $9000 of computer time -- not all that bad. Plugging in the cost numbers for another production supercomputer, Kraken, gives a slightly lower cost.
Actually, I was at a high-performance physics computing conference this summer in which a genetic oncologist talked about some of the computational challenges in cancer genomics and said, basically, "There's lots of room over here if you physics folks want something else to chew on." It won't be cured by brute-force computing alone, but there are certainly computational challenges where a few million core-hours would be welcome.
If you insist that it's your body and your choice, then, well, make that choice. Saying that organ donation is the default is a whole lot different than mandating it.
The current law states that if you die without a will, your family gets your stuff. Is this a violation of your rights? Of course not -- it's people trying to put your stuff to best use once you're dead. Don't like it, write a will.
Don't like the fact that, if you die, some sick person will get your kidneys? Write a living will.
The film-digital comparison is a tricky one.
As I understand it, because of the way film grain works, film has very good definition for fine high-contrast detail (which is somewhat above the Nyquist frequency of even the best FX sensors), but when you start talking about lower-contrast detail, the grain structure of the film obscures the detail.
So if I'm trying to read black type on white paper at a very long distance, I want film, but if I'm trying to make out low-contrast feather detail on a distant bird, top-end digital is better (if more expensive).