How many minutes can you expose before dark noise becomes a problem? With CCD, I am limited to about 30 seconds. Can the CMOS really handle 10-minute or longer exposures?
In fact, astronomers nowadays use digital CCD detectors as their standard imaging equipment.
Sure, there are great CCDs for astronomy. I don't question that. I am addressing assertions about film versus CCDs in general-purpose cameras. These cameras don't use low-light optimized CCDs. For one thing, astronomers cool their CCDs with liquid nitrogen, and I don't think there would be much market for an SLR with a Dewar.
As to the background light of the sky question, that's a good idea, but it's not the main difference. I can take any commercial DSLR and shoot with a lens cap on, the viewfinder blocked, and with the camera inside a black bag and get a solid white saturated frame with only 10 minutes exposure or so.
It may not be what you're looking for because it's an API more than a PDF-writing application, but Reportlab is a great high-level pdf-writing API for python. It's quite easy to write scripts to query DBs and generate good reports. It's also great for charting/graphics. It includes great documentation and lots of example code.
Plus, being open source, it's easy to read the code.
It's not so bad to clean the sensor yourself. There are a lot of decent instructions out there on the web.
Re:It's a half-frame. Focal length issues.
on
Digital 35mm SLRs?
·
· Score: 1
In the meantime, I can afford to buy a lot of film for my old Olympus OM-2 with the $7000 I'm not spending on a DSLR.;)
I pay about $0.50 per frame for film, developing, and scanning (negatives and CD only; no prints). Unless you're getting much cheaper lab work than I am, this adds up to about 14,000 frames, which is about what I would shoot in 2-2.5 years.
If you went with a lower-end DSLR (Digital Rebel, D100, etc.), you could pay for the camera with film savings in a year.
Meanwhile, you might be interested to note that Nikon has started producing a "DX" line of lenses designed only for DSLRs with APS-sized sensors.
There's more to it than that. For many things you may be correct, but unlike the CD vs. vinyl crowd, there is an easy objective test for film vs. digital: Try to shoot a 30-minute star trail exposure with any digital camera on the market. Even better, make it 2 hours to get nice long trails!
The difference will not be subjective. The film camera will have star trails and the digital will be completely saturated with dark noise. That's an objective measurement.
I'm with you on everything but latitude. You really start to see the difference between digital and film on anything longer than about 10 seconds if you look carefully. I can take decent shots up to maybe 1 minute, but forget 5 minutes, much less 30. No star trails from digicams.
Latitude is not so bad for digital. In principle a CCD digitized to 12 bits, as most cameras do, is capable of 12 zones. B&W film gets about 8-9 depending on the emulsion. Color print film is somewhat worse and slide film has almost no latitude (Velvia is about the worst on this, but its color saturation is so beautiful that sometimes it's worth all the hassle of lighting to get those colors!). When you actually go to print the image, you can't get better than 8 zones of latitude from any paper I know, so you have to dodge and burn if you're going to fit a 9-zone negative onto paper without losing shadows or highlights.
Of course you never really get 12 bits of latitude from a CCD, but it's pretty typical in my tests to get at least 8 zones, which means that your output device (printer, CRT, LCD) will be the limiting factor. This is much as it is in the darkroom, where you have a hard time finding printing paper that will match the range you get on your negatives.
What's really differnent about digital imaging from film is that the CCD's transfer function stays pretty linear all the way down to black. Towards the white end, it also stays quite linear until it gets very close to saturation. This is a lot different from the sigmoid film curves we've all known and loved since Ansel Adams published "The Negative." This means you have to think out your high- and low-ends more carefully.
About your 4x5, one big difference between view cameras and 35mm is that almost nobody shoots thousands of frames per year with a view camera. The whole point of the camera is to spend half an hour setting up your shot and getting it right in one or two exposures. With 35mm you pay for the digital sensor in a year or less with the savings on film and processing. With the 4x5 you'd be waiting a long time to pay back the film costs. Even more with an 8x10.
Pro. vs. Consumer cameras (more than megapixels)
on
Digital 35mm SLRs?
·
· Score: 2, Informative
The D1x and D1h don't have the megapixels that the D100 has, but they have much better metering and autofocus modules, as well as better capabilities for burst photography.
The D100 lets you shoot 4 frames and then you have to wait a minute for it to write the frames to the CF card.
The D1x lets you shoot about 8 frames before the buffer fills, and the D1h lets you shoot something like 40 frames. This matters to some people. These two cameras also have a much better & faster autofocus module, which I don't need but which can make a lot of difference for someone shooting a footballer through a 300mm lens.
Also, if you plan on shooting under physically rough conditions, you might want a rugged magnesium body that will survive dropping and getting water splashed on it. The D100 is fine for people like me or you, but if I were a professional journalist a plastic body might not take the beating a pro's camera is subjected to in the field. At the same time, if the picture's going to end up on newsprint you don't need 6 megapixels to get adequate resolution.
So you and I are better with something more like a D100, but for the pros there are good reasons to drop 3-5 grand on a rugged high-performance camera.
As to long-term valuation, in 5 years I expect my D100 to take as good pictures as it does now. I haven't sold any of my film cameras and probably will not sell the D100 when I eventually buy a new body, so what's the problem with valuation?
A good question to ask would be, do I really want an SLR? The advantages of an SLR are mostly: large catalog of interchangeable lenses, accurate framing and metering with different lenses, and no parallax for close-up shots. If you don't want to buy lots of lenses, then the first two advantages are moot. For film, many photographers prefer rangefinder cameras because they are smaller, lighter, and less obtrusive for photojournalism and candid portraiture. Similarly, you could buy a small non-SLR digital camera. All the mechanicals on SLRs (mirror, shutter, lens mount) are expensive and going with a well-made non-SLR might be a better investment.
If you are serious about an SLR, you probably are going to invest many thousands of dollars in optics. This is where you pretty much want to choose a lens system and buy a camera body that will work with the lenses you want to buy. The two major choices are Nikon and Canon. As another poster has written, Nikon doesn't have anything in the range of the Digital Rebel, but at $1500, the D100 is not a bad buy. Both Nikon and Canon DSLRs are getting great reviews. I've got a D100 and am perfectly happy with it, but Canons are fine as well. As to different models within a given brand, the more expensive bodies tend to have better autofocusing and metering and more flexibility in exposure modes. You'll find that autofocusing on low-end DSLRs (<$2000) is like AF on low-end film SLRs (<$800): slow and sometimes inaccurate, but generally quite adequate if you're not shooting sporting events through bit telephotos.
In the long run, you are likely to spend a lot more money on lenses than on your camera body, so it's worth thinking carefully about which line of lenses you want to commit to. If you end up in five years with several thousand invested in Canon or Nikon lenses you will be pretty much be locked into that brand. For this reason, I would focus on the long term and maybe pay several hundred dollars more for a body from one brand than I would for another if I liked the lenses better (or already had a substantial investment in glass) from that company. If you don't have a lot of glass already, Canon makes great lenses and lots of other companies make EOS-compatible lenses, so you would do fine with the Canon body.
As to cost, a $1500 digital SLR is about equal to a $500 film SLR and the Digital Rebel is probably about equal to the lowest-end Canon Rebels. What's worth considering when you weigh these costs is that film is not cheap if you take a lot of pictures. My estimates is that if you just buy good color negative film and have a decent lab process it and burn it onto CD (no prints), you are talking about something like $0.50 per frame, more if you want high-resolution scans of your negatives. I shoot about 5,000 frames per year, so at this rate eliminating the film saves me something around $2500 per year.
As to image quality, the current crop of DSLRs, even with their APS-sized sensors (18x24 mm instead of the 24x36mm film format) give you enough resolution to do about 7x10 inch prints at 300dpi (that's the standard for magazine printing) and you can really go somewhat larger (12x18 or so) with enough quality that you'd be quite happy to hang the print on your wall.
Nations might be a more appropriate work to point to as "the root of much modern economic theory,"
After all, nobody would listen to Marx if he wrote this:
In civilized society, it is only among the inferior ranks of people that the scantiness of subsistence can set limits to the further multiplication of the human species ; and it can do so in no other way than by destroying a great part of the children which their fruitful marriages produce.... It is in this manner that the demand for men, like that for any other commodity, necessarily regulates the production of men, quickens it when it goes on too slowly, and stops it when it advances too fast.
You are right to be skeptical. However, the sources of the quotes are readily available. They're just not on line, so I linked to a page that had bibliographic citations to the primary sources on dead trees and videotape.
I have seen the "Private Universe" videotape and you can order it from the link I supplied if you want to see it too. My favorite part is when Professor Stephen Thernstrom, coauthor of the anti-affirmative action screed America in Black and White confidently demonstrates his ignorance of basic science, thus illustrating that a lot of fat white men don't deserve to teach at Harvard either!
If you're interested in the survey, it was performed by Jon D. Miller, the director of the International Center for the Advancement of Scientific Literacy, Professor of Political Science at Northern Illinois University and former vice president of the Chicago Academy of Science. Miller has published a large number of telephone surveys of scientific literacy, and he does address methodological details, such as the ones you raise, in his publications. Two good places to go are J.D. Miller, "The measurement of civic scientific literacy," Public Understanding of Science, Vol. 7, pp. 203-23 (1998) and the book, J.D. Miller, R. Pardo, and F. Niwa, Public Perceptions of Science and Technology: A Comparative Study of
the European Union, the United States, Japan, and Canada (Madrid: BBV Foundation, 1997) and other publications referenced therein.
Some results of other surveys he has performed include:
48% of the respondents knew that light traveles faster than sound.
48% knew that the earth orbits the sun and not visa versa.
47% knew that humans and dinosaurs did not live concurrently.
45% were familiar with evolution as a scientific concept.
41% knew that the electron was smaller than an atom.
37% knew that lasers worked with light.
32% could identify the meaning of the words "Big Bang."
It's simple: don't mix bases. If you're expressing the number of gigabytes in decimal, then it should use a decimal billion. If you express the number of gigabytes in base 2, 8, or 16, then you can use 2^30 to represent a gigabyte.
Thus,
120 gigabytes would mean 120 x 10^9 bytes
0120 gigabytes would mean 0120 x 8^10 bytes, or 85 x 10^9 decimal
0x120 gigabytes would be 0x120 x 16^7.5 bytes, or 309 x 10^9 decimal
But real geeks would only buy a 1,111,000 gigabyte disk (129 x 10^9 decimal)
Americans are phenomenally ignorant about climate. Most do not even know why summer is hotter than winter. As the AAAS Project 2061 describes it,
A classic video made at a Harvard University graduation illustrates what I mean (
Private Universe Project, 1989). In the video, young graduates and faculty--still in their caps and gowns-- answer this question: Why is it warm in the summer and cold in the winter? Twenty-two out of 25 got the answer wrong. The typical answer was that it's warmer in the summer because the earth is closer to the sun. (The correct answer is that it's warmer then because the tilt of the earth, which remains constant as the earth orbits the sun, puts each hemisphere at an angle to receive maximum sunlight during the summer. The distance from the earth to the sun varies very little--actually, the earth is a little closer to the sun in January.)
More than half of the US population doesn't know that the earth orbits the sun or how scientists figured out that it does. Almost no one can explain what the phrase "orbits the sun" even means. Worse still, few can distinguish between an evidence-based explanation of how the physical world works and an opinion-based one.
Thanks for the correction. I was only speaking about the Infiniiums and falsely generalized to the whole product line.
It does surprise me that users love the feature that upgrading the software breaks the ability read old binary data files captured with that scope. I suppose it shows how much users value their data.
Hmm the Agilent equipment I've been using lately runs HP-UX. Nothing like a nice lightweight embedded OS...
Tektronix uses 2k as well though...and I've yet to see either crash.
My Agilent Infiniium scopes run Win95 and while they usually run well, I've definitely had my share of freezes and BSODs.
My favorite aspect of the Infiniium line is the fact that they don't document the binary file format for storing data and that they change the binary format arbitrarily across different releases of the scope software, with no provision for backward-compatibility, so different Infiniium scopes can't read each others' data files and even for the same scope, if you upgrade the scope software you may not be able to read older data files recorded on that very scope.
Agilent. They're the ones who decided it would be a good idea if all HP scopes ran Windows 95 as their embedded OS. I always wanted a scope that would BSOD.
Here is a description of Donald Knuth's writing process, as related by one of his students in Mathematical Writing (p. 14):
His first copy is written in pencil. Some people compose at a terminal, but Don says, "The
speed at which I write by hand is almost perfectly synchronized with the speed at which
I think. I type faster than I think so I have to stop, and that interrupts the flow."
In the process of typing his handwritten copy into the computer he edits his composition
for flow, so that it will read well at normal reading speed. Somewhere around here the
text gets TeXed, but the description of this stage was tangled up with the description of
the process of rewriting the composition. Of course, rewriting does not all occur at any
one stage. As Don said, "You see things in different ways on the different passes. Some
things look good in longhand but not in type."
The problem being that once a commercial technology ("commercial off-the-shelf" or COTS in milspeak) starts to leak into a closed architecture application, it becomes almost impossible for manufactuers to resist the pressure to use all the features of the commercial technology to reduce cost.
Perhaps you missed the item in the story about how ATMs already use COTS. The switch the O'Reilly story describes is from one COTS operating system (OS/2) to another (Windows).
Since the standard OS/2 distro has a TCP/IP stack, I don't see why a windows-based ATM is more likely than an OS/2-based one to add the stack.
Not according to Click and Clack. The other week on their radio show, they were advising a caller to run her car dry and then mark the true empty level on her fuel gauge with a Sharpie.
In WTO-world, corporations can move their jobs across borders but workers cannot follow. This one-sidedness pushes salaries down everywhere, as companies seek the cheapest available labor.
Maybe I'm slow, but it would seem to me that if workers could chase jobs across borders, the increased supply of labor would drive wages down, not up.
That's why the US doesn't open our borders to all workers who want to come here---if everyone migrated to the US, wages would plummet.
Access to the source code is a precondition for this
If someone redistributes free software without source code, under what legal grounds can you sue? You can sue only on the grounds that the copyright was violated by redistributing the software without complying with the licensing terms. The FSF carefully copyrights all GNU code because copyright law provides the only legal protection the GPL has.
Without copyright, all software would be in the public domain. The GPL and other Free Software philosophy explain why public domain does not protect freedom as much as copyright law does.
Slashdot Mirror
Just to be pedantic, fluorescent lights flicker at 120 Hz because there are two zero-crossings of the line voltage in each cycle.
How many minutes can you expose before dark noise becomes a problem? With CCD, I am limited to about 30 seconds. Can the CMOS really handle 10-minute or longer exposures?
As to the background light of the sky question, that's a good idea, but it's not the main difference. I can take any commercial DSLR and shoot with a lens cap on, the viewfinder blocked, and with the camera inside a black bag and get a solid white saturated frame with only 10 minutes exposure or so.
Plus, being open source, it's easy to read the code.
It's not so bad to clean the sensor yourself. There are a lot of decent instructions out there on the web.
I pay about $0.50 per frame for film, developing, and scanning (negatives and CD only; no prints). Unless you're getting much cheaper lab work than I am, this adds up to about 14,000 frames, which is about what I would shoot in 2-2.5 years.
If you went with a lower-end DSLR (Digital Rebel, D100, etc.), you could pay for the camera with film savings in a year.
Meanwhile, you might be interested to note that Nikon has started producing a "DX" line of lenses designed only for DSLRs with APS-sized sensors.
There's more to it than that. For many things you may be correct, but unlike the CD vs. vinyl crowd, there is an easy objective test for film vs. digital: Try to shoot a 30-minute star trail exposure with any digital camera on the market. Even better, make it 2 hours to get nice long trails!
The difference will not be subjective. The film camera will have star trails and the digital will be completely saturated with dark noise. That's an objective measurement.
I'm with you on everything but latitude. You really start to see the difference between digital and film on anything longer than about 10 seconds if you look carefully. I can take decent shots up to maybe 1 minute, but forget 5 minutes, much less 30. No star trails from digicams.
Latitude is not so bad for digital. In principle a CCD digitized to 12 bits, as most cameras do, is capable of 12 zones. B&W film gets about 8-9 depending on the emulsion. Color print film is somewhat worse and slide film has almost no latitude (Velvia is about the worst on this, but its color saturation is so beautiful that sometimes it's worth all the hassle of lighting to get those colors!). When you actually go to print the image, you can't get better than 8 zones of latitude from any paper I know, so you have to dodge and burn if you're going to fit a 9-zone negative onto paper without losing shadows or highlights.
Of course you never really get 12 bits of latitude from a CCD, but it's pretty typical in my tests to get at least 8 zones, which means that your output device (printer, CRT, LCD) will be the limiting factor. This is much as it is in the darkroom, where you have a hard time finding printing paper that will match the range you get on your negatives.
What's really differnent about digital imaging from film is that the CCD's transfer function stays pretty linear all the way down to black. Towards the white end, it also stays quite linear until it gets very close to saturation. This is a lot different from the sigmoid film curves we've all known and loved since Ansel Adams published "The Negative." This means you have to think out your high- and low-ends more carefully.
About your 4x5, one big difference between view cameras and 35mm is that almost nobody shoots thousands of frames per year with a view camera. The whole point of the camera is to spend half an hour setting up your shot and getting it right in one or two exposures. With 35mm you pay for the digital sensor in a year or less with the savings on film and processing. With the 4x5 you'd be waiting a long time to pay back the film costs. Even more with an 8x10.
The D1x and D1h don't have the megapixels that the D100 has, but they have much better metering and autofocus modules, as well as better capabilities for burst photography.
The D100 lets you shoot 4 frames and then you have to wait a minute for it to write the frames to the CF card.
The D1x lets you shoot about 8 frames before the buffer fills, and the D1h lets you shoot something like 40 frames. This matters to some people. These two cameras also have a much better & faster autofocus module, which I don't need but which can make a lot of difference for someone shooting a footballer through a 300mm lens.
Also, if you plan on shooting under physically rough conditions, you might want a rugged magnesium body that will survive dropping and getting water splashed on it. The D100 is fine for people like me or you, but if I were a professional journalist a plastic body might not take the beating a pro's camera is subjected to in the field. At the same time, if the picture's going to end up on newsprint you don't need 6 megapixels to get adequate resolution.
So you and I are better with something more like a D100, but for the pros there are good reasons to drop 3-5 grand on a rugged high-performance camera.
As to long-term valuation, in 5 years I expect my D100 to take as good pictures as it does now. I haven't sold any of my film cameras and probably will not sell the D100 when I eventually buy a new body, so what's the problem with valuation?
If you are serious about an SLR, you probably are going to invest many thousands of dollars in optics. This is where you pretty much want to choose a lens system and buy a camera body that will work with the lenses you want to buy. The two major choices are Nikon and Canon. As another poster has written, Nikon doesn't have anything in the range of the Digital Rebel, but at $1500, the D100 is not a bad buy. Both Nikon and Canon DSLRs are getting great reviews. I've got a D100 and am perfectly happy with it, but Canons are fine as well. As to different models within a given brand, the more expensive bodies tend to have better autofocusing and metering and more flexibility in exposure modes. You'll find that autofocusing on low-end DSLRs (<$2000) is like AF on low-end film SLRs (<$800): slow and sometimes inaccurate, but generally quite adequate if you're not shooting sporting events through bit telephotos.
In the long run, you are likely to spend a lot more money on lenses than on your camera body, so it's worth thinking carefully about which line of lenses you want to commit to. If you end up in five years with several thousand invested in Canon or Nikon lenses you will be pretty much be locked into that brand. For this reason, I would focus on the long term and maybe pay several hundred dollars more for a body from one brand than I would for another if I liked the lenses better (or already had a substantial investment in glass) from that company. If you don't have a lot of glass already, Canon makes great lenses and lots of other companies make EOS-compatible lenses, so you would do fine with the Canon body.
As to cost, a $1500 digital SLR is about equal to a $500 film SLR and the Digital Rebel is probably about equal to the lowest-end Canon Rebels. What's worth considering when you weigh these costs is that film is not cheap if you take a lot of pictures. My estimates is that if you just buy good color negative film and have a decent lab process it and burn it onto CD (no prints), you are talking about something like $0.50 per frame, more if you want high-resolution scans of your negatives. I shoot about 5,000 frames per year, so at this rate eliminating the film saves me something around $2500 per year.
As to image quality, the current crop of DSLRs, even with their APS-sized sensors (18x24 mm instead of the 24x36mm film format) give you enough resolution to do about 7x10 inch prints at 300dpi (that's the standard for magazine printing) and you can really go somewhat larger (12x18 or so) with enough quality that you'd be quite happy to hang the print on your wall.
After all, nobody would listen to Marx if he wrote this:
I have seen the "Private Universe" videotape and you can order it from the link I supplied if you want to see it too. My favorite part is when Professor Stephen Thernstrom, coauthor of the anti-affirmative action screed America in Black and White confidently demonstrates his ignorance of basic science, thus illustrating that a lot of fat white men don't deserve to teach at Harvard either!
If you're interested in the survey, it was performed by Jon D. Miller, the director of the International Center for the Advancement of Scientific Literacy, Professor of Political Science at Northern Illinois University and former vice president of the Chicago Academy of Science. Miller has published a large number of telephone surveys of scientific literacy, and he does address methodological details, such as the ones you raise, in his publications. Two good places to go are J.D. Miller, "The measurement of civic scientific literacy," Public Understanding of Science, Vol. 7, pp. 203-23 (1998) and the book, J.D. Miller, R. Pardo, and F. Niwa, Public Perceptions of Science and Technology: A Comparative Study of the European Union, the United States, Japan, and Canada (Madrid: BBV Foundation, 1997) and other publications referenced therein.
Some results of other surveys he has performed include:
It does surprise me that users love the feature that upgrading the software breaks the ability read old binary data files captured with that scope. I suppose it shows how much users value their data.
Tektronix uses 2k as well though...and I've yet to see either crash.
My Agilent Infiniium scopes run Win95 and while they usually run well, I've definitely had my share of freezes and BSODs.
My favorite aspect of the Infiniium line is the fact that they don't document the binary file format for storing data and that they change the binary format arbitrarily across different releases of the scope software, with no provision for backward-compatibility, so different Infiniium scopes can't read each others' data files and even for the same scope, if you upgrade the scope software you may not be able to read older data files recorded on that very scope.
I would work at 7-11 in a moment if they would give me a $2 million advance against my 7-11 salary.
Agilent. They're the ones who decided it would be a good idea if all HP scopes ran Windows 95 as their embedded OS. I always wanted a scope that would BSOD.
This experiment is new only in that the atom is nearly at rest in the cavity.
Perhaps you missed the item in the story about how ATMs already use COTS. The switch the O'Reilly story describes is from one COTS operating system (OS/2) to another (Windows).
Since the standard OS/2 distro has a TCP/IP stack, I don't see why a windows-based ATM is more likely than an OS/2-based one to add the stack.
Yes, I too miss the dead-on accuracy of Communist spedometers, tovarishch Drinkypoo.
Not according to Click and Clack. The other week on their radio show, they were advising a caller to run her car dry and then mark the true empty level on her fuel gauge with a Sharpie.
Maybe I'm slow, but it would seem to me that if workers could chase jobs across borders, the increased supply of labor would drive wages down, not up. That's why the US doesn't open our borders to all workers who want to come here---if everyone migrated to the US, wages would plummet.
If someone redistributes free software without source code, under what legal grounds can you sue? You can sue only on the grounds that the copyright was violated by redistributing the software without complying with the licensing terms. The FSF carefully copyrights all GNU code because copyright law provides the only legal protection the GPL has.
Without copyright, all software would be in the public domain. The GPL and other Free Software philosophy explain why public domain does not protect freedom as much as copyright law does.