I'm surprised your linked page didn't mention Iran, which suddenly opened its nuclear program to international inspection just a few days after watching Saddam get dragged up out of his hole in the ground.
Although.png can store 16 bits per channel, none of that is overrange, so it can't be used for this purpose.
Whether or not any information is overrange just depends on the exposure. That's why it's best to archive the original image before doing anything to it, even though it might look better in print if you boost the contrast (compress the dynamic range.) I won't argue that it might be more *convenient* to have specific "overrange" information in the file; this just records your decision of how much of the range to use for output. But ultimately a 16 bit floating-point format can't hold more information than a 16 bit fixed-point format.
As for the extra bits, I didn't mean to imply they were never useful, only that they are overkill for archiving originals. This is because they don't come into play until you start editing the picture. Once editing starts of course all bets are off; for instance you could (for some strange reason) decrease the image's contrast drastically, so you're only using between 50% and 55% exposure. Then you'd be using those previously insignificant bits. If you were using an 8 bit format everything that used to be from 0-255 would now be between about 128 and 133, so you'd have lost a lot. But in an original, unedited photo (which is what you'd normally have a lot of in an archive) I still say everything past 10 bits or so is noise.
And yes, capturing the grain is important. It's part of the character of the image....
No, if you really want to archive your images, preserving as much of the information that's on the film as possible, you'd need to scan them at the highest resolution your scanning device could manage and store them in either RAW or a floating-point high dynamic range format.
Grain may be considered part of the image artistically, but it's in principle no different than jpeg compression artifacts - it doesn't come from the world, it's a side effect of the mechanism.
As to the resolution for storing film, certainly it takes a high resolution to store the image data PLUS the grain. The more relevant question for most people is at what resolution an image taken with a digital camera is enough to look as good as a 35mm frame when blown up and printed. No way a 35mm frame produces the equivalent of 6000x4000 color pixels - at that resolution the grain will be very evident. Still, that's only about 25% as many pixels as the poster I was responding to suggested - already a 75% file size reduction!
The.png format, by the way, can store 16 bits per channel, which is overkill. (Too bad gimp is restricted to 8 bits per channel though). 32 bits per channel is silly. All those extra bits (beyond 10 or so) are insignificant, in the sense that neither film nor digital is that accurate; those extra bits are just noise. Some actual information may be pushed down into those bits if you compress part of the range during image processing, but when you are mass archiving unedited photos, there's no point in storing all those insignificant bits.
Then again, I suppose that unless you keep a generator to run your computers in case of a power outage, it wouldn't really matter if the internet is not working while you're power is out.
Exactly! This is where you realize all that typing was for nothing, hit the back button, and start thinking of some other comment to make.
It's the principle that matters though!
Or you can drum up some sort of wacky conclusion and submit it anyways.
I guess our station is a mix. They play NPR stuff twice a day, but the rest of the time it's shows like "native american news" and "democracy now" (and believe me, Monsanto and "democracy now" go together like Microsoft and Stallman).
So they discounted the viruses and email crap that require some user to click the attachment called 'Im a virus, click me now'.
They didn't ignore JUST that. It sounds like they ignored every virus and worm that spread themselves automatically, even if due to an rpc bug or what have you.
You, know, those hundreds of default.ida and scripts/..%252f.. requests you get every day? According to these guys the cracked machines behind those requests don't exist, or at least don't count.
Nevertheless I'm going to take a closer look and see how I can secure my linux boxes better. I'm surprised linux fared so badly, because many of the services running on linux (apache, sshd, ntp) are the very same ones running on the bsd boxes which did better.
Pick a resolution for colour negative film? 10000 x 8000 sound reasonable? That's about 2.4E+8 bytes per picture, or 8.6E+9 bytes per roll of film, equivalent.
I don't think so. At that resolution you're capturing every grain in the film, at least if it's 35 mm film. That grain is not really part of the image, it's an artifact. Kodak states "(2048 x 3072 pixels) captures all the image data 35 mm film has to offer." There, we reduced file size by a factor of 12. Now, I hope you're not storing uncompressed tiffs? They'd be around half the size (depending on image) as compressed.png. That brings us to a 96% reduction from your figure. And that's without touching lossy compression - which I doubt you would touch, even though you don't mind scanning and storing away all the grain of film.
There's no objective way to exactly compare film/digital resolutions, but your estimate is certainly biased towards film.
Give me a break, the article had a whole paragraph on existing automatic prescription machines:
Current automatic diagnostic technologies are expensive, fragile and error-prone. Because they rely on a patient looking at electronically generated images a few inches away from his or her face, they can lead to incorrect diagnoses. Plus, highly skilled people are required to operate these machines.
I know it's kind of flaky, but I like "American Idol" because I think it is an honest talent show. True, they consider stage presence and showmanship as well as singing, but the people who do well really are talented and entertaining, and I don't think they're using any Autotune. In fact, one of the judges on the show (the black guy formerly from Rush) has been talking in interviews about how the music industry is sagging because it's so out of touch, and how many of the people who do well on the show probably wouldn't get to sign a deal otherwise, like the fat guy from last year.
Believe it or not, Autotune [antarestech.com] already exists! This product is the sole reason people like Britney Spears and Justin Timberlake can be called "singers"
...it's VERY useful for minor pitch correction... I'd never use it live, though.
Live? The only time Spears or Timberlake performs live is in their own nightmares.
I was pretty excited with this genius idea until I read the "500-600 watts" part. Ouch! I wonder if he could use a different alloy, or insulate the strings, or something else to make the power consumption reasonable without changing the sound? Maybe they could make an easy way for the consumer to manually tune the piano so it's just a little bit sharp, so only small corrections would be necessary. Maybe an LED by each string could blink if the note is too low at room temp? I dunno.
I would really like this idea to work. There are a lot of un-tuned pianos out there!
...it would reduce false negatives/positives by 50%.
Not necessarily, that assumes the results of this filter and whatever you augment it with are uncorrelated. In practice it's likely that different anti-spam filters have difficulty with the same messages - the very clever spams, or the friend of a friend who only emails you once in a while.
I started designing a network for distributed storage a few years ago. I decided I wouldn't even care to use it *myself* unless there was some way to avoid having (fragments of) this stuff on my hard drive.
The best I came up with was a distributed blacklist, where you could specify urls of lists of files you didn't want. Somebody would maintain one for child pornography, somebody else for weapons, etc, and you could choose from them. Of course now we've seen that idea play out with spam, and how it is flawed.
I am all for research. But seeing as how there are always limited resources, you have to do as much as possible with the dollars you get. That's where the ISS falls down (as does a return to the moon as you state.) The real question is, what research has been done on the ISS, which couldn't have been done without it - that is, on the shuttle and by unmanned craft? That hundred billion dollars could have funded *hundreds* of scientifically worthwhile missions.
The shuttle itself can get to a higher orbit than the ISS, and the shuttle can and has been used to carry out microgravity experiments. (And for that matter, most of those experiments could be carried out without people onboard - maybe even more effectively, since people scooting around and using the toilet etc. tend to make vibrations.)
1) The move to 64 bits gives a natural break for some other improvements, such as more registers and new instructions.
2) There isn't a performance hit, especially since the AMD chips run 32 bit code so well, and I'm sure Intel's chip will too.
3) The price premium for 64 bit chips is very small, even now.
4) It's better to do the switch while it's still forward-looking - *before* we start resorting to kluges to address more memory. 1GB memory is very common for desktop machines now. If I buy a computer today, I'd like to have some room for upgrading RAM later. In the past this has always extended my computers' lifetimes.
If you look very carefully at what Intel has actually officially said the whole time, you'll see that they simply said they would provide a solution when the appropriate OS support and perceived need becomes available, and that is EXACTLY what has happened here.
So are you saying this had nothing to do with AMD? With a straight face?
It's not new-ness. A law requiring this sort of software is no different than a law requiring authors to insert government approved passages into their works, or musicians to add a patriotic verse to the end of their songs. You're perfectly OK with forcing mysterious government garbage into my code, and forcing me to pay the license fee no doubt.
I USED to think the idea of the government forcing us all to use approved or monitored computing equipment was ridiculous, but now it doesn't take much imagination at all. Game consoles and DVD players are already there, though the motivation is commercial rather than political. And I look at the recent struggle for cryptography rights. This isn't very different.
Pencil and paper are OK. The main problem is that handwriting is too slow, and you can't "touch write" (you have to look down). I feel the more time spent concentrating on the lecture rather than taking notes, the better.
What I have found to work pretty well is a Palm + fold-out keyboard. This is also a pretty cheap solution; you can get the Palm and the keyboard for a couple hundred bucks, and it might make it easier to get by without a laptop. Go with a B&W screen model and you won't have to worry about plugging it in very often at all. It's a nice feeling to have all my notes in one place, with full-text searching, but then again I can't really say that it will double your GPA overnight.
A laptop screen standing on the desk in front of you is too large and distracting IMHO. Also many laptops make some fan and hdd noise, which really makes you feel conspicuous.
The rovers are cutting edge aerospace, but I haven't seen anything to make me think they're on the forefront of computer vision. The obstacle avoidance done on the rover was first done on earth many years ago. It's not just the NASA hardware that's (necessarily) conservative; the surface of MARS isn't a race so apparently it's better to drive the rovers conservatively and mostly manually, which is what they do.
The rovers aren't even autonomous in real time. They stop, take pictures, plan the next few feet, execute blindly, then stop and open their eyes again to start the next episode. That's not what DARPA is looking for. And the system only looks ahead by a few feet. You might think it's just a matter of adding more computing horsepower, but handling all the disorienting motion from looking while moving is a whole different problem.
The DARPA contest will hopefully be won by somebody pushing the field forward, not by recycling a technology time-tested enough to go on a rover.
Slashdot Mirror
I'm surprised your linked page didn't mention Iran, which suddenly opened its nuclear program to international inspection just a few days after watching Saddam get dragged up out of his hole in the ground.
As for the extra bits, I didn't mean to imply they were never useful, only that they are overkill for archiving originals. This is because they don't come into play until you start editing the picture. Once editing starts of course all bets are off; for instance you could (for some strange reason) decrease the image's contrast drastically, so you're only using between 50% and 55% exposure. Then you'd be using those previously insignificant bits. If you were using an 8 bit format everything that used to be from 0-255 would now be between about 128 and 133, so you'd have lost a lot. But in an original, unedited photo (which is what you'd normally have a lot of in an archive) I still say everything past 10 bits or so is noise.
As to the resolution for storing film, certainly it takes a high resolution to store the image data PLUS the grain. The more relevant question for most people is at what resolution an image taken with a digital camera is enough to look as good as a 35mm frame when blown up and printed. No way a 35mm frame produces the equivalent of 6000x4000 color pixels - at that resolution the grain will be very evident. Still, that's only about 25% as many pixels as the poster I was responding to suggested - already a 75% file size reduction!
The .png format, by the way, can store 16 bits per channel, which is overkill. (Too bad gimp is restricted to 8 bits per channel though). 32 bits per channel is silly. All those extra bits (beyond 10 or so) are insignificant, in the sense that neither film nor digital is that accurate; those extra bits are just noise. Some actual information may be pushed down into those bits if you compress part of the range during image processing, but when you are mass archiving unedited photos, there's no point in storing all those insignificant bits.
I guess our station is a mix. They play NPR stuff twice a day, but the rest of the time it's shows like "native american news" and "democracy now" (and believe me, Monsanto and "democracy now" go together like Microsoft and Stallman).
You, know, those hundreds of default.ida and scripts/..%252f.. requests you get every day? According to these guys the cracked machines behind those requests don't exist, or at least don't count.
Nevertheless I'm going to take a closer look and see how I can secure my linux boxes better. I'm surprised linux fared so badly, because many of the services running on linux (apache, sshd, ntp) are the very same ones running on the bsd boxes which did better.
There's no objective way to exactly compare film/digital resolutions, but your estimate is certainly biased towards film.
Apparently Kodak is "the #1 supplier of photo imaging kiosks at retail." So I think they're way ahead of you.
Well, I dunno. Britain has owned real estate in a lot of places around the world at one time or another.
I know it's kind of flaky, but I like "American Idol" because I think it is an honest talent show. True, they consider stage presence and showmanship as well as singing, but the people who do well really are talented and entertaining, and I don't think they're using any Autotune. In fact, one of the judges on the show (the black guy formerly from Rush) has been talking in interviews about how the music industry is sagging because it's so out of touch, and how many of the people who do well on the show probably wouldn't get to sign a deal otherwise, like the fat guy from last year.
I would really like this idea to work. There are a lot of un-tuned pianos out there!
The best I came up with was a distributed blacklist, where you could specify urls of lists of files you didn't want. Somebody would maintain one for child pornography, somebody else for weapons, etc, and you could choose from them. Of course now we've seen that idea play out with spam, and how it is flawed.
The shuttle itself can get to a higher orbit than the ISS, and the shuttle can and has been used to carry out microgravity experiments. (And for that matter, most of those experiments could be carried out without people onboard - maybe even more effectively, since people scooting around and using the toilet etc. tend to make vibrations.)
I for one will pay a small premium to get 1 larger stick instead of 2 smaller ones. That way I can upgrade without chucking my original ram.
I had one of the early HP Kayak computers with Rambus memory. HP ended up having to send out an entire *daughter board* to remedy the Rambus issues.
1) The move to 64 bits gives a natural break for some other improvements, such as more registers and new instructions.
2) There isn't a performance hit, especially since the AMD chips run 32 bit code so well, and I'm sure Intel's chip will too.
3) The price premium for 64 bit chips is very small, even now.
4) It's better to do the switch while it's still forward-looking - *before* we start resorting to kluges to address more memory. 1GB memory is very common for desktop machines now. If I buy a computer today, I'd like to have some room for upgrading RAM later. In the past this has always extended my computers' lifetimes.
I USED to think the idea of the government forcing us all to use approved or monitored computing equipment was ridiculous, but now it doesn't take much imagination at all. Game consoles and DVD players are already there, though the motivation is commercial rather than political. And I look at the recent struggle for cryptography rights. This isn't very different.
What I have found to work pretty well is a Palm + fold-out keyboard. This is also a pretty cheap solution; you can get the Palm and the keyboard for a couple hundred bucks, and it might make it easier to get by without a laptop. Go with a B&W screen model and you won't have to worry about plugging it in very often at all. It's a nice feeling to have all my notes in one place, with full-text searching, but then again I can't really say that it will double your GPA overnight.
A laptop screen standing on the desk in front of you is too large and distracting IMHO. Also many laptops make some fan and hdd noise, which really makes you feel conspicuous.
The rovers aren't even autonomous in real time. They stop, take pictures, plan the next few feet, execute blindly, then stop and open their eyes again to start the next episode. That's not what DARPA is looking for. And the system only looks ahead by a few feet. You might think it's just a matter of adding more computing horsepower, but handling all the disorienting motion from looking while moving is a whole different problem.
The DARPA contest will hopefully be won by somebody pushing the field forward, not by recycling a technology time-tested enough to go on a rover.