The real numbers (other than countable subsets thereof) cannot be real, as that would violate the Bekenstein bound and let you do crazy things like build superturing machines. Not in this universe...
Well, Wine is always an option. Photoshop runs in Wine without a problem.
Additionally, one of the most important graphics production applications, Autodesk's Maya, runs natively on Linux and has absolutely no difference from the Windows version.
And if still didn't really sink in, let me give an extreme artificial example for the purposes of illustrating the mathematical point: consider a light with a completely flat spectral response, and a light where its spectrum has a square wave shape, with each 10 nm long portion of the spectral response alternating between zero intensity and double the intensity of the first light. The two lights obviously have the same intensity and same color temperature and white balance.
Consider a material which has a spectral reflectance which is the same as the second light, but shifted by 10 nm, so that each 10 nm segment of wavelengths where the second light is at zero intensity, the material has unity reflectance, and where the second light has its full intensity, the material has zero reflectance.
The material would be white under the first light and black under the second light. As extreme this example is, the less extreme cases are still common enough that metameric failure and related color rendering phenomena remain active fields of study.
The human eye corrects color for overall color temperature/white balance with the presumption of a smooth, approximately blackbody spectrum, which is then shifted more towards the reds or blues (in the case of daylight, by the filtering of the atmosphere, and in the case of incandescent bulbs, by the change of filament temperature and any actual filters). This shift is what the human vision correction compensates for.
If the spectrum is not smooth, approximately blackbody, but uneven with peaks and troughs, then the color rendering errors when you use the light to light up various materials are of a complex nature that cannot be corrected by a simple weighting towards the blue or red wavelengths. Think of it this way: color temperature is a compressed summary measure that ignores the details of each spectrum. A number of problems occur, such as metameric failure (see http://en.wikipedia.org/wiki/Metamerism_(color)#Metameric_failure and http://www.kurabo.co.jp/el/world/en/img/room/color/page1/6_image01.jpg which assumes the light and sun are matched to the same color temperature but have different spectra) and the reverse, as well as some material colors completely changing in appearance.
Mathematically, consider a light spectrum discretized into a vector where each component of the vector represents the relative intensity of a small range of wavelenghts. Same for the material spectral reflectance, and same with the sensitivity response curves of each of the three types of cone cells in your eye. The resulting color appearance of a material lit by the light is then done by taking the element-by-element product of the first two vectors, and then creating a three element vector where each component is the dot-product of the intermediate vector mentioned before and the response vector for each of the three cone cells. Now here's the part I think you had not realized: when you take two lights with the same color temperature but different spectra (i.e. color temperature being a weighted average of the spectral power distribution over these vectors), then in general when you carry the mathematical operation above using the same material in each case, you will get different 3-vectors (perception) as an end result. This is the explanation behind metamerism and related color rendering phenomena. Looking directly at the lights, they look the same, but when you introduce the middle step of lighting actual materials (multiplying with the material reflectances), then suddenly those appear different for each of the lights, regardless of matching light color temperature/white balance.
Well, halogens are incandescents. They are able to run somewhat hotter than non-halogen incandescents because the halogen gas redeposits some of the tungsten that now sublimates faster from the filament under the increased temperature. Higher temperature means the blackbody spectrum is shifted a bit higher into the blues and a bit less is wasted in infrared.
> I find the 5000k to to feel "blue", not white
It is. Daylight is normally even higher than that (in places over 6000 K) because the sky scatters a lot of the blue light back down.
The important thing is actually that human perception of light being warm or cold depends on the overall light intensity. When intensity is low, the same light appears cooler colored than when its intensity is high. If you take 6000 K daylight and reduce in intensity by a couple orders of magnitude to get an indoor 6000 K light it would look very blueish. On the other hand, boosting the intensity of a 3500 K bulb to noon daylight levels would result in a very orange-looking light. This is an artifact of human perception, not physics.
The 4700 K Solux ones look great indoors at nigh or when it's heavily overcast outside. But when I turn it on while there's bright sunlight outside, the Solux looks yellowish. Yet I did not go for the 5000 K Solux as I know that at those times I actually will be using the light, it will look too blue.
The lifespan and efficiency of Solux is pretty good for incandescents but the biggest problem is that they are MR16 format and the needed 12 V fixtures are becoming less common, as most new track lighting uses 120 V GU10 bulbs. Solux now has 120 V bulbs that fit standard old-school sockets but they're too expensive for my taste. The other, smaller downside, with Solux is that you need to have a black inner surface of the shade behind the bulb to absorb the yellow-red light that goes through the dichroic filter reflector, else you end up defeating the very way they achieve the nice spectrum. I've found the hard way that you have to use a shade that's already black, as if you just paint one that way, it gets hotter than it's usually designed for...
Well, I guess some people are more anal than others about quality. Or rather, they concentrate on different things: for some, the wine has to be perfect, for others, their pixels. For me, it's many things, but light is one, especially as I live in a place (Vancouver) that's overcast and rainy half of the year. During the dark times, fluorescent lights of all types don't make me feel better, and HIDs the same, but daylight bulbs do. I also have paintings and film photographs (gasp! lol) and they look far better under the Solux bulbs. People's skin also looks more natural. I don't have flowers anymore but when I did, they looked completely different under CFL light than daylight due to some of the spikes in the violet end of the CFL spectrum. Same for my fish. LED lights are getting better but still not quite there, and I think the only way they can be made to match daylight in the future would be by using a combination of more phosphors with emissions in more discrete parts of the spectrum combined with filtering. I don't know how practical that is.
Efficiency is a non-issue here in Canada except in the 4 warmest months of the year. Any heat produced by the light is offset by the same extra electricity going to the heaters instead.
When efficiency starts impacting quality significantly (color rendering ability, in this case), then the focus ought to be shifted to power generation.
There are a number of severe errors in your post. There's a difference between the apparent color of the light when you look at it (what color temperature is about) and the color rendering of objects it lights up, which depend on the specifics of its spectrum. When you look into the light, ones with different spectra can look the same if they're matched to the same color temperature. As I explained in my linked post, this tells you nothing about how objects will look. A CFL matched to noon-daytime temperature still has a shitty color rendering index because the spectrum is not smooth and most surfaces look significantly different than under actual noon-sunlight!! For incandescents, however, the reddish end of the spectrum is easily filtered by the bulb reflectors on high end MR16 bulbs. Thus, I can buy halogen incandescent easily at a choice of color temperatures: 3500 K, 4700 K, 5000 K (Solux and their competitors) and professional use ones at 6000+ K (museums, movies). Unlike a CFL, all of these have smooth spectra that are filtered to match the sunlight at corresponding times of day, and an incandescent at a given color temperature is infinitely better color renderer than a CFL at the same color temperature. But, you say, noon is around 6500 K! Not so fast, old man: studies show that human color perception is influenced by overall brightness levels, so that in a dim environment a 5000 K light appears cool, whereas in a very bright environment it appears too warm. Sunlight at noon is a couple orders of magnitude brighter than typical home lighting, so that even a 5000 K light is too cool for most indoor use, unless you have a grow op.
The color temperature is a minor factor compared to the CRI (Color Rendering Index), and incandescents are still kings there due to the smooth spectrum. This makes them easy to filter, and high-end MR16-sized halogens do that using wavelength-selective reflectors and match daylight almost exactly. That's impossible to do with fluorescents/HIDs/LEDs and their spiky spectra fucking up colors. http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
You're right, who wants to use lamps that have the highest Color Rendering Index in their optimal form (MR16 halogens with filtering reflectors? I'd love to see your reasoned reply to my detailed argument here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
LEDs still don't have as good Color Rendering Index as high end incandescents, because the spectrum is not as smooth as a blackbody radiator--the incandescent, which can be easily filtered to raise the color temperature and match daylight.
They also suck anywhere you need a high Color Rendering Index due to their spiky spectra fucking up color appearance of many surfaces. The good MR16 halogens have reflectors that leak some of the longer wavelengths through, which can only be done with the smooth spectra of blackbody emitters--incandescents--and they match daylight almost exactly.
Color temperature is not what's important, it's the color rendering index. LEDs have spiky spectra that fuck up color reproduction, though not nearly as bad as fluorescents and HIDs. Incandescents have CRI around 97, due to their smooth blackbody spectrum which is basically like sunlight but tinted to red/yellow. But the lower color temperature is easy to correct--that's exactly what the best daylight-matched MR16 track lighting bulbs do, by having their reflectors leak through just the right amount of longer wavelengths instead of reflecting them. Filtering the narrow spikes of non-blackbody bulbs is hopeless (not to mention removes big portions of the light energy and thus killing their primary benefit-efficiency). More details in my post here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
It's not that they radiate more into the red. It's that incandescents have a much higher color rendering index (~97), because they emit a smooth blackbody spectrum. Objects lit by them have the appearance of being lit by daylight that has been tinted towards the low end of the spectrum. On the other hand, fluorescents, HIDs, and LEDs have spiky spectrums which fuck up the color reproduction. Incandescents' smooth spectrum can also easily be filtered to exactly match daylight (and some bulbs do), whereas narrow spikes cannot be effectively filtered. More info in my post here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
Color temperature tells you nothing about color reproduction. CRI is a far more important metric. You can match a 100 different lights to the same color temperature, and objects will still look completely different under each of those light. See my post at http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269 for more detail.
Unfortunately, only incandescent bulbs can be made to match the spectrum of sunlight (with appropriate filtering). This is very important for a number of reasons, including cases where color reproduction is critical (museums, etc.), and where there is not much natural sunlight (Seattle and Vancouver half of the year, or an office with insufficient or poorly directed windows). There have been a number of studies correlating productivity with daylight, and also studies linking low amount of daylight with depression.
Color reproduction suffers tremendously with the horribly spiky spectrum of fluorescent and high intensity discharge bulbs, and while white LEDs are better, their spectrum still has significant humps that make them unsuitable if you want to emulate daylight properly. This is not a simple matter of white balance and using so-called "full-spectrum" bulbs. The color of objects one observes is the product of three functions: the light source spectrum, the surface reflectance spectrum, and the spectral sensitivities of the human eye. Although the eye reduces color to only three dimensions due to having cones with sensitivities centered at the usual RGB wavelengths, it does not mean that a white-balanced light source with three narrow spikes centered at the same wavelengths is anywhere near sufficient. The reason for this is that the three types of retinal cone cells each have fairly broad sensitivity ranges. This means that, while staring at such a light source would be the same as staring at a light source with a smooth spectrum, things change when you introduce the reflectance of the surface of objects. Then, the lights that when observed directly looked the same will produce very different renditions of the colored object--because its reflectance has a spectral distribution that doesn't correlate with that of the R, G, and B peaks of retinal color sensitivity. Your "full-spectrum" fluorescent bulb will have its spectral spikes in general not match the spectral spikes of the reflectance of different surfaces that you're observing with this light. The result is colors that look completely different for bulbs matched to the same white balance point and color temperature.
Daylight from the sun has a smooth spectrum, because it is a blackbody emitter. Incandescent bulbs are also blackbody emitters. Unfortunately, due to the lack of a material that can withstand sufficient temperature, we have to run them at lower than optimal temperature and more of their energy tends towards longer wavelengths: heat (giving inefficiency) and red and yellow colors (giving a tint and low color temperature). Halogen incandescents are a little bit better as they run hotter, but its the reflectors on their MR16 incarnations (common in track lighting) that make an important difference, as they're designed to preferentially leak some of the reds and yellows and improve the color temperature.
High end MR16 incandescents such as Solux (which I use in my desk lamps) match daylight almost perfectly. This is made possible by the smooth spectrum of a blackbody emitter--the heated filament. Trying to filter narrow, high power spikes in fluorescent/HID/LED light spectra requires precise narrow-band filters, which is extremely impractical.
Slashdot Mirror
The real numbers (other than countable subsets thereof) cannot be real, as that would violate the Bekenstein bound and let you do crazy things like build superturing machines. Not in this universe...
Luckily, I live in Canada where the right to offend people was recently restored by parliament: http://www2.macleans.ca/2012/06/19/five-years-two-tribunals-a-raft-of-secret-hearings-a-supreme-court-challenge-how-the-battle-for-free-speech-was-won/ Good thing my country has politicians who, unlike the parent poster, value freedom of expression over political correctness.
Well, Wine is always an option. Photoshop runs in Wine without a problem.
Additionally, one of the most important graphics production applications, Autodesk's Maya, runs natively on Linux and has absolutely no difference from the Windows version.
And if still didn't really sink in, let me give an extreme artificial example for the purposes of illustrating the mathematical point: consider a light with a completely flat spectral response, and a light where its spectrum has a square wave shape, with each 10 nm long portion of the spectral response alternating between zero intensity and double the intensity of the first light. The two lights obviously have the same intensity and same color temperature and white balance.
Consider a material which has a spectral reflectance which is the same as the second light, but shifted by 10 nm, so that each 10 nm segment of wavelengths where the second light is at zero intensity, the material has unity reflectance, and where the second light has its full intensity, the material has zero reflectance.
The material would be white under the first light and black under the second light. As extreme this example is, the less extreme cases are still common enough that metameric failure and related color rendering phenomena remain active fields of study.
The human eye corrects color for overall color temperature/white balance with the presumption of a smooth, approximately blackbody spectrum, which is then shifted more towards the reds or blues (in the case of daylight, by the filtering of the atmosphere, and in the case of incandescent bulbs, by the change of filament temperature and any actual filters). This shift is what the human vision correction compensates for.
If the spectrum is not smooth, approximately blackbody, but uneven with peaks and troughs, then the color rendering errors when you use the light to light up various materials are of a complex nature that cannot be corrected by a simple weighting towards the blue or red wavelengths. Think of it this way: color temperature is a compressed summary measure that ignores the details of each spectrum. A number of problems occur, such as metameric failure (see http://en.wikipedia.org/wiki/Metamerism_(color)#Metameric_failure and http://www.kurabo.co.jp/el/world/en/img/room/color/page1/6_image01.jpg which assumes the light and sun are matched to the same color temperature but have different spectra) and the reverse, as well as some material colors completely changing in appearance.
Mathematically, consider a light spectrum discretized into a vector where each component of the vector represents the relative intensity of a small range of wavelenghts. Same for the material spectral reflectance, and same with the sensitivity response curves of each of the three types of cone cells in your eye. The resulting color appearance of a material lit by the light is then done by taking the element-by-element product of the first two vectors, and then creating a three element vector where each component is the dot-product of the intermediate vector mentioned before and the response vector for each of the three cone cells. Now here's the part I think you had not realized: when you take two lights with the same color temperature but different spectra (i.e. color temperature being a weighted average of the spectral power distribution over these vectors), then in general when you carry the mathematical operation above using the same material in each case, you will get different 3-vectors (perception) as an end result. This is the explanation behind metamerism and related color rendering phenomena. Looking directly at the lights, they look the same, but when you introduce the middle step of lighting actual materials (multiplying with the material reflectances), then suddenly those appear different for each of the lights, regardless of matching light color temperature/white balance.
Well, halogens are incandescents. They are able to run somewhat hotter than non-halogen incandescents because the halogen gas redeposits some of the tungsten that now sublimates faster from the filament under the increased temperature. Higher temperature means the blackbody spectrum is shifted a bit higher into the blues and a bit less is wasted in infrared.
> I find the 5000k to to feel "blue", not white
It is. Daylight is normally even higher than that (in places over 6000 K) because the sky scatters a lot of the blue light back down.
The important thing is actually that human perception of light being warm or cold depends on the overall light intensity. When intensity is low, the same light appears cooler colored than when its intensity is high. If you take 6000 K daylight and reduce in intensity by a couple orders of magnitude to get an indoor 6000 K light it would look very blueish. On the other hand, boosting the intensity of a 3500 K bulb to noon daylight levels would result in a very orange-looking light. This is an artifact of human perception, not physics.
The 4700 K Solux ones look great indoors at nigh or when it's heavily overcast outside. But when I turn it on while there's bright sunlight outside, the Solux looks yellowish. Yet I did not go for the 5000 K Solux as I know that at those times I actually will be using the light, it will look too blue.
The lifespan and efficiency of Solux is pretty good for incandescents but the biggest problem is that they are MR16 format and the needed 12 V fixtures are becoming less common, as most new track lighting uses 120 V GU10 bulbs. Solux now has 120 V bulbs that fit standard old-school sockets but they're too expensive for my taste. The other, smaller downside, with Solux is that you need to have a black inner surface of the shade behind the bulb to absorb the yellow-red light that goes through the dichroic filter reflector, else you end up defeating the very way they achieve the nice spectrum. I've found the hard way that you have to use a shade that's already black, as if you just paint one that way, it gets hotter than it's usually designed for...
Well, I guess some people are more anal than others about quality. Or rather, they concentrate on different things: for some, the wine has to be perfect, for others, their pixels. For me, it's many things, but light is one, especially as I live in a place (Vancouver) that's overcast and rainy half of the year. During the dark times, fluorescent lights of all types don't make me feel better, and HIDs the same, but daylight bulbs do. I also have paintings and film photographs (gasp! lol) and they look far better under the Solux bulbs. People's skin also looks more natural. I don't have flowers anymore but when I did, they looked completely different under CFL light than daylight due to some of the spikes in the violet end of the CFL spectrum. Same for my fish. LED lights are getting better but still not quite there, and I think the only way they can be made to match daylight in the future would be by using a combination of more phosphors with emissions in more discrete parts of the spectrum combined with filtering. I don't know how practical that is. Efficiency is a non-issue here in Canada except in the 4 warmest months of the year. Any heat produced by the light is offset by the same extra electricity going to the heaters instead.
When efficiency starts impacting quality significantly (color rendering ability, in this case), then the focus ought to be shifted to power generation.
There are a number of severe errors in your post. There's a difference between the apparent color of the light when you look at it (what color temperature is about) and the color rendering of objects it lights up, which depend on the specifics of its spectrum. When you look into the light, ones with different spectra can look the same if they're matched to the same color temperature. As I explained in my linked post, this tells you nothing about how objects will look. A CFL matched to noon-daytime temperature still has a shitty color rendering index because the spectrum is not smooth and most surfaces look significantly different than under actual noon-sunlight!! For incandescents, however, the reddish end of the spectrum is easily filtered by the bulb reflectors on high end MR16 bulbs. Thus, I can buy halogen incandescent easily at a choice of color temperatures: 3500 K, 4700 K, 5000 K (Solux and their competitors) and professional use ones at 6000+ K (museums, movies). Unlike a CFL, all of these have smooth spectra that are filtered to match the sunlight at corresponding times of day, and an incandescent at a given color temperature is infinitely better color renderer than a CFL at the same color temperature. But, you say, noon is around 6500 K! Not so fast, old man: studies show that human color perception is influenced by overall brightness levels, so that in a dim environment a 5000 K light appears cool, whereas in a very bright environment it appears too warm. Sunlight at noon is a couple orders of magnitude brighter than typical home lighting, so that even a 5000 K light is too cool for most indoor use, unless you have a grow op.
You can get a 50 W near-perfect daylight halogen for $7.
The color temperature is a minor factor compared to the CRI (Color Rendering Index), and incandescents are still kings there due to the smooth spectrum. This makes them easy to filter, and high-end MR16-sized halogens do that using wavelength-selective reflectors and match daylight almost exactly. That's impossible to do with fluorescents/HIDs/LEDs and their spiky spectra fucking up colors. http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
Color warmth (a.k.a. color temperature) is a minor factor compared to CRI (Color Rendering Index). I don't want to repeat a long explanation so here's my previous one: http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
Maybe people that want a decent Color Rendering Index? http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
I'm pretty sure that, though you've not realized it, it's not the warmth per se you prefer, but the smooth spectrum (the sun also has a smooth spectrum, as both are blackbody radiators). http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
What do you mean "generate light more directly"? This is an extremely unscientific comment. Incandescent bulbs' filaments are blackbody radiators. That's pretty direct, if you ask me. http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
Actually, incandescents have a huge advantage, if you have the right kind of incandescents: http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
There is one aspects of CCFLs that you can't fix, and that's their horrible spiky spectra. http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
LED spectra still are not good enough. I'll avoid a long post again and simply redirect you to http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
You're right, who wants to use lamps that have the highest Color Rendering Index in their optimal form (MR16 halogens with filtering reflectors? I'd love to see your reasoned reply to my detailed argument here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
LEDs still don't have as good Color Rendering Index as high end incandescents, because the spectrum is not as smooth as a blackbody radiator--the incandescent, which can be easily filtered to raise the color temperature and match daylight.
They also suck anywhere you need a high Color Rendering Index due to their spiky spectra fucking up color appearance of many surfaces. The good MR16 halogens have reflectors that leak some of the longer wavelengths through, which can only be done with the smooth spectra of blackbody emitters--incandescents--and they match daylight almost exactly.
Color temperature is not what's important, it's the color rendering index. LEDs have spiky spectra that fuck up color reproduction, though not nearly as bad as fluorescents and HIDs. Incandescents have CRI around 97, due to their smooth blackbody spectrum which is basically like sunlight but tinted to red/yellow. But the lower color temperature is easy to correct--that's exactly what the best daylight-matched MR16 track lighting bulbs do, by having their reflectors leak through just the right amount of longer wavelengths instead of reflecting them. Filtering the narrow spikes of non-blackbody bulbs is hopeless (not to mention removes big portions of the light energy and thus killing their primary benefit-efficiency). More details in my post here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
It's not that they radiate more into the red. It's that incandescents have a much higher color rendering index (~97), because they emit a smooth blackbody spectrum. Objects lit by them have the appearance of being lit by daylight that has been tinted towards the low end of the spectrum. On the other hand, fluorescents, HIDs, and LEDs have spiky spectrums which fuck up the color reproduction. Incandescents' smooth spectrum can also easily be filtered to exactly match daylight (and some bulbs do), whereas narrow spikes cannot be effectively filtered. More info in my post here http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269
Color temperature tells you nothing about color reproduction. CRI is a far more important metric. You can match a 100 different lights to the same color temperature, and objects will still look completely different under each of those light. See my post at http://politics.slashdot.org/comments.pl?sid=3143051&cid=41459269 for more detail.
Unfortunately, only incandescent bulbs can be made to match the spectrum of sunlight (with appropriate filtering). This is very important for a number of reasons, including cases where color reproduction is critical (museums, etc.), and where there is not much natural sunlight (Seattle and Vancouver half of the year, or an office with insufficient or poorly directed windows). There have been a number of studies correlating productivity with daylight, and also studies linking low amount of daylight with depression.
Color reproduction suffers tremendously with the horribly spiky spectrum of fluorescent and high intensity discharge bulbs, and while white LEDs are better, their spectrum still has significant humps that make them unsuitable if you want to emulate daylight properly. This is not a simple matter of white balance and using so-called "full-spectrum" bulbs. The color of objects one observes is the product of three functions: the light source spectrum, the surface reflectance spectrum, and the spectral sensitivities of the human eye. Although the eye reduces color to only three dimensions due to having cones with sensitivities centered at the usual RGB wavelengths, it does not mean that a white-balanced light source with three narrow spikes centered at the same wavelengths is anywhere near sufficient. The reason for this is that the three types of retinal cone cells each have fairly broad sensitivity ranges. This means that, while staring at such a light source would be the same as staring at a light source with a smooth spectrum, things change when you introduce the reflectance of the surface of objects. Then, the lights that when observed directly looked the same will produce very different renditions of the colored object--because its reflectance has a spectral distribution that doesn't correlate with that of the R, G, and B peaks of retinal color sensitivity. Your "full-spectrum" fluorescent bulb will have its spectral spikes in general not match the spectral spikes of the reflectance of different surfaces that you're observing with this light. The result is colors that look completely different for bulbs matched to the same white balance point and color temperature.
Daylight from the sun has a smooth spectrum, because it is a blackbody emitter. Incandescent bulbs are also blackbody emitters. Unfortunately, due to the lack of a material that can withstand sufficient temperature, we have to run them at lower than optimal temperature and more of their energy tends towards longer wavelengths: heat (giving inefficiency) and red and yellow colors (giving a tint and low color temperature). Halogen incandescents are a little bit better as they run hotter, but its the reflectors on their MR16 incarnations (common in track lighting) that make an important difference, as they're designed to preferentially leak some of the reds and yellows and improve the color temperature.
High end MR16 incandescents such as Solux (which I use in my desk lamps) match daylight almost perfectly. This is made possible by the smooth spectrum of a blackbody emitter--the heated filament. Trying to filter narrow, high power spikes in fluorescent/HID/LED light spectra requires precise narrow-band filters, which is extremely impractical.