You don't call any library code, do you? Or any system DLLs? That's good! Modern disassemblers can recognize those entry points and they name the routines, along with all variables used in those calls.
You'd be suprised at how readable ASM can be, even without variable names, when written by someone's C compiler. And how many game developers these days have ever seen ASM, let alone code in it enough to be tricky? (Brian Hook, once from idSoft, said that ASM was obsolete and he recommended against anyone learning it...) The companies have better things to do than make their lead coders (John Carmack, etc) write copy protection.
Especially when the copy protection is based on a standard scheme like SecuROM. Most of those games don't actually take any human intervention to crack, honestly.
The best one I ever heard of was on someone's web page, a description of a a game they wrote for the Mac. It didn't do CD checks, being much earlier, but it was *nasty*. They realized you shouldn't check the results of a test until later. You shouldn't even set a variable and test it later, that's too easy to find. They would do things like set a variable (many routines ago) to fail the test, and then in the check routine they'd fail to set it to pass. And they'd do things like intentionally screw up a jump table with an off-by-one error if the check failed. That way the game would crash later, fifteen minutes or so. And it'd look like a common error. Devious stuff! Must have been more fun to work on than the game. Alas, it sold for early 90s Macs and the market was too small, so it vanished without a trace.
Your link is supposed to prove something? XP is a nasty OS, like ME. I'd rule both of them out for a gaming environment. Of the leftovers (98se and 2k) there isn't much speed difference either way, sometimes 98, sometimes 2k. But 98 is more compatible with many games. Masters of Orion 2 doesn't run in 2k (for me anyways) and GTA2 (multiplayer fun) are two examples.
If all you do is play the latest MS-certifies games, 2k will be fine. But if you boot into something for gaming, 98 is what it should be.
For general use, 2k is pretty nice, two-week uptimes are fairly common and it handles zombie processes much better than 98. But it's slower to boot and needs more ram.
I'm probably what you'd call far left. I believe social programs (some, within reason, like universal schooling) are good. I believe people should be able to do anything they want in their bedroom.
I'm also a very big fan of the US Constitution (despite not even being a USian). I think that in almost all cases it's a better guide to government that the people we elect, regardless of party.
I'd call myself a strict constitutionalist, especially compared to most of the people who use that label.
So, which side am I on? Right, because I support a constitution that's the only thing keeping corrupt politicians from passing any law they're paid for, or left just because I don't buy into the Republican party?
The labels are harmful. Everything is so partisan and you're missing the big pictures. They're all corrupt, with one or two exceptions, on both sides.
The whole 'the required code is copyright' argument has been tossed out fairly recently by the US courts. I haven't heard the final results, but it seems the judge was fairly clear on it.
Copyright covers creative work. If there's only one possible boot code, it's not a creative work, and thus not copyrightable.
Ditto with trademarks. I think Sony uses their trademark in the boot code "This boot code is properly licensed from Sony(tm)(yadda)(yadda)" and the judge cast doubts on that type of scheme as well, though I don't think it was directly at hand.
Same sort of reasoning too. If you are required to duplicate a trademarked phrase, logo, shape, etc, to make a product work, then the trademark fails the generic-mark test, or something like that.
And your argument about the mod chip being a circumvention device is hurt a lot by the existence of XBox Linux, fatally if the justice is honest. It's only required to have a significant legal use, not that the majority of uses be legal.
Of course, it'll be another Kaplan with some judge who worked with MS during law school, or something. No outright bribes, but dishonesty still.
This is a common question so maybe I should answer it.
If you're doing a lot of number crunching or data manipulation (in big sets, with hashes, etc) you're probably spending most of your time in the libraries which are written in C. In fact, being that they're written by programmers skilled in that specific area, you're probably getting better performance than if you wrote them yourself.
Perl isn't an interpretted language, in the traditional sense. In most BASICs, when the execution comes back to a given line it's parsed again, executed, and dumped. If anything, they usually only cache a line or two to help tight loops. Perl is interpretted/compiled all at once, when you start.
Runtime speed is a little slower than other languages, but it's mainly because you've got a lot of runtime checks and hidden memory allocation turned on. Use C++ with automatic array expanding and garbage collecting and you'll see the same kind of performance hits.
That said, the ease of perl causes a lot of features to be misused by programmers who don't know how long it'll take. If you have two pieces of data (a header name and value for instance) it's common to toss them into a hash to keep them related. This isn't really a good idea unless you need to look them up by the header name. If you're just going to dump them out in arbitrary order, you should probably use two arrays in sync. Pre-allocate them to avoid a little delay at every operation. This way you avoid the overhead of the hashing algorithm that you're never going to use, and the slightly-slower lookups compared to an array.
You can also do more complex things this way. I've seen people use hashes here and read the list out by sorting the keys to the hash and iterating through. They'll then do this a few times, sorting at every step. If you want these arrays sorted, but you still don't care about finding a specific header, use an array of two-element arrays, sort the master based on the first element, and not only do you avoid almost all the overhead of hashes, but you have a permanently sorted array, no need to sort at every use.
These programming "errors" are worse in Perl than most older languages because they're easier to implement. In C you'd have to find a library function to create hashes, or write your own. If you started to write your own you'd quickly realize how many cycles you were burning and probably find an easier way unless your application demanded it. In perl (and many little "scripting" languages) you can do so much in a single command that you may not realize.
This is why if I were hiring I'd only take programmers with a "traditional" background of C or other low-level language, before they got to the Perl, Java, Python, or whatever modern rapid-development language we were using. ASM experience is even a plus. Nobody understands the cost of a routine like someone who programmed in ASM. And it's worth thinking about. Usually you say that requiring 512MB of ram ($40 these days) is worth it to save an hour or two of programming, but hopefully at other times you realize a CGI on a busy site can't be that greedy.
So, in conclusion. Perl isn't traditionally interpretted. It's almost as fast, or faster, than C for anything that spends much time in libraries (most code). Most of what slows down Perl (or Python, or Java, or C++, etc) is programmers who don't know what routines they really need to use. The cause of this is usually not enough experience in less "helpful" languages.
The resolution is 6.3, correct. It's from 6.3M monochrome pixels. The color information is from 6.3M pixels, but 2M of each color. They're dithered across the pixels monochrome picture. Because the human eye isn't as sensitive to small color variations, this doesn't look all that bad.
You currently have large corporations bribing standards bodies and government agencies to make their proprietary products a standard. (Microsoft)
If all standards were required to be LGPLed you'd always be able to come along and work with a competitors product, regardless of their wishes.
I'd also say that operating systems and interconnection systems (file system drivers, etc) should all be required to be GPLed. (If you wish to sell to government or government funded industry.) OSes aren't a commodity in the same way as the software that runs on them. They exist just to enable software to work. Competition in this area always seems to involve sabotage of APIs and hiding functionality to allow some products to work better than others.
But, other than the OS and the protocols/file formats/etc, I don't think software needs to be GPLed. And before you ask, that leaves a lot.
Does your church group print up pamphlets with "Tough Questions for Athiests" on them?
Obviously there aren't black and white issues. A parent not wanting to trust ritalin for their child based on a hurried diagnosis is different than one refusing a blood transfusion for a child who lost a lot of blood from a deep cut.
Just the same, abortion is an issue with no good hard rules. Even very pro-abortion people would dislike the idea of 9th month abortions and few reasonable anti-abortionists really hold to the view that every sperm is sacred, or that the instant a sperm and egg touch, that they are morally equivalent to a child.
If you've studied any biology it's hard to get worked up about first trimester abortions. They're probably reasonable later, but except for hidden health complications, there are few valid reasons for waiting this long so limiting abortions to the first trimester is a reasonable compromise.
Considering religion is bunk, it's not a tough choice.
But, I'm libertarian enough to respect your wishes to die over some silly beliefs.
I'm not libertarian enough to allow you to take your child with you. They're not of age yet and having a parent make these choices is worse than any child abuse.
If I ever saw a CS refusing medical treatment for their children I'd make sure they required a lot of treatment for themselves. Until everyone has the idea that this is abuse/murder these dogs will continue in their dangerously stupid practices.
There's an even better Outlook patch, it's the only guaranteed one. Unplug the network cable. It's about as effective as the patch you mention. Unplug the power cable and Windows will stop crashing too!
Fug, disabling the capability to send attachments instead of fixing the client to not be swiss-cheese. What a useless fix.
You can't blame an OS for the services a user installs on it. Windows comes with Outlook, it's standard. If there's a bug in outlook, there's a bug in *EVERY COPY OF WINDOWS* until it's fixed. Even after it's patched, broken systems are still around.
I haven't patched Apache on Linux but I'm not vulnerable. Know why? Because I didn't enable it.
Windows users don't have to enable Outlook (Express) or IE, they're there by default. A hole in one of those is a pretty big flaw.
Had IIS never been installed by default, MS wouldn't have gotten half the flack for Code Red that it did. But most of the CR sources are some unpatched box in a closet, or on someone's desk, where nobody realizes it's running IIS.
Half of the security flaw in MS products is the lousy code, the other half is MS themselves.
btw, re your sig. You haven't got any ideas what real usage number are. Right now I'm counting as a hit for IE6.0 in XP, but I'm really using Mozilla in Linux with the prefbar addon to spoof user-agents. Most Linux users do something like this because so many sites are intentionally crippled to look for IE specifically. And polls are notoriously stuffed by trolls like you who love to point out the results as if they meant any more than a Florida election.
I doubt you'd notice any difference in the Sigma picture if you didn't know to look for it. It looked nice, but not revolutionary. In fact, the cathedral picture showed considerable purple fringing on the right side of objects. An effect I don't even get with my G2.
The foveon sensor is nice, and eventually we may all use one when we've maxed out the spatial resolution of 35mm lenses, but until then spatial resolution is much more important.
The human eye is much more sensitive to brightness differences, and fine detail, than to color resoltion. JPEGs throw away color and keep brightness detail. It's intentional because humans just don't notice the color as much.
Also, because JPEGs are lossy in the color area you shouldn't pay much attention to foveon pictures until you see them in PNG or some other lossy format considering that JPG throws away exactly what you claim to be seeing.
If you get 50mp from film, and you can print a 4mp image at 8x10, I'd expect you to be able to print roughly a 35x28 print from 35mm film without any visible grain.
You wouldn't accept a digital print that started to show pixels, why accept film when it starts to show grain? The only reason we do is because we always had to.
Really though, take a picture of the same resolution chart that the 1Ds maxes out with a Canon EOS 1v (That's the current high-end film camera, right?) and the same lenses as on the 1Ds. If you can't resolve the same ammount of detail, then digital wins. But, we've seen the results from Luminous Landscape that show a picture where 35mm didn't resolve detail that the 1Ds resolved easily.
And the noise difference! WOW! Digital was crystal clear, the only noise in the picture was the JPG artifacts from being shrunk to fit the web.
Not quite. They truly have 6.3 million pixels in the D60, but those are monochrome. Each is behind a filter and senses a certain colour of light. The spatial resolution is truly 6.3 mega-pixel. The color resolution is 1/3 that claimed.
Luckily, the human eye is less sensitive to slight color variations than to brightness variations. This is exactly what JPEG uses to compress pictures. These cameras a high-resolution where it matters and low-resolution where they can get away with it.
Of course, having one spatial pixel sample all three colors will be great, but only if these sensors are as dense as in traditional digital sensors, otherwise we'll get better color (good) as the expense of spatial resolution (quite bad).
Oh yeah. I forgot that most MF cameras come with free film and developing. That's handy. Without that, the very low (zero) incremental cost of digital would have won out in the end.
Actually, MF photographers are some who are really hoping for Digital in the next year or two. It doesn't have to completely match MF, it just has to produce the size of output they do. Who cares if MF could blow up to 6'x10' when you only ever sell 3'x2' posters? And the cost of a picture in MF is what, $5 or more, with all the expenses?
Consider the cost of the materials and a Canon EOS 1D already matches the cost of a similar quality film camera if you figure the TCO over a year or more.
Backing up my old media is pretty simple. I don't burn awkward disks onto more awkward disks, I copy thing into an 'Old' directory in my home directory and roughly categorize it. I've got an 'Apple2' directory in there containing programs I wrote in school twenty years ago. Those were copied onto my PC over a serial cable when I got rid of the Apple2 a few years back and there they sit until I want them. I back up my data, but unless my HD crashes suddenly (not the usual pattern) I'll just copy everything over to a new drive without a bit of trouble. (I also distribute my old backups to friends and family, every time I make a new backup I give the old one to someone else, so that if I lose everything I can go back to my offsite backups. I've recently started using a safe-deposit box for this.)
I can see that re-archiving pictures every few years would be a pain, but you don't really do it that way.
And, when I want to use those old apple2 programs, they're right there. I load an emulator, point it at the disk image, and go. (I have also written an Apple2 filesystem driver (DOS3.3) as well, to pull off individual files, but *that* is beyond the abilities of most users.)
And really, my new HDs are always much bigger than the last, what's 5GB of total "personal files" on 400GB of storage?
You should probably try the Nikon for a while before considering it. The controls are a bit of a nightmare, where I picked up a G2 and started playing with it without needing to look in the manual.
The quality is about the same, despite the 1MP difference. In fact, some people feel the G2 is sharper at 4 than the CP5k is at 5.
The CP5700 is nice with an 8x optical zoom but the widest F-stop at telephoto is pretty bad, where the G2 goes from f2.0 - 2.5 over its 3x zoom.
Whatever you end up doing, check out www.dpreview.com before buying, they've got test shots of everything and side-by-side comparisons of most of the big cameras (G2, CP5k, Sony 707, etc)
Or, just go on EBay and buy a D30 or D60, they use the EOS line of lenses and are *very* low-noise.
So you get a digital system with so many bins that they can easily classify even the rnadom microphone vibrations from the warmth of the air. At that point you'd completely captured the sound to the limit of the capture equipment (microphone) regardless of how you store it.
Also, analog storage is just as imprecise in the real world. Vinyl isn't perfectly smooth, look at it under a magnifying glass and you'll see the grain of the material. That's the limit to what you can retrieve during playback.
As to the watch... Your second hand may show the sub-second time more easily, but I doubt your watch it accurate enough for it to matter. Who cares if you think the time is 4:15:32.026 when the time is really 4:15:37? If your friend's watch is accurate to the second he's got a better clock. Then, consider that he can get a watch that shows hundredths of second in an unambiguous way. Can you really distinguish a second-hand's position to a hundredth-second resolution?
People write articles for a reason. You could try to read one.
The lens was placed on a tripod, the Canon 1Ds digital body was used, then an equivalent film model, one of Canon's best, was used. Exactly the same lens and locations, exactly the same settings.
The film was scanned with a film scanner.
The resolution was "only" 3500dpi or so, but why bother going higher? Noise was already overwhelming the picture and details that were showing up on the digital were completely lost in the 35mm picture.
The MF (645) film took a more detailed picture than the 35mm but the noise was still an issue.
The MF could be blown up farther, but for even a 3'x2' poster the digital would be good and the lack of noise would likely produce a superior image.
You can't just keep enlarging film. Where do you propose to get the lenses for your 16x20 film camera?
Look at the detail out of the pro-sumer 4mp cameras with the tiny sensors. Make those full-frame and you're looking at 50mp+, with standard 35mm lenses.
When you need resolution to match large format you go up to medium format with a super-dense sensor and you get 300mp+, completely blowing away LF...
Anyways, it's likely that the future of very-high-resolution photography is taking many pictures and overlaying them in software. Already grainy videotape can be used to extract 35mm quality pictures.
Film can continually be enlarged, but it's too ridiculous to imagine it being used at super-large sizes. Digital will expand the other way, gaining more resolution until 35mm lenses simply max out, and then for people doing very sensitive work, they'll start working on maxing out MF lenses.
Actually, even relatively "low-end" scanners (You know, only $2000) are well beyond the film they scan. You start picking up so much grain that you don't get usefull detail anymore. Sure, an 8000dpi scan pulls all the detail out of a picture, but all its done is finely resolve the grains.
The site that proclaimed film to be dead printed, with the standard process, a film picture and compared it to a digital printed with a similar process and they still didn't find a win for film.
That was with the D60, last year's camera. This years, the 1Ds has twice the pixels and easily blows past film. Examine the negative with a loupe and you won't see the detail that the digital displays.
And even if that detail was on the negative, what would you do with it if anytime you scanned it for printing into a magazine you'd lose most of that?
Slashdot Mirror
It does, yes.
But I don't feel that it's an adequate way of indicating the speaker. When it's not clear by context there should be an explicit statement.
You don't call any library code, do you? Or any system DLLs? That's good! Modern disassemblers can recognize those entry points and they name the routines, along with all variables used in those calls.
You'd be suprised at how readable ASM can be, even without variable names, when written by someone's C compiler. And how many game developers these days have ever seen ASM, let alone code in it enough to be tricky? (Brian Hook, once from idSoft, said that ASM was obsolete and he recommended against anyone learning it...) The companies have better things to do than make their lead coders (John Carmack, etc) write copy protection.
Especially when the copy protection is based on a standard scheme like SecuROM. Most of those games don't actually take any human intervention to crack, honestly.
The best one I ever heard of was on someone's web page, a description of a a game they wrote for the Mac. It didn't do CD checks, being much earlier, but it was *nasty*. They realized you shouldn't check the results of a test until later. You shouldn't even set a variable and test it later, that's too easy to find. They would do things like set a variable (many routines ago) to fail the test, and then in the check routine they'd fail to set it to pass. And they'd do things like intentionally screw up a jump table with an off-by-one error if the check failed. That way the game would crash later, fifteen minutes or so. And it'd look like a common error. Devious stuff! Must have been more fun to work on than the game. Alas, it sold for early 90s Macs and the market was too small, so it vanished without a trace.
Your link is supposed to prove something? XP is a nasty OS, like ME. I'd rule both of them out for a gaming environment. Of the leftovers (98se and 2k) there isn't much speed difference either way, sometimes 98, sometimes 2k. But 98 is more compatible with many games. Masters of Orion 2 doesn't run in 2k (for me anyways) and GTA2 (multiplayer fun) are two examples.
If all you do is play the latest MS-certifies games, 2k will be fine. But if you boot into something for gaming, 98 is what it should be.
For general use, 2k is pretty nice, two-week uptimes are fairly common and it handles zombie processes much better than 98. But it's slower to boot and needs more ram.
It's also a British-ism, I think. I see it quite frequently in Canada.
Makes sense to me just from a common-sense point of view.
I wish we could get people to stop leaving off closing quotation marks at the end of a paragraph. It may be optional, but it's sloppy.
It's nothing to do with parties.
I'm probably what you'd call far left. I believe social programs (some, within reason, like universal schooling) are good. I believe people should be able to do anything they want in their bedroom.
I'm also a very big fan of the US Constitution (despite not even being a USian). I think that in almost all cases it's a better guide to government that the people we elect, regardless of party.
I'd call myself a strict constitutionalist, especially compared to most of the people who use that label.
So, which side am I on? Right, because I support a constitution that's the only thing keeping corrupt politicians from passing any law they're paid for, or left just because I don't buy into the Republican party?
The labels are harmful. Everything is so partisan and you're missing the big pictures. They're all corrupt, with one or two exceptions, on both sides.
The whole 'the required code is copyright' argument has been tossed out fairly recently by the US courts. I haven't heard the final results, but it seems the judge was fairly clear on it.
Copyright covers creative work. If there's only one possible boot code, it's not a creative work, and thus not copyrightable.
Ditto with trademarks. I think Sony uses their trademark in the boot code "This boot code is properly licensed from Sony(tm)(yadda)(yadda)" and the judge cast doubts on that type of scheme as well, though I don't think it was directly at hand.
Same sort of reasoning too. If you are required to duplicate a trademarked phrase, logo, shape, etc, to make a product work, then the trademark fails the generic-mark test, or something like that.
And your argument about the mod chip being a circumvention device is hurt a lot by the existence of XBox Linux, fatally if the justice is honest. It's only required to have a significant legal use, not that the majority of uses be legal.
Of course, it'll be another Kaplan with some judge who worked with MS during law school, or something. No outright bribes, but dishonesty still.
This is a common question so maybe I should answer it.
If you're doing a lot of number crunching or data manipulation (in big sets, with hashes, etc) you're probably spending most of your time in the libraries which are written in C. In fact, being that they're written by programmers skilled in that specific area, you're probably getting better performance than if you wrote them yourself.
Perl isn't an interpretted language, in the traditional sense. In most BASICs, when the execution comes back to a given line it's parsed again, executed, and dumped. If anything, they usually only cache a line or two to help tight loops. Perl is interpretted/compiled all at once, when you start.
Runtime speed is a little slower than other languages, but it's mainly because you've got a lot of runtime checks and hidden memory allocation turned on. Use C++ with automatic array expanding and garbage collecting and you'll see the same kind of performance hits.
That said, the ease of perl causes a lot of features to be misused by programmers who don't know how long it'll take. If you have two pieces of data (a header name and value for instance) it's common to toss them into a hash to keep them related. This isn't really a good idea unless you need to look them up by the header name. If you're just going to dump them out in arbitrary order, you should probably use two arrays in sync. Pre-allocate them to avoid a little delay at every operation. This way you avoid the overhead of the hashing algorithm that you're never going to use, and the slightly-slower lookups compared to an array.
You can also do more complex things this way. I've seen people use hashes here and read the list out by sorting the keys to the hash and iterating through. They'll then do this a few times, sorting at every step. If you want these arrays sorted, but you still don't care about finding a specific header, use an array of two-element arrays, sort the master based on the first element, and not only do you avoid almost all the overhead of hashes, but you have a permanently sorted array, no need to sort at every use.
These programming "errors" are worse in Perl than most older languages because they're easier to implement. In C you'd have to find a library function to create hashes, or write your own. If you started to write your own you'd quickly realize how many cycles you were burning and probably find an easier way unless your application demanded it. In perl (and many little "scripting" languages) you can do so much in a single command that you may not realize.
This is why if I were hiring I'd only take programmers with a "traditional" background of C or other low-level language, before they got to the Perl, Java, Python, or whatever modern rapid-development language we were using. ASM experience is even a plus. Nobody understands the cost of a routine like someone who programmed in ASM. And it's worth thinking about. Usually you say that requiring 512MB of ram ($40 these days) is worth it to save an hour or two of programming, but hopefully at other times you realize a CGI on a busy site can't be that greedy.
So, in conclusion. Perl isn't traditionally interpretted. It's almost as fast, or faster, than C for anything that spends much time in libraries (most code). Most of what slows down Perl (or Python, or Java, or C++, etc) is programmers who don't know what routines they really need to use. The cause of this is usually not enough experience in less "helpful" languages.
The resolution is 6.3, correct. It's from 6.3M monochrome pixels. The color information is from 6.3M pixels, but 2M of each color. They're dithered across the pixels monochrome picture. Because the human eye isn't as sensitive to small color variations, this doesn't look all that bad.
Or did I not understand your question?
Actually, I think you do need those regulations.
You currently have large corporations bribing standards bodies and government agencies to make their proprietary products a standard. (Microsoft)
If all standards were required to be LGPLed you'd always be able to come along and work with a competitors product, regardless of their wishes.
I'd also say that operating systems and interconnection systems (file system drivers, etc) should all be required to be GPLed. (If you wish to sell to government or government funded industry.) OSes aren't a commodity in the same way as the software that runs on them. They exist just to enable software to work. Competition in this area always seems to involve sabotage of APIs and hiding functionality to allow some products to work better than others.
But, other than the OS and the protocols/file formats/etc, I don't think software needs to be GPLed. And before you ask, that leaves a lot.
Does your church group print up pamphlets with "Tough Questions for Athiests" on them?
Obviously there aren't black and white issues. A parent not wanting to trust ritalin for their child based on a hurried diagnosis is different than one refusing a blood transfusion for a child who lost a lot of blood from a deep cut.
Just the same, abortion is an issue with no good hard rules. Even very pro-abortion people would dislike the idea of 9th month abortions and few reasonable anti-abortionists really hold to the view that every sperm is sacred, or that the instant a sperm and egg touch, that they are morally equivalent to a child.
If you've studied any biology it's hard to get worked up about first trimester abortions. They're probably reasonable later, but except for hidden health complications, there are few valid reasons for waiting this long so limiting abortions to the first trimester is a reasonable compromise.
Considering religion is bunk, it's not a tough choice.
But, I'm libertarian enough to respect your wishes to die over some silly beliefs.
I'm not libertarian enough to allow you to take your child with you. They're not of age yet and having a parent make these choices is worse than any child abuse.
If I ever saw a CS refusing medical treatment for their children I'd make sure they required a lot of treatment for themselves. Until everyone has the idea that this is abuse/murder these dogs will continue in their dangerously stupid practices.
There's an even better Outlook patch, it's the only guaranteed one. Unplug the network cable. It's about as effective as the patch you mention. Unplug the power cable and Windows will stop crashing too!
Fug, disabling the capability to send attachments instead of fixing the client to not be swiss-cheese. What a useless fix.
Face it, he's got a point and you missed it.
You can't blame an OS for the services a user installs on it. Windows comes with Outlook, it's standard. If there's a bug in outlook, there's a bug in *EVERY COPY OF WINDOWS* until it's fixed. Even after it's patched, broken systems are still around.
I haven't patched Apache on Linux but I'm not vulnerable. Know why? Because I didn't enable it.
Windows users don't have to enable Outlook (Express) or IE, they're there by default. A hole in one of those is a pretty big flaw.
Had IIS never been installed by default, MS wouldn't have gotten half the flack for Code Red that it did. But most of the CR sources are some unpatched box in a closet, or on someone's desk, where nobody realizes it's running IIS.
Half of the security flaw in MS products is the lousy code, the other half is MS themselves.
btw, re your sig. You haven't got any ideas what real usage number are. Right now I'm counting as a hit for IE6.0 in XP, but I'm really using Mozilla in Linux with the prefbar addon to spoof user-agents. Most Linux users do something like this because so many sites are intentionally crippled to look for IE specifically. And polls are notoriously stuffed by trolls like you who love to point out the results as if they meant any more than a Florida election.
I doubt you'd notice any difference in the Sigma picture if you didn't know to look for it. It looked nice, but not revolutionary. In fact, the cathedral picture showed considerable purple fringing on the right side of objects. An effect I don't even get with my G2.
The foveon sensor is nice, and eventually we may all use one when we've maxed out the spatial resolution of 35mm lenses, but until then spatial resolution is much more important.
The human eye is much more sensitive to brightness differences, and fine detail, than to color resoltion. JPEGs throw away color and keep brightness detail. It's intentional because humans just don't notice the color as much.
Also, because JPEGs are lossy in the color area you shouldn't pay much attention to foveon pictures until you see them in PNG or some other lossy format considering that JPG throws away exactly what you claim to be seeing.
If you get 50mp from film, and you can print a 4mp image at 8x10, I'd expect you to be able to print roughly a 35x28 print from 35mm film without any visible grain.
You wouldn't accept a digital print that started to show pixels, why accept film when it starts to show grain? The only reason we do is because we always had to.
Really though, take a picture of the same resolution chart that the 1Ds maxes out with a Canon EOS 1v (That's the current high-end film camera, right?) and the same lenses as on the 1Ds. If you can't resolve the same ammount of detail, then digital wins. But, we've seen the results from Luminous Landscape that show a picture where 35mm didn't resolve detail that the 1Ds resolved easily.
And the noise difference! WOW! Digital was crystal clear, the only noise in the picture was the JPG artifacts from being shrunk to fit the web.
Not quite. They truly have 6.3 million pixels in the D60, but those are monochrome. Each is behind a filter and senses a certain colour of light. The spatial resolution is truly 6.3 mega-pixel. The color resolution is 1/3 that claimed.
Luckily, the human eye is less sensitive to slight color variations than to brightness variations. This is exactly what JPEG uses to compress pictures. These cameras a high-resolution where it matters and low-resolution where they can get away with it.
Of course, having one spatial pixel sample all three colors will be great, but only if these sensors are as dense as in traditional digital sensors, otherwise we'll get better color (good) as the expense of spatial resolution (quite bad).
You are truly quasar-level stupid. Festingly ignorant.
It's a joke. The spelling is intentional.
Oh yeah. I forgot that most MF cameras come with free film and developing. That's handy. Without that, the very low (zero) incremental cost of digital would have won out in the end.
Actually, MF photographers are some who are really hoping for Digital in the next year or two. It doesn't have to completely match MF, it just has to produce the size of output they do. Who cares if MF could blow up to 6'x10' when you only ever sell 3'x2' posters? And the cost of a picture in MF is what, $5 or more, with all the expenses?
Consider the cost of the materials and a Canon EOS 1D already matches the cost of a similar quality film camera if you figure the TCO over a year or more.
Backing up my old media is pretty simple. I don't burn awkward disks onto more awkward disks, I copy thing into an 'Old' directory in my home directory and roughly categorize it. I've got an 'Apple2' directory in there containing programs I wrote in school twenty years ago. Those were copied onto my PC over a serial cable when I got rid of the Apple2 a few years back and there they sit until I want them. I back up my data, but unless my HD crashes suddenly (not the usual pattern) I'll just copy everything over to a new drive without a bit of trouble. (I also distribute my old backups to friends and family, every time I make a new backup I give the old one to someone else, so that if I lose everything I can go back to my offsite backups. I've recently started using a safe-deposit box for this.)
I can see that re-archiving pictures every few years would be a pain, but you don't really do it that way.
And, when I want to use those old apple2 programs, they're right there. I load an emulator, point it at the disk image, and go. (I have also written an Apple2 filesystem driver (DOS3.3) as well, to pull off individual files, but *that* is beyond the abilities of most users.)
And really, my new HDs are always much bigger than the last, what's 5GB of total "personal files" on 400GB of storage?
You should probably try the Nikon for a while before considering it. The controls are a bit of a nightmare, where I picked up a G2 and started playing with it without needing to look in the manual.
The quality is about the same, despite the 1MP difference. In fact, some people feel the G2 is sharper at 4 than the CP5k is at 5.
The CP5700 is nice with an 8x optical zoom but the widest F-stop at telephoto is pretty bad, where the G2 goes from f2.0 - 2.5 over its 3x zoom.
Whatever you end up doing, check out www.dpreview.com before buying, they've got test shots of everything and side-by-side comparisons of most of the big cameras (G2, CP5k, Sony 707, etc)
Or, just go on EBay and buy a D30 or D60, they use the EOS line of lenses and are *very* low-noise.
So you get a digital system with so many bins that they can easily classify even the rnadom microphone vibrations from the warmth of the air. At that point you'd completely captured the sound to the limit of the capture equipment (microphone) regardless of how you store it.
Also, analog storage is just as imprecise in the real world. Vinyl isn't perfectly smooth, look at it under a magnifying glass and you'll see the grain of the material. That's the limit to what you can retrieve during playback.
As to the watch... Your second hand may show the sub-second time more easily, but I doubt your watch it accurate enough for it to matter. Who cares if you think the time is 4:15:32.026 when the time is really 4:15:37? If your friend's watch is accurate to the second he's got a better clock. Then, consider that he can get a watch that shows hundredths of second in an unambiguous way. Can you really distinguish a second-hand's position to a hundredth-second resolution?
People write articles for a reason. You could try to read one.
The lens was placed on a tripod, the Canon 1Ds digital body was used, then an equivalent film model, one of Canon's best, was used. Exactly the same lens and locations, exactly the same settings.
The film was scanned with a film scanner.
The resolution was "only" 3500dpi or so, but why bother going higher? Noise was already overwhelming the picture and details that were showing up on the digital were completely lost in the 35mm picture.
The MF (645) film took a more detailed picture than the 35mm but the noise was still an issue.
The MF could be blown up farther, but for even a 3'x2' poster the digital would be good and the lack of noise would likely produce a superior image.
You can't just keep enlarging film. Where do you propose to get the lenses for your 16x20 film camera?
Look at the detail out of the pro-sumer 4mp cameras with the tiny sensors. Make those full-frame and you're looking at 50mp+, with standard 35mm lenses.
When you need resolution to match large format you go up to medium format with a super-dense sensor and you get 300mp+, completely blowing away LF...
Anyways, it's likely that the future of very-high-resolution photography is taking many pictures and overlaying them in software. Already grainy videotape can be used to extract 35mm quality pictures.
Film can continually be enlarged, but it's too ridiculous to imagine it being used at super-large sizes. Digital will expand the other way, gaining more resolution until 35mm lenses simply max out, and then for people doing very sensitive work, they'll start working on maxing out MF lenses.
Actually, even relatively "low-end" scanners (You know, only $2000) are well beyond the film they scan. You start picking up so much grain that you don't get usefull detail anymore. Sure, an 8000dpi scan pulls all the detail out of a picture, but all its done is finely resolve the grains.
The site that proclaimed film to be dead printed, with the standard process, a film picture and compared it to a digital printed with a similar process and they still didn't find a win for film.
That was with the D60, last year's camera. This years, the 1Ds has twice the pixels and easily blows past film. Examine the negative with a loupe and you won't see the detail that the digital displays.
And even if that detail was on the negative, what would you do with it if anytime you scanned it for printing into a magazine you'd lose most of that?
Quite right. The small size difference isn't worth going to some too-proprietary format.
The lack of DRM is a huge plus. I'll never buy anything with it.