Just about every technological advance, by nature of its disruptiveness, harms some people and helps others, but that doesn't make it wrong. The entertainment industry can either fight a rearguard action to delay the inevitable, hurting a lot of people along the way, or embrace the reality that copying data is cheap and easy and find new ways to profit from the situation. It's always been the mark of a good business person that he or she finds opportunities rather than complaining about the situation. Right now it looks like EMI is starting to understand the situation; let's see how long it takes the rest of the industry (and the movie industry) to figure it out.
If it turns out that a handful of mega-stars supported by large multinational companies is not the most efficient way to deliver entertainment, I see little loss to society as a whole, and it would surely be to the benefit of a much larger set of artists.
(a) He says OOXML is great not because the specification itself is a work of engineering genius, but because out in the Real World is easier to implement than ODF. That might not be for a good reason (OOXML is similar to existing World formats in structure, and so existing code is easily modified to use it, where ODF requires an entirely new approach and so is far harder to add to existing software), but it's certainly a different story than Miguel just blindly loving the OOXML spec.
It might be easier for Microsoft to implement it for that reason, but that doesn't apply to anyone else, none of whom started from that code base. That's not a very good reason for a standard. And StarOffice/OpenOffice have about as good support for legacy MS formats as anyone, but that didn't seem to make it any harder to adopt ODF.
(b) The patent protection claim is exactly what it sounds like, except for the fact that there are NO known parents which Moonlight or Mono infringe. It's a simple of matter of, "if something comes up, we won't sue your customers." Those same companies (Microsoft and the MPEGLA group) are still totally free to sue the developers and companies behind FFMPEG, Linux, GNOME, KDE, Apache, X.org, OpenOffice.org, etc. Nothing about the protection Novell offers will increase the risk of those lawsuits - all it does is decrease the risk for people who download from them. It's a nice gesture that some suit-wearing types give a fuck about, and the rest of us are free to ignore just like we ignore the patent minefield for every other project, all of which are guaranteed to be infringing _something_.
It strikes me as a wee bit cynical -- not to mention rather far outside of the mainstream of conventional GPL/LGPL usage -- to say "we're releasing this under the LGPL, but if you know what's good for you, you'll download it from us". It's not a violation of the LGPL for Novell to do this with their own code, obviously (at least, obvious to me, but I'm not a lawyer), but it's going to make it awfully dangerous for any other project to reuse this code.
Also, what distribution formats will be covered by the patent protection claim? Source? Binary-only? Does this mean that only operating systems that Novell chooses to compile Moonlight for will be "safe"?
If it sounds like I don't trust anything Microsoft touches with a ten foot pole, that's correct -- Microsoft hasn't given anyone much reason to trust them. This looks like yet another attempt to bust the GPL/LGPL, this time in a way that keeps them safe from the FSF -- by using someone who actually owns the copyright on the code.
I would urge the people making the decisions at Novell to go back and read the story of King Midas. Very, very carefully.
4) Keep any distractions that aren't absolutely necessary away from your folks.
5) Don't sign up for any critical technical work (you'll just slip the project).
6) Oh, in case I forgot to mention, fight like crazy for your people.
7) If your people are working overtime/weekends for more than a well-bounded, short term crunch, it's a sign of a problem, not a good sign at all. Kick your folks out and figure out what's going wrong, and fix it.
8) Don't constantly look over everyone's shoulder.
I'm no fan of Microsoft, but there's a book published by Microsoft Press by Steve Maguire or Steve McConnell (whichever one of them didn't write Code Complete) about technical management. It's actually very good indeed.
The combination of manager and architect sounds like a very dangerous one to me. It's going to be very difficult for you to get honest technical feedback from people who report to you, and architects need that kind of candid feedback. An architectural error that goes unchallenged for political reasons is bad news. When I was a manager, I once led a software integration team (a more suitable role for a manager), and even that caused some tension. Not with the person who happened to report to me -- she had no trouble pushing back when needed -- but with another engineer who found it uncomfortable when I had to go around and get commitments from people. I was actually selected for that role because initially no one on the team reported to me (I was the one leadership/management role who didn't have an interest in a particular component within the project), but things change...
Whether 100 Mb/sec is "just fine" depends upon what you're doing. Likewise, whether an occasional audio skip matters or not depends upon what you're doing.
For home users who are more interested in multimedia playback, and who have slow internet connections, that's probably correct. In other contexts, where you're processing a high volume data stream and only occasionally use audio, it might be the other way around.
In any case, this all sounds like balderdash to me. Modern processors are easily powerful enough to handle gigabit networking and MP3 playback. I've played back full quality video and audio on a 700 MHz PIII processor, and done gigabit networking on processors much slower than the difference between that and even a slow modern processor. So it sounds to me like there's something screwy going on here. But I've certainly seen plenty of screwy network drivers.
I neither run Windows nor know anything about the Windows kernel, so this is of little more than amusement to me. I'd be inclined to guess from what I'm reading that something in the audio code is holding onto a lock excessively long, and that particular lock is preventing the network stack from running (even processing interrupts), for long enough that the card is being forced to drop a lot of inbound packets. I wouldn't be too surprised if this really is only on the receive side, but that's going to have to be tested by someone who actually runs Vista. The best way to test it would be to blast large UDP or ICMP packets from the Vista box to something that is known to be able to receive fast enough, and see how much actually gets sent with and without audio playing. Don't use TCP because if there is a performance loss you won't know whether it's because ACK's are being dropped or because the Vista box simply can't transmit fast enough.
Likewise, to test receive performance, use UDP from a fast system (or even ICMP, if you can get information out of the kernel about what it's receiving) rather than TCP. There are too many things that can screw up TCP performance.
Of course, none of that will really tell you what's going on inside; it will just give you better measurements. But you can't fix something if you can't measure it.
I presume this is through a dealer finance program, not through a bank. You're not getting a free ride; the price is higher than if you paid cash or arranged outside financing.
You know those deals "0% financing or $1500 cash back"? Well, that's basically a way of increasing the effective interest rate. Essentially, you're prepaying $1500 in interest by going with that "0% financing", so it's not really 0% financing at all.
You might be able to get 13% on your investment if you're lucky, but you're not likely to get anything close to that on a consistent basis.
The Inspiron 8000-8200 don't have integrated video. They're a little dated, but I'm quite happy with my 8200 which I built from parts (I wanted to reuse the very nice 1600x1200 screen that I had upgraded my previous 8000 to). The 8200 is reasonably easy to service as these things can go, and it's easy to get parts on eBay.
Styling seems to be very in for laptops these days. However, I think the current trend of 17" WXGA screens is absurd. My processor is slow by today's rather excessive standards (P4-1.8, although I could apparently go up to 2.6), but 1.5 GB of RAM really helps.
You can get real wide-angle glass, although you'll need to get the Canon 5D (or 1Ds/1DsII), which have full-frame sensors. The Kodak DCS (they make, or at least used to make, Canon and Nikon mount versions) SLR's are also full-frame.
The dynamic range issue, alas, is very real, and it makes shooting sunsets (and other very high dynamic range situations) very difficult, where good color print film has no difficulty. It produces weird color fringes around the sun that look very unnatural. Shooting RAW helps, since the camera actually captures much more information than can be represented in 8 bits. I have a hacked up version of dcraw that does a much better job in some situations. On the plus side, digital noise is usually much less severe than film grain, although at high ISO it's more objectionable.
As far as cost goes, certainly a digital body is much more expensive than a film body (a Digital Rebel is in the same price range as an EOS 3, and a 30D is in the same range as a 1V, and it goes up from there). However, the consumables costs of a digital body is much, much lower. A roll of 36 exposures costs between $10 and $20 (say, $.25~$.50/frame), counting the purchase price and the processing price, and storage is a hassle. If you shoot RAW+large JPEG, a 20D frame is about 12 MB. Assuming storage costs $1/GB, that's $.01/frame, comfortably better than an order of magnitude cheaper. You can also discard frames you don't want to keep. Making copies (I didn't say "prints") is also much cheaper unless you already have a scanner, and a good film scanner costs at least $500.
Assuming that a film body costs $300 and a digital body costs $700 (comparing the Elan to the Digital Rebel -- the features and build quality of the Digital Rebel are much closer to the Elan than to the film Rebel), after 2000 frames or so you're ahead with the digital body.
As for the "better film" argument, while you can't outright upgrade the digital body (it's stuck with the sensor it came with), you effectively have a lot of different film types with you (choosing ISO and processing parameters) and can switch with every frame.
I'm told by our OS X packager (Gutenprint) that the standard OS X installer template must have a license that it will present and ask for agreement -- essentially, the click-through license is a mandatory field, although you can put whatever you like in there. I don't remember offhand exactly what we put in there (since it has been a while since I've installed it on OS X), but I had been thinking of putting the GPL along with a statement up front that merely installing or using Gutenprint does not require agreement with the GPL and that therefore by clicking "I agree" the user is not stating agreement with the GPL.
It says something that Apple doesn't think anyone's ever going to distribute a software package that doesn't carry a EULA, though.
As I see it, the history of computing is one of repeated waves that crash against the beach as the next wave gathers behind it. In terms of corporate IT, this is marked by shifts back and forth between centralized and distributed computing, but the phenomenon encompasses all of computing.
What happens in each phase is that the new model is adopted first at the grass roots -- individual users or small groups see the "new thing" as a way to get something done, and start using it under the radar. It's frequently said that something needs to have an order of magnitude better price/performance ratio in order to be adopted over the existing way of doing things -- these waves are sufficiently radical that that's exactly what happens. This is the "decentralized" phase.
Word spreads about the new way of doing things, and it gets used more and more widely. Eventually, the corporate bureaucracy realizes what's going on and takes over, partly because it saves each group from having to support itself and partly because the "new way" becomes more sophisticated and more powerful, and outgrows the ability of small, isolated groups to manage. This is the "centralized" phase.
Time marches on, people are chafing under the iron grip of the corporate bureaucracy, and something new comes along...
On the supplier side, what happens is that the new technology is just as capable as the old, for a fraction of the cost (and size, and power requirement, and everything else), so the supplier wants to incorporate the new technology.
As I see it, the actual specific waves that we've experienced are:
Mainframes: these were the first computers receiving wide acceptance outside of the research lab. As much as people like to joke about them, they were a huge advance over what was used before -- hand calculation, slide rules, mechanical or electro-mechanical adding machines, and the like.
Minicomputers: this had its roots in the 1960's, but really took off only the 1970's. These were often sold and used as "departmental" machines, and were much smaller and simpler to operate than mainframes. They could also be used interactively.
Workstations: as microprocessors became more powerful, they approached (and often exceeded) minicomputers in processing power. At this point it was possible for someone to actually have a computer on his or her desk, under that person's control that was almost as capable as a departmental minicomputer for a fraction of the cost. A lot of people found that these workstations actually made very nice servers. I remember (in the late 1980's/early 1990's) helping to install a bunch of "pizza boxes" (SPARCstation I and II-class machines) with small, cheap SCSI drives that could do just about everything a VAX 8800 or Sun 4/490 could do. What they couldn't do was more than made up for by the fact that we could put 3-4 of these, plus several GB of disk, in a 19" rack that could otherwise hold a 4/490. For ease of administration, we wound up cloning these machines and coming up with a couple of standard loads. We discouraged people from modifying their own machines to make it easier to administer the network.
PC's: cheap microprocessors became even cheaper and just as powerful as workstation-class processors, while small commodity disk drives gained capacity. Eventually someone came up with the bright idea that machines like this could have substantial use of their own, not just as desktops.
What happened on the vendor side, of course, is that each wave offered tremendous cost savings. It was a lot easier to design a system around a microprocessor than to use LSI or discrete logic. Meanwhile, these microprocessor-based systems became more sophisticated, with increased I/O capacity and more processors (Sun E10K was basically 64 UltraSPARC microprocessors with lots of memory and I/O capacity), and commodity microprocessors, motherboards, memory, and disk attacked proprietary systems. Each wave
The (by now quite ancient) Inspiron 8000 and related series are very easy to upgrade. I had an 8000 for about 4 years, during which time I replaced the keyboard, CPU (upgrade), LCD panel (upgrade), display (twice, hinges broke), fixed optical drive (CD/DVD->DVD+-RW), hard drive (twice, once due to failure and once for upgrade). Last year I got the shell+mobo of an 8200. I reused the hard drive, optical drive, and display from my 8000, got a keyboard, processor, memory, and the cheapest graphics card I could find (I don't do games beyond minesweeper and use a very minimalist KDE theme with no visual effects). Works quite nicely, and I have a nice machine for the money that I can fix and upgrade.
Image quality is limited by optics and by the size (overall size -- in millimeters) of the sensor. With trashy optics and a tiny sensor, it doesn't matter how many (ever tinier) pixels one jams onto that sensor; the image quality just won't be there.
I advise people who want a point and shoot (which has a much bigger sensor than a cell phone, but much smaller than an SLR) to not bother with more than 3 megapixels or so. There are some expensive "advanced" non-SLR digital cameras that have sensors comparable to SLR sensors, but most point and shoot cameras have sensors no more than 1/2" on the diagonal. Most digital SLR's are about 1.2" diagonal, and 35 mm SLR's (such as film cameras and the EOS 5D and 1Ds digital cameras) are a bit less than 2" on the diagonal.
The basic issue here is the pixel size, which (along with lens speed) controls how much light the pixel can gather. The larger the pixel, the more light it can gather and the less noise it will have. There's also the wavelength of light to consider; as the pixel gets smaller, the ability to resolve between neighboring pixels becomes less.
The sweet spot for digital SLR's with APS-C sensors with 1.5 or 1.6 cropping factors (such as the low to midrange Canon and Nikon cameras) seems to be about 8-10 megapixels. Canon's latest offering (the 30D) stayed at 8 MP. The Nikon D200 is 10 megapixels, but it's noisier at high ISO settings than the 20D/30D. This would suggest that full-frame (35 mm) digital SLR's won't get much above 20 megapixels (based on pixel size), and you'll have to go to medium format to get much more than that. If the Foveon sensor ever gets perfected the marketing numbers will triple (since each position would have a sensor for each color), but the grid won't change.
It's possible to reduce noise by lowering the effective ISO (in other words, allowing more light into the sensor by requiring longer exposures). So while the EOS 20D has excellent noise performace even at ISO 800, a typical point and shoot (with its tiny sensor) will be very noisy above ISO 100 or 200.
Finally, there's the matter of the lens. My own tests suggest that I only get the full 8 megapixel resolution out of the 20D if I use a good lens (such as the 85 f/1.8 or 200 f/2.8 prime lenses), well stopped down and very carefully focused, and otherwise in good conditions (on a tripod or with a very short exposure). I recently took a shot at sunset with a 1 second exposure at f/16 with my 200 mm lens and there was very sharp single pixel detail. Even slight blur will very quickly reduce the useful pixel count; if it's blurred to the extent that there's no useful detail at less than 2 pixel resolution, you're effectively at the 2 megapixel level.
So what does all of this mean? Camera phones have tiny sensors, with cheap lenses, and can't have long exposures. However many pixels the sensor may have, I'd be surprised if the effective resolution of the output is more than a few hundred thousand pixels.
It's a neat idea, but how well is it going to scale to 10,000 or more objects (say, 6-16 megapixel images)? A lot of interfaces of this kind seem to work very well on small sets of images (or whatever), but founder when they scale up.
I only have about 12000 images, but professionals might easily accumulate 50,000 or more images per year, in some cases using medium format backs with 35 megapixels and 16 bit color depth. While the storage requirements for something like that might still be a bit daunting (each image of that size would be 200 MB if stored in uncompressed TIFF format, so this would be 5 TB/year), any good image management tool has to handle large scale.
I like KPhotoAlbum (formerly KimDaBa) myself. While not particularly elegant visually, it's fast and has excellent search capabilities and metadata organization.
Actually, early in the 5-year period the NAO was positive most of the time (in particular, the winter of 2001-2002 had an extreme positive NAO all winter, which led to the eastern US being extremely warm), which most likely corresponded with strong troughing over Greenland. The next three winters were much more dominated by a negative phase of the NAO. For Greenland that would almost certainly translate into net warming over the 5-year period (just like one could argue that it demonstrated strong cooling over the eastern US).
It certainly does make for interesting speculation about what would happen if the thermohaline circulation were to shut off, as some people predict will happen if ice melt becomes too rapid. One possible outcome would be that the NAO would more or less lock in positive (i. e. a deep trough over the midlatitudes in the Atlantic), which would typically result in strong ridging (much warmer weather) over the eastern US.
There are, if you can believe it, real flashlights that are almost as complicated. This one has a thirteen page user's guide in addition to a quick start guide and a reference card.
(The LED Museum is a site that all nerds should bookmark. I believe it's been Slashdotted before.)
Doesn't quite make it into a 10D -- as you note, the body is very different, and the FPS and buffering just aren't there. However, certainly the firmware hack does enable some very useful functionality. I store only a small JPEG in my RAW files and sometimes use mirror lockup; flash exposure compensation is also very useful.
The limitations of the Rebel aside, it's a great camera. In addition to the landscape work I enjoy, I also do event photography for a club that I belong to. As limiting as 2.5 fps and 4 frames may seem to be, I rarely run into problems with that, despite a distinctly run and gun shooting style (usually flash recharge gets me first, even at ISO 800). I wouldn't consider a 10D; it just doesn't have enough over the Rebel to justify it, and the Rebel has one objective advantage -- the ability to use EF/S lenses. The 20D is another matter, although the Rebel's easily good enough that I'm not about to shell out $1400 after only a year.
Indeed. I've been doing some work on the "bag" that I carry for different things, and I've recently added a 50 f/1.8, 85 f/1.8, and 200 f/2.8, and remove the 75-300 f/4-5.6 IS that I carry to events (I've kept the 28-135 f/3.5-5.6 and 12-24 f/4.5-5.6). When I first tried it, the results were a big improvement; the 200 f/2.8 is much brighter than the 75-300, so it's easier to focus in a dark room and needs less flash power, and this lens wide open is still sharp (it's a Canon "L" lens). The 50 and 85 aren't L lenses, but they're still wonderfully sharp and even faster, and they're compact. It does mean more lens changing, but it's not too bad.
I don't want to ditch the 28-135 -- having zoom in this range (equivalent to 45-200 or so) is very useful, and this lens autofocuses well -- but the 75-300 seems to be better as a wildlife lens outdoors, where the stabilizer helps a lot and the extra length is more important.
When I go to weddings, I always take fast prime lenses with me, even when I'm not the official photographer. I particularly enjoy taking natural-light photographs at the ceremony; they look so much better than the blown out flash photos that I've often seen from professionals (they use, or at least used to use, slow film with extremely stopped-down lenses for depth of field). My style is quite different from most pros, and if I'm not the hired photographer I prefer not to overlap in what I take, so I shoot a lot of long-telephoto close cropped candids and available-light ceremony shots.
It certainly does take more skill to shoot an event well. I do think that the mystique around wedding photographers is a bit overblown (unless you really like the style of your photographer, and there are some excellent ones), but simply knowing how to handle an SLR isn't enough to do a good job. You do need to have some understanding of lighting and composition, and either the equipment to back it up or consciously limiting yourself to what your equipment can do. At our wedding, we put disposable cameras on the tables. Most of the shots weren't much good -- people trying to photograph dancing and such. At one table, the shots were well-composed and properly lit. Not exactly coincidentally, two very good photographers were seated at that table, who knew the limits of the equipment and didn't try to do something that wouldn't work.
The depth of field depends upon the F stop (absolute aperture divided by focal length) and the magnification of the image on the sensor/film, to be more precise. The aperture (F-stop) is also a measure of how bright the lens is -- a factor of 2 larger F-stop (F/4 vs. F/2) means 1/4 as much as light.
Cheap digital cameras frequently have remarkably bright lenses -- f/2.8 is very common. The depth of field is quite large because the magnification of the image on the sensor is very low, because the sensor is so small. Even in macro mode, the magnification is less than on a 35 mm SLR (much less a medium or large format camera), because the sensor is small.
Obviously, saying that you only get a narrow depth of field with a 35 mm SLR is nonsense. If you use a wide angle lens (with less magnification), or you stop way down (making the aperture tiny, or a large F number), you can get a very deep field of sharp focus. The problem is that a telephoto lens stopped down that much won't gather much light, so you need a very long exposure. A compact camera at the equivalent of 105 mm (but actually 20 mm) at f/2.8 would require f/16 to get the same depth of field, or 5 stops (32x) slower.
What's really happening here is that the compact camera has the pixels much closer together than the 35 mm SLR. This makes the pixels much smaller and somewhat less sensitive to light, so you do need a bit more exposure. Of course, tiny pixels do mean more noise and possible diffraction problems, so it isn't free.
I agree with you that the Rebel is crippled in some ways compared to the 1D and even 10D and 20D (such as frame rates, buffering, and controls), but in image quality, the Rebel is identical to the 10D and is mighty close to the 1D. It uses the same sensor as the 10D.
The Rebel can do just as good a job as the 10D, as long as you understand its limitations (e. g. it's not a good camera for photographing motor sports, due to the specific limitations). However, I've used it to photograph fast action, and just so long as I'm aware of the 4 frame limit and work around it, it does fine. That said, the 20D is a far better camera in that regard.
SLR means "single lens reflex". The "reflex" bit is the mirror that either sends the light to the viewfinder (the focusing screen) or allows it through to the shutter and then the film. You can have an SLR with a fixed lens, and some do exist; they used to be called "bridge" or "ZLR" (zoom lens reflex).
The closer the flash is to the lens (and therefore in general the smaller the camera), the more of a problem you'll have with redeye, which is the direct reflection of light from the back of the retina. Using an SLR with a shoe-mounted flash will generally decrease redeye, because the angle between the flash, the eye, and the lens will be much greater. The way I shoot I have a lot of problems with redeye; I like to take candid portraits from very long distances with long lenses, so even a shoe-mount flash is too close to the lens, and I should really use a flash bracket. When I take a candid portrait with a 300 mm lens from 30 feet away, the eyes often glow bright red.
Photo studios don't have problems with this because the lights are far away from the camera, and the shooting distance is fairly short.
That's really the only solution for the problem. You can use software to remove redeye after the fact, but pretty much all compact cameras will have this problem.
You might be surprised what pros are using, actually. I've interacted with three professionals this year: at a bat mitzvah, a wedding, and an event covered by a press photographer. Two of them used digital SLR's -- the bat mitzvah photographer (who appeared to be the highest price one) was using a Nikon D2H, which is a 4 (!) megapixel camera, and the press guy was using a 4 or 6 megapixel camera. The guy at the wedding was using a high-end Sony Cybershot -- this is a 7 or 8 megapixel camera, not an SLR, but still a high-end camera. And another friend of mine, who just turned professional, is shooting a Nikon D100, I believe. This is also a 6 megapixel camera. He does a lot of landscapes and nature photography.
The pro at the bat mitzvah was clearly an experienced event photographer; his lenses, flash gear, and assistant were clearly set up for weddings and the like. The fact that he felt that 4 megapixels was sufficient is a sure indication that consumers buying compact cameras with 8 megapixels are paying for something they simply don't need. It may actually be harmful, since the pixels are so small that they're less sensitive, requiring more light (i. e. longer shutter speeds) to reduce noise. The D2H is truly a professional camera (in terms of body construction and operating functionality), and it's a current model. The Rebel isn't, but that's not due to image quality -- the image quality is right up there with just about anything -- but rather due to operating features and construction (it's less rugged than, say, the 1N/1D; it only delivers 2.5 frames/second for 4 frames; the controls are crippled in various minor ways that don't matter too much to amateurs but are a big deal to working professionals). But I suspect a lot of pros are using the 10D and now the 20D rather than spending a lot more money on the 1D.
Would I like to have a 1Ds II? With 16 megapixels and a full frame sensor, sure:-) But at $8000 it's more than a bit out of my price range. To be honest, the full frame is probably more important, since it preserves the wide angle lenses. It also needs lots of pixels, since otherwise it would lose detail with long lenses due to the pixels being larger. Someone commented to me that the 1Ds II is Canon's entry into the medium format market, and their point is well taken -- the Leaf medium format back, which is only (!) 22 megapixel, is about $20K.
The truth is that the detail I can get from a 6 megapixel frame is quite astounding. I have to be very careful with my technique to even fully exploit that resolution -- use a sharp lens (or at least an optimal aperture with a less than stellar lens), along with a tripod, image stabilization, or very short exposure. It's possible to get more detail from very sharp film (such as Fuji Velvia or Kodak Ektar 25, and surely Fuji Provia), but even Ektar 25, which is probably the sharpest and finest grained film ever made, is noisier than digital at ISO 400 or even 800.
On my last trip, I brought along my EOS 1N and a few rolls of Ektar 25 for a very specific purpose -- to use my 12-24 mm Sigma for real hyperwide shots; there's still a big difference between 12 mm and 19 mm (the 35 mm equivalent of 12 mm on the Rebel), even though 19 mm is still considered superwide. Everything else was digital.
BTW, on the subject of the LCD display, I just don't like them for composing; I prefer to compose through an optical viewfinder. Partly it's sharpness/resolution for critical focusing, partly it's what I'm used to, and partly it's because LCD panels respond slowly. I think it's also less distracting, because I have my eye right up against it. SLR's do of course have mirror delay, but even the Rebel's quite good in that department, perhaps not quite as good as the 1N, but on a par with the A2.
Good suggestions by Skyshadow; nothing I disagree with.
I have a Digital Rebel myself, with a collection of glass ranging from a Sigma 12-24 to a Canon 75-300 IS and several prime (fixed focal length) lenses. It's a great camera; the autofocus is better than the EOS 1N (the previous generation pro film body), and it's almost as responsive in a lot of ways. The 20D is a great camera, from what little I've used (fired a few frames at a show).
I bought a Canon A75 for my wife, at the much-maligned Best Buy (it was on sale, and the price was reasonable). I specifically wanted her to be able to share my memory cards, and Canon's about the only P&S maker that uses compact flash. It's also a very nice camera for what it is. It can't do what the Rebel can in sheer flexibility, and shooting RAW with the Rebel lets me process the images to get more highlight detail, but it's surprisingly flexible if you want to get more creative. It's a bit bigger than the S series, and not as flexible as the G series, but it's a good camera for a good price.
I completely agree about the megapixels and zoom range. One thing to beware of is "digital zoom", which is a polite word for "cropping". A 3x digital zoom effectively converts a 3 MP camera to a 1/3 MP camera (3x linear zoom, 1/9 the area). I suppose if you want to just stick it into a printer with a card reader that's OK, but understand what's going on. I also agree about image stabilization with a 10x zoom! This is typically equivalent to about 400 mm on a 35 mm camera, which means that to have a decent shot of getting a good hand held shot you need an exposure of 1/400 second or faster. My sister in law bought one of those cameras. Shooting indoors, even with flash (in a brightly lit room), I was unable to get a sharp image, despite being careful in my technique. Image stabilization helps a lot -- with that focal length, if you're reasonably careful, you can get a sharp shot at 1/50 ~ 1/100.
One other point -- look at the 35 mm equivalent for the zoom range. Most of these cameras have more extension on the long end than on the wide end; the wide end is typically equivalent to 35~38 mm, which is only barely "wide angle". Some compact cameras have a wide end of 24-28 mm, which is markedly wider angle, and usually enough for good landscapes. The downside is that a 3x zoom with a wide end of 28 (equivalent) only goes to 90 mm or thereabouts, which is reasonable for portraits but isn't exactly much good for real telephoto work. 35 mm lenses go out to 105 or so, which isn't markedly longer, so I would suggest getting a compact with a wide end equivalent to 28 mm or so. The actual focal length isn't very useful for comparison, since sensors aren't all of the same size and the angle of coverage depends upon the lens focal length and the sensor size, so "35 mm equivalent" is the best comparison.
Re:And this surprises us because...
on
NYT on EA Games
·
· Score: 4, Insightful
Employers should (for the sake of the business) try to get the most they can out of their workforce. However, at least in a typical high tech development environment, that doesn't mean the most hours of work. Practices such as those described suggest that either management assumes that there will always be some incremental gain in output for another hour worked, or have other reasons (such as a power trip, or some wild notion of "team building").
My own take, when I was a manager at a large company you've heard of, was that I wanted people to work smart rather than merely working hard. Granted, there are rare times when it's necessary to put in more time (late nights, weekends, off hours) to complete a key short-term deliverable, but people working long hours constantly isn't a sign of good management, but rather poor management. Employees who get tired will start making mistakes, and that's expensive (remember that the later a bug is found the more expensive -- by a large margin -- it is to fix.
The other key point here is that hiring (including the salaries of the hiring manager, HR, interviewers, and training) is expensive. In my experience, it takes a while in my line of business (system development) for even a very good new hire to really pay their way. It has also been said that the difference in productivity between a top programmer and a marginal programmer is 100 to 1. If you work from those assumptions, the way to extract the maximum useful output is to hire good people, encourage them to work efficiently, and otherwise treat them well.
I like to say that if someone who reported to me accomplished everything they were expected in a high quality manner to in 10 hours a week I'd have no problem with it. My own experience is that some people like to work in quick bursts, some people really do like to put in a lot of hours, and some people simply work steadily. However they prefer to do so is fine by me. I do have a bit of a problem with people who do the same thing over and over again (often spending a lot of time on it) without trying to find a better way of doing it. I like to say that I'm too lazy to do the same thing twice. Computers don't get upset if they're asked to do the same thing over and over again, and I prefer to move on to something new.
Obviously, there are people who don't see it that way. Rest assured, though, there are companies and managers who do take a reasonable approach to this, and that the whole industry isn't a sweatshop.
A click to raise option is fine -- as long as it's an option. Personally, I use a keyboard accelerator to raise windows, but with neither autoraise nor click to raise enabled (in KDE), and with mouse strictly follows focus. Yes, that does mean that I sometimes operate in partially obscured windows, but that's by choice.
I recall reading way back when that sulphur goes through an interesting variety of state changes as it's heated -- it first becomes a free-flowing liquid, then the viscosity increases in discrete steps as it's further heated, and then it becomes a free flowing liquid again and finally boils. It never really solidifies, but the viscosity markedly increases as it's heated to the point that it becomes taffy-like or harder.
Slashdot Mirror
Just about every technological advance, by nature of its disruptiveness, harms some people and helps others, but that doesn't make it wrong. The entertainment industry can either fight a rearguard action to delay the inevitable, hurting a lot of people along the way, or embrace the reality that copying data is cheap and easy and find new ways to profit from the situation. It's always been the mark of a good business person that he or she finds opportunities rather than complaining about the situation. Right now it looks like EMI is starting to understand the situation; let's see how long it takes the rest of the industry (and the movie industry) to figure it out.
If it turns out that a handful of mega-stars supported by large multinational companies is not the most efficient way to deliver entertainment, I see little loss to society as a whole, and it would surely be to the benefit of a much larger set of artists.
It might be easier for Microsoft to implement it for that reason, but that doesn't apply to anyone else, none of whom started from that code base. That's not a very good reason for a standard. And StarOffice/OpenOffice have about as good support for legacy MS formats as anyone, but that didn't seem to make it any harder to adopt ODF.
It strikes me as a wee bit cynical -- not to mention rather far outside of the mainstream of conventional GPL/LGPL usage -- to say "we're releasing this under the LGPL, but if you know what's good for you, you'll download it from us". It's not a violation of the LGPL for Novell to do this with their own code, obviously (at least, obvious to me, but I'm not a lawyer), but it's going to make it awfully dangerous for any other project to reuse this code.
Also, what distribution formats will be covered by the patent protection claim? Source? Binary-only? Does this mean that only operating systems that Novell chooses to compile Moonlight for will be "safe"?
If it sounds like I don't trust anything Microsoft touches with a ten foot pole, that's correct -- Microsoft hasn't given anyone much reason to trust them. This looks like yet another attempt to bust the GPL/LGPL, this time in a way that keeps them safe from the FSF -- by using someone who actually owns the copyright on the code.
I would urge the people making the decisions at Novell to go back and read the story of King Midas. Very, very carefully.
1) Your people come first.
2) Support your people, come hell or high water.
3) See #1.
4) Keep any distractions that aren't absolutely necessary away from your folks.
5) Don't sign up for any critical technical work (you'll just slip the project).
6) Oh, in case I forgot to mention, fight like crazy for your people.
7) If your people are working overtime/weekends for more than a well-bounded, short term crunch, it's a sign of a problem, not a good sign at all. Kick your folks out and figure out what's going wrong, and fix it.
8) Don't constantly look over everyone's shoulder.
I'm no fan of Microsoft, but there's a book published by Microsoft Press by Steve Maguire or Steve McConnell (whichever one of them didn't write Code Complete) about technical management. It's actually very good indeed.
The combination of manager and architect sounds like a very dangerous one to me. It's going to be very difficult for you to get honest technical feedback from people who report to you, and architects need that kind of candid feedback. An architectural error that goes unchallenged for political reasons is bad news. When I was a manager, I once led a software integration team (a more suitable role for a manager), and even that caused some tension. Not with the person who happened to report to me -- she had no trouble pushing back when needed -- but with another engineer who found it uncomfortable when I had to go around and get commitments from people. I was actually selected for that role because initially no one on the team reported to me (I was the one leadership/management role who didn't have an interest in a particular component within the project), but things change...
Whether 100 Mb/sec is "just fine" depends upon what you're doing. Likewise, whether an occasional audio skip matters or not depends upon what you're doing.
For home users who are more interested in multimedia playback, and who have slow internet connections, that's probably correct. In other contexts, where you're processing a high volume data stream and only occasionally use audio, it might be the other way around.
In any case, this all sounds like balderdash to me. Modern processors are easily powerful enough to handle gigabit networking and MP3 playback. I've played back full quality video and audio on a 700 MHz PIII processor, and done gigabit networking on processors much slower than the difference between that and even a slow modern processor. So it sounds to me like there's something screwy going on here. But I've certainly seen plenty of screwy network drivers.
I neither run Windows nor know anything about the Windows kernel, so this is of little more than amusement to me. I'd be inclined to guess from what I'm reading that something in the audio code is holding onto a lock excessively long, and that particular lock is preventing the network stack from running (even processing interrupts), for long enough that the card is being forced to drop a lot of inbound packets. I wouldn't be too surprised if this really is only on the receive side, but that's going to have to be tested by someone who actually runs Vista. The best way to test it would be to blast large UDP or ICMP packets from the Vista box to something that is known to be able to receive fast enough, and see how much actually gets sent with and without audio playing. Don't use TCP because if there is a performance loss you won't know whether it's because ACK's are being dropped or because the Vista box simply can't transmit fast enough.
Likewise, to test receive performance, use UDP from a fast system (or even ICMP, if you can get information out of the kernel about what it's receiving) rather than TCP. There are too many things that can screw up TCP performance.
Of course, none of that will really tell you what's going on inside; it will just give you better measurements. But you can't fix something if you can't measure it.
I presume this is through a dealer finance program, not through a bank. You're not getting a free ride; the price is higher than if you paid cash or arranged outside financing.
You know those deals "0% financing or $1500 cash back"? Well, that's basically a way of increasing the effective interest rate. Essentially, you're prepaying $1500 in interest by going with that "0% financing", so it's not really 0% financing at all.
You might be able to get 13% on your investment if you're lucky, but you're not likely to get anything close to that on a consistent basis.
The Inspiron 8000-8200 don't have integrated video. They're a little dated, but I'm quite happy with my 8200 which I built from parts (I wanted to reuse the very nice 1600x1200 screen that I had upgraded my previous 8000 to). The 8200 is reasonably easy to service as these things can go, and it's easy to get parts on eBay.
Styling seems to be very in for laptops these days. However, I think the current trend of 17" WXGA screens is absurd. My processor is slow by today's rather excessive standards (P4-1.8, although I could apparently go up to 2.6), but 1.5 GB of RAM really helps.
You can get real wide-angle glass, although you'll need to get the Canon 5D (or 1Ds/1DsII), which have full-frame sensors. The Kodak DCS (they make, or at least used to make, Canon and Nikon mount versions) SLR's are also full-frame.
The dynamic range issue, alas, is very real, and it makes shooting sunsets (and other very high dynamic range situations) very difficult, where good color print film has no difficulty. It produces weird color fringes around the sun that look very unnatural. Shooting RAW helps, since the camera actually captures much more information than can be represented in 8 bits. I have a hacked up version of dcraw that does a much better job in some situations. On the plus side, digital noise is usually much less severe than film grain, although at high ISO it's more objectionable.
As far as cost goes, certainly a digital body is much more expensive than a film body (a Digital Rebel is in the same price range as an EOS 3, and a 30D is in the same range as a 1V, and it goes up from there). However, the consumables costs of a digital body is much, much lower. A roll of 36 exposures costs between $10 and $20 (say, $.25~$.50/frame), counting the purchase price and the processing price, and storage is a hassle. If you shoot RAW+large JPEG, a 20D frame is about 12 MB. Assuming storage costs $1/GB, that's $.01/frame, comfortably better than an order of magnitude cheaper. You can also discard frames you don't want to keep. Making copies (I didn't say "prints") is also much cheaper unless you already have a scanner, and a good film scanner costs at least $500.
Assuming that a film body costs $300 and a digital body costs $700 (comparing the Elan to the Digital Rebel -- the features and build quality of the Digital Rebel are much closer to the Elan than to the film Rebel), after 2000 frames or so you're ahead with the digital body.
As for the "better film" argument, while you can't outright upgrade the digital body (it's stuck with the sensor it came with), you effectively have a lot of different film types with you (choosing ISO and processing parameters) and can switch with every frame.
I'm told by our OS X packager (Gutenprint) that the standard OS X installer template must have a license that it will present and ask for agreement -- essentially, the click-through license is a mandatory field, although you can put whatever you like in there. I don't remember offhand exactly what we put in there (since it has been a while since I've installed it on OS X), but I had been thinking of putting the GPL along with a statement up front that merely installing or using Gutenprint does not require agreement with the GPL and that therefore by clicking "I agree" the user is not stating agreement with the GPL.
It says something that Apple doesn't think anyone's ever going to distribute a software package that doesn't carry a EULA, though.
As I see it, the history of computing is one of repeated waves that crash against the beach as the next wave gathers behind it. In terms of corporate IT, this is marked by shifts back and forth between centralized and distributed computing, but the phenomenon encompasses all of computing.
What happens in each phase is that the new model is adopted first at the grass roots -- individual users or small groups see the "new thing" as a way to get something done, and start using it under the radar. It's frequently said that something needs to have an order of magnitude better price/performance ratio in order to be adopted over the existing way of doing things -- these waves are sufficiently radical that that's exactly what happens. This is the "decentralized" phase.
Word spreads about the new way of doing things, and it gets used more and more widely. Eventually, the corporate bureaucracy realizes what's going on and takes over, partly because it saves each group from having to support itself and partly because the "new way" becomes more sophisticated and more powerful, and outgrows the ability of small, isolated groups to manage. This is the "centralized" phase.
Time marches on, people are chafing under the iron grip of the corporate bureaucracy, and something new comes along...
On the supplier side, what happens is that the new technology is just as capable as the old, for a fraction of the cost (and size, and power requirement, and everything else), so the supplier wants to incorporate the new technology.
As I see it, the actual specific waves that we've experienced are:
What happened on the vendor side, of course, is that each wave offered tremendous cost savings. It was a lot easier to design a system around a microprocessor than to use LSI or discrete logic. Meanwhile, these microprocessor-based systems became more sophisticated, with increased I/O capacity and more processors (Sun E10K was basically 64 UltraSPARC microprocessors with lots of memory and I/O capacity), and commodity microprocessors, motherboards, memory, and disk attacked proprietary systems. Each wave
The (by now quite ancient) Inspiron 8000 and related series are very easy to upgrade. I had an 8000 for about 4 years, during which time I replaced the keyboard, CPU (upgrade), LCD panel (upgrade), display (twice, hinges broke), fixed optical drive (CD/DVD->DVD+-RW), hard drive (twice, once due to failure and once for upgrade). Last year I got the shell+mobo of an 8200. I reused the hard drive, optical drive, and display from my 8000, got a keyboard, processor, memory, and the cheapest graphics card I could find (I don't do games beyond minesweeper and use a very minimalist KDE theme with no visual effects). Works quite nicely, and I have a nice machine for the money that I can fix and upgrade.
Image quality is limited by optics and by the size (overall size -- in millimeters) of the sensor. With trashy optics and a tiny sensor, it doesn't matter how many (ever tinier) pixels one jams onto that sensor; the image quality just won't be there.
I advise people who want a point and shoot (which has a much bigger sensor than a cell phone, but much smaller than an SLR) to not bother with more than 3 megapixels or so. There are some expensive "advanced" non-SLR digital cameras that have sensors comparable to SLR sensors, but most point and shoot cameras have sensors no more than 1/2" on the diagonal. Most digital SLR's are about 1.2" diagonal, and 35 mm SLR's (such as film cameras and the EOS 5D and 1Ds digital cameras) are a bit less than 2" on the diagonal.
The basic issue here is the pixel size, which (along with lens speed) controls how much light the pixel can gather. The larger the pixel, the more light it can gather and the less noise it will have. There's also the wavelength of light to consider; as the pixel gets smaller, the ability to resolve between neighboring pixels becomes less.
The sweet spot for digital SLR's with APS-C sensors with 1.5 or 1.6 cropping factors (such as the low to midrange Canon and Nikon cameras) seems to be about 8-10 megapixels. Canon's latest offering (the 30D) stayed at 8 MP. The Nikon D200 is 10 megapixels, but it's noisier at high ISO settings than the 20D/30D. This would suggest that full-frame (35 mm) digital SLR's won't get much above 20 megapixels (based on pixel size), and you'll have to go to medium format to get much more than that. If the Foveon sensor ever gets perfected the marketing numbers will triple (since each position would have a sensor for each color), but the grid won't change.
It's possible to reduce noise by lowering the effective ISO (in other words, allowing more light into the sensor by requiring longer exposures). So while the EOS 20D has excellent noise performace even at ISO 800, a typical point and shoot (with its tiny sensor) will be very noisy above ISO 100 or 200.
Finally, there's the matter of the lens. My own tests suggest that I only get the full 8 megapixel resolution out of the 20D if I use a good lens (such as the 85 f/1.8 or 200 f/2.8 prime lenses), well stopped down and very carefully focused, and otherwise in good conditions (on a tripod or with a very short exposure). I recently took a shot at sunset with a 1 second exposure at f/16 with my 200 mm lens and there was very sharp single pixel detail. Even slight blur will very quickly reduce the useful pixel count; if it's blurred to the extent that there's no useful detail at less than 2 pixel resolution, you're effectively at the 2 megapixel level.
So what does all of this mean? Camera phones have tiny sensors, with cheap lenses, and can't have long exposures. However many pixels the sensor may have, I'd be surprised if the effective resolution of the output is more than a few hundred thousand pixels.
It's a neat idea, but how well is it going to scale to 10,000 or more objects (say, 6-16 megapixel images)? A lot of interfaces of this kind seem to work very well on small sets of images (or whatever), but founder when they scale up.
I only have about 12000 images, but professionals might easily accumulate 50,000 or more images per year, in some cases using medium format backs with 35 megapixels and 16 bit color depth. While the storage requirements for something like that might still be a bit daunting (each image of that size would be 200 MB if stored in uncompressed TIFF format, so this would be 5 TB/year), any good image management tool has to handle large scale.
I like KPhotoAlbum (formerly KimDaBa) myself. While not particularly elegant visually, it's fast and has excellent search capabilities and metadata organization.
Actually, early in the 5-year period the NAO was positive most of the time (in particular, the winter of 2001-2002 had an extreme positive NAO all winter, which led to the eastern US being extremely warm), which most likely corresponded with strong troughing over Greenland. The next three winters were much more dominated by a negative phase of the NAO. For Greenland that would almost certainly translate into net warming over the 5-year period (just like one could argue that it demonstrated strong cooling over the eastern US).
It certainly does make for interesting speculation about what would happen if the thermohaline circulation were to shut off, as some people predict will happen if ice melt becomes too rapid. One possible outcome would be that the NAO would more or less lock in positive (i. e. a deep trough over the midlatitudes in the Atlantic), which would typically result in strong ridging (much warmer weather) over the eastern US.
There are, if you can believe it, real flashlights that are almost as complicated. This one has a thirteen page user's guide in addition to a quick start guide and a reference card.
(The LED Museum is a site that all nerds should bookmark. I believe it's been Slashdotted before.)
Doesn't quite make it into a 10D -- as you note, the body is very different, and the FPS and buffering just aren't there. However, certainly the firmware hack does enable some very useful functionality. I store only a small JPEG in my RAW files and sometimes use mirror lockup; flash exposure compensation is also very useful.
The limitations of the Rebel aside, it's a great camera. In addition to the landscape work I enjoy, I also do event photography for a club that I belong to. As limiting as 2.5 fps and 4 frames may seem to be, I rarely run into problems with that, despite a distinctly run and gun shooting style (usually flash recharge gets me first, even at ISO 800). I wouldn't consider a 10D; it just doesn't have enough over the Rebel to justify it, and the Rebel has one objective advantage -- the ability to use EF/S lenses. The 20D is another matter, although the Rebel's easily good enough that I'm not about to shell out $1400 after only a year.
Indeed. I've been doing some work on the "bag" that I carry for different things, and I've recently added a 50 f/1.8, 85 f/1.8, and 200 f/2.8, and remove the 75-300 f/4-5.6 IS that I carry to events (I've kept the 28-135 f/3.5-5.6 and 12-24 f/4.5-5.6). When I first tried it, the results were a big improvement; the 200 f/2.8 is much brighter than the 75-300, so it's easier to focus in a dark room and needs less flash power, and this lens wide open is still sharp (it's a Canon "L" lens). The 50 and 85 aren't L lenses, but they're still wonderfully sharp and even faster, and they're compact. It does mean more lens changing, but it's not too bad.
I don't want to ditch the 28-135 -- having zoom in this range (equivalent to 45-200 or so) is very useful, and this lens autofocuses well -- but the 75-300 seems to be better as a wildlife lens outdoors, where the stabilizer helps a lot and the extra length is more important.
When I go to weddings, I always take fast prime lenses with me, even when I'm not the official photographer. I particularly enjoy taking natural-light photographs at the ceremony; they look so much better than the blown out flash photos that I've often seen from professionals (they use, or at least used to use, slow film with extremely stopped-down lenses for depth of field). My style is quite different from most pros, and if I'm not the hired photographer I prefer not to overlap in what I take, so I shoot a lot of long-telephoto close cropped candids and available-light ceremony shots.
It certainly does take more skill to shoot an event well. I do think that the mystique around wedding photographers is a bit overblown (unless you really like the style of your photographer, and there are some excellent ones), but simply knowing how to handle an SLR isn't enough to do a good job. You do need to have some understanding of lighting and composition, and either the equipment to back it up or consciously limiting yourself to what your equipment can do. At our wedding, we put disposable cameras on the tables. Most of the shots weren't much good -- people trying to photograph dancing and such. At one table, the shots were well-composed and properly lit. Not exactly coincidentally, two very good photographers were seated at that table, who knew the limits of the equipment and didn't try to do something that wouldn't work.
The depth of field depends upon the F stop (absolute aperture divided by focal length) and the magnification of the image on the sensor/film, to be more precise. The aperture (F-stop) is also a measure of how bright the lens is -- a factor of 2 larger F-stop (F/4 vs. F/2) means 1/4 as much as light.
Cheap digital cameras frequently have remarkably bright lenses -- f/2.8 is very common. The depth of field is quite large because the magnification of the image on the sensor is very low, because the sensor is so small. Even in macro mode, the magnification is less than on a 35 mm SLR (much less a medium or large format camera), because the sensor is small.
Obviously, saying that you only get a narrow depth of field with a 35 mm SLR is nonsense. If you use a wide angle lens (with less magnification), or you stop way down (making the aperture tiny, or a large F number), you can get a very deep field of sharp focus. The problem is that a telephoto lens stopped down that much won't gather much light, so you need a very long exposure. A compact camera at the equivalent of 105 mm (but actually 20 mm) at f/2.8 would require f/16 to get the same depth of field, or 5 stops (32x) slower.
What's really happening here is that the compact camera has the pixels much closer together than the 35 mm SLR. This makes the pixels much smaller and somewhat less sensitive to light, so you do need a bit more exposure. Of course, tiny pixels do mean more noise and possible diffraction problems, so it isn't free.
I agree with you that the Rebel is crippled in some ways compared to the 1D and even 10D and 20D (such as frame rates, buffering, and controls), but in image quality, the Rebel is identical to the 10D and is mighty close to the 1D. It uses the same sensor as the 10D.
The Rebel can do just as good a job as the 10D, as long as you understand its limitations (e. g. it's not a good camera for photographing motor sports, due to the specific limitations). However, I've used it to photograph fast action, and just so long as I'm aware of the 4 frame limit and work around it, it does fine. That said, the 20D is a far better camera in that regard.
SLR means "single lens reflex". The "reflex" bit is the mirror that either sends the light to the viewfinder (the focusing screen) or allows it through to the shutter and then the film. You can have an SLR with a fixed lens, and some do exist; they used to be called "bridge" or "ZLR" (zoom lens reflex).
The closer the flash is to the lens (and therefore in general the smaller the camera), the more of a problem you'll have with redeye, which is the direct reflection of light from the back of the retina. Using an SLR with a shoe-mounted flash will generally decrease redeye, because the angle between the flash, the eye, and the lens will be much greater. The way I shoot I have a lot of problems with redeye; I like to take candid portraits from very long distances with long lenses, so even a shoe-mount flash is too close to the lens, and I should really use a flash bracket. When I take a candid portrait with a 300 mm lens from 30 feet away, the eyes often glow bright red.
Photo studios don't have problems with this because the lights are far away from the camera, and the shooting distance is fairly short.
That's really the only solution for the problem. You can use software to remove redeye after the fact, but pretty much all compact cameras will have this problem.
You might be surprised what pros are using, actually. I've interacted with three professionals this year: at a bat mitzvah, a wedding, and an event covered by a press photographer. Two of them used digital SLR's -- the bat mitzvah photographer (who appeared to be the highest price one) was using a Nikon D2H, which is a 4 (!) megapixel camera, and the press guy was using a 4 or 6 megapixel camera. The guy at the wedding was using a high-end Sony Cybershot -- this is a 7 or 8 megapixel camera, not an SLR, but still a high-end camera. And another friend of mine, who just turned professional, is shooting a Nikon D100, I believe. This is also a 6 megapixel camera. He does a lot of landscapes and nature photography.
:-) But at $8000 it's more than a bit out of my price range. To be honest, the full frame is probably more important, since it preserves the wide angle lenses. It also needs lots of pixels, since otherwise it would lose detail with long lenses due to the pixels being larger. Someone commented to me that the 1Ds II is Canon's entry into the medium format market, and their point is well taken -- the Leaf medium format back, which is only (!) 22 megapixel, is about $20K.
The pro at the bat mitzvah was clearly an experienced event photographer; his lenses, flash gear, and assistant were clearly set up for weddings and the like. The fact that he felt that 4 megapixels was sufficient is a sure indication that consumers buying compact cameras with 8 megapixels are paying for something they simply don't need. It may actually be harmful, since the pixels are so small that they're less sensitive, requiring more light (i. e. longer shutter speeds) to reduce noise. The D2H is truly a professional camera (in terms of body construction and operating functionality), and it's a current model. The Rebel isn't, but that's not due to image quality -- the image quality is right up there with just about anything -- but rather due to operating features and construction (it's less rugged than, say, the 1N/1D; it only delivers 2.5 frames/second for 4 frames; the controls are crippled in various minor ways that don't matter too much to amateurs but are a big deal to working professionals). But I suspect a lot of pros are using the 10D and now the 20D rather than spending a lot more money on the 1D.
Would I like to have a 1Ds II? With 16 megapixels and a full frame sensor, sure
The truth is that the detail I can get from a 6 megapixel frame is quite astounding. I have to be very careful with my technique to even fully exploit that resolution -- use a sharp lens (or at least an optimal aperture with a less than stellar lens), along with a tripod, image stabilization, or very short exposure. It's possible to get more detail from very sharp film (such as Fuji Velvia or Kodak Ektar 25, and surely Fuji Provia), but even Ektar 25, which is probably the sharpest and finest grained film ever made, is noisier than digital at ISO 400 or even 800.
On my last trip, I brought along my EOS 1N and a few rolls of Ektar 25 for a very specific purpose -- to use my 12-24 mm Sigma for real hyperwide shots; there's still a big difference between 12 mm and 19 mm (the 35 mm equivalent of 12 mm on the Rebel), even though 19 mm is still considered superwide. Everything else was digital.
BTW, on the subject of the LCD display, I just don't like them for composing; I prefer to compose through an optical viewfinder. Partly it's sharpness/resolution for critical focusing, partly it's what I'm used to, and partly it's because LCD panels respond slowly. I think it's also less distracting, because I have my eye right up against it. SLR's do of course have mirror delay, but even the Rebel's quite good in that department, perhaps not quite as good as the 1N, but on a par with the A2.
Good suggestions by Skyshadow; nothing I disagree with.
I have a Digital Rebel myself, with a collection of glass ranging from a Sigma 12-24 to a Canon 75-300 IS and several prime (fixed focal length) lenses. It's a great camera; the autofocus is better than the EOS 1N (the previous generation pro film body), and it's almost as responsive in a lot of ways. The 20D is a great camera, from what little I've used (fired a few frames at a show).
I bought a Canon A75 for my wife, at the much-maligned Best Buy (it was on sale, and the price was reasonable). I specifically wanted her to be able to share my memory cards, and Canon's about the only P&S maker that uses compact flash. It's also a very nice camera for what it is. It can't do what the Rebel can in sheer flexibility, and shooting RAW with the Rebel lets me process the images to get more highlight detail, but it's surprisingly flexible if you want to get more creative. It's a bit bigger than the S series, and not as flexible as the G series, but it's a good camera for a good price.
I completely agree about the megapixels and zoom range. One thing to beware of is "digital zoom", which is a polite word for "cropping". A 3x digital zoom effectively converts a 3 MP camera to a 1/3 MP camera (3x linear zoom, 1/9 the area). I suppose if you want to just stick it into a printer with a card reader that's OK, but understand what's going on. I also agree about image stabilization with a 10x zoom! This is typically equivalent to about 400 mm on a 35 mm camera, which means that to have a decent shot of getting a good hand held shot you need an exposure of 1/400 second or faster. My sister in law bought one of those cameras. Shooting indoors, even with flash (in a brightly lit room), I was unable to get a sharp image, despite being careful in my technique. Image stabilization helps a lot -- with that focal length, if you're reasonably careful, you can get a sharp shot at 1/50 ~ 1/100.
One other point -- look at the 35 mm equivalent for the zoom range. Most of these cameras have more extension on the long end than on the wide end; the wide end is typically equivalent to 35~38 mm, which is only barely "wide angle". Some compact cameras have a wide end of 24-28 mm, which is markedly wider angle, and usually enough for good landscapes. The downside is that a 3x zoom with a wide end of 28 (equivalent) only goes to 90 mm or thereabouts, which is reasonable for portraits but isn't exactly much good for real telephoto work. 35 mm lenses go out to 105 or so, which isn't markedly longer, so I would suggest getting a compact with a wide end equivalent to 28 mm or so. The actual focal length isn't very useful for comparison, since sensors aren't all of the same size and the angle of coverage depends upon the lens focal length and the sensor size, so "35 mm equivalent" is the best comparison.
Employers should (for the sake of the business) try to get the most they can out of their workforce. However, at least in a typical high tech development environment, that doesn't mean the most hours of work. Practices such as those described suggest that either management assumes that there will always be some incremental gain in output for another hour worked, or have other reasons (such as a power trip, or some wild notion of "team building").
My own take, when I was a manager at a large company you've heard of, was that I wanted people to work smart rather than merely working hard. Granted, there are rare times when it's necessary to put in more time (late nights, weekends, off hours) to complete a key short-term deliverable, but people working long hours constantly isn't a sign of good management, but rather poor management. Employees who get tired will start making mistakes, and that's expensive (remember that the later a bug is found the more expensive -- by a large margin -- it is to fix.
The other key point here is that hiring (including the salaries of the hiring manager, HR, interviewers, and training) is expensive. In my experience, it takes a while in my line of business (system development) for even a very good new hire to really pay their way. It has also been said that the difference in productivity between a top programmer and a marginal programmer is 100 to 1. If you work from those assumptions, the way to extract the maximum useful output is to hire good people, encourage them to work efficiently, and otherwise treat them well.
I like to say that if someone who reported to me accomplished everything they were expected in a high quality manner to in 10 hours a week I'd have no problem with it. My own experience is that some people like to work in quick bursts, some people really do like to put in a lot of hours, and some people simply work steadily. However they prefer to do so is fine by me. I do have a bit of a problem with people who do the same thing over and over again (often spending a lot of time on it) without trying to find a better way of doing it. I like to say that I'm too lazy to do the same thing twice. Computers don't get upset if they're asked to do the same thing over and over again, and I prefer to move on to something new.
Obviously, there are people who don't see it that way. Rest assured, though, there are companies and managers who do take a reasonable approach to this, and that the whole industry isn't a sweatshop.
A click to raise option is fine -- as long as it's an option. Personally, I use a keyboard accelerator to raise windows, but with neither autoraise nor click to raise enabled (in KDE), and with mouse strictly follows focus. Yes, that does mean that I sometimes operate in partially obscured windows, but that's by choice.
I recall reading way back when that sulphur goes through an interesting variety of state changes as it's heated -- it first becomes a free-flowing liquid, then the viscosity increases in discrete steps as it's further heated, and then it becomes a free flowing liquid again and finally boils. It never really solidifies, but the viscosity markedly increases as it's heated to the point that it becomes taffy-like or harder.
Or my memory could be AFU.