I'll back this up, in case anyone doesn't believe him. After I bought a colorimeter and calibrated my display, gradients have almost no stepping (even though the calibration process actually removes colors, because it maps to a subset of the available colors). And I don't even have particularly nice monitors, like the 2405FPW.
I find it amusing how most people don't even realize how poorly calibrated their monitors are. If they don't come out poorly calibrated from the factory or the store, someone fiddles with the picture settings and skews everything way off.
Try this little experiment: Post a picture of something online, then ask a few different people to describe it. It's amusing how many widely different descriptions you get of the same colors.
Why do so many people not care about having sharp eyesight? I was one of those people, so I'll try to answer this for you.
Frankly, most daily tasks don't require good eyesight. I don't even bother wearing my glasses unless I'm reading signs or driving or something. And my level of eyesight actually requires correction; a lot of people have less-than-perfect eyesight that's still legal to drive with.
When I go to the movie theater or watch a DVD on a big screen or something (if I'm watching on my laptop, I can already see every pixel at a comfortable viewing distance), I do put on my glasses so I can enjoy the sharpness (if it's that sort of movie; some movies are better without being pixel-perfect sharp).
However, for everyday life, it provides marginal benefit. And corrective lenses inevitably introduce other kinds of distortion, which I find give me a headache. Certainly if I want to make sure something is straight and level, I take off my glasses, because I can't trust my lenses to match what my brain has been wired over the years to perceive as straight.
The human eye isn't linear in color response, but neither are monitors. While the actual technologies have various response curves, they're calibrated to take, say, an 8-bit input, and map it to a power curve (gamma) which closely models human eye response.
This means, essentially, that every step between colors on the RGB scale looks like the same difference in intensity, even though it's not. That's why color gradients don't look like crap; RGB in the computer world is perceptually linear, rather than physically linear.
This raises an interesting wrinkle in 3D rendering, because lighting calculations need to be performed in physical, linear RGB (for obvious reasons; lighting is a physical process), but the output needs to be in gamma-corrected perceptual RGB for output to the monitor. Get this wrong and everything will look too dark, and gradients won't look right. (This also applies to texture maps, which are generally stored in perceptual RGB.)
The same thing also applies in the case of antialiasing, which is based on the linear intensity produced by subpixel rendering...
Especially considering that most people buying these will be big tech geeks. Which are mostly men. I actually rather have my doubts about that. The stereotypical big tech geek is a gamer who probably cares more about response time (which these color accurate displays generally don't have), home theater enthusiast who wants a big screen (but probably doesn't care particularly about color accuracy, within reason), or is reasonable enough not to spend $3000 on a monitor that only provides marginal benefit to them.
The real customer for a display like this, people who can actually spend that $3000 and call it an investment, are creative types. You know, graphic artists, filmmakers, photographers, etc., both professional and highly motivated amateur. Probably a large fraction of those are female. And in any case, probably have good color perception if they're in the market for something like this to begin with.
There are some interesting comments about whether or not the human eye can actually distinguish all these colors, but I think they miss the point about the true purpose of the extra bits.
It's so you can throw them away.
Achieving color accuracy requires a lot more than just having a lot of precision. If any given display can output 2^30 different shades, that still doesn't get you accuracy, because you want any given 3x8-bit color to map to a precise one of those 2^30 shades.
The extra bits give you room to make minor adjustments to get exactly the color you want. You'll notice how they mention a laundry list of color spaces that they support, each with a slightly different mapping from 24-bit color to what this monitor outputs.
Dynamic range is a red herring; these displays aren't designed to produce high dynamic range (check out the BrightSide monitor if you want to see where that tech is going). They're designed to be perfect, idealized versions of what you've got in your living room. It doesn't do you much good to proof on a supermonitor which doesn't resemble the final output device. (And yes, output to film stock does provide plenty of opportunity for dynamic range, but that's still not the point of these monitors.)
Incidentally, for those who don't understand the bit about the "wide color gamut" enabled by LEDs, color spaces (such as the Adobe RGB, sRGB, NTSC, and so on spaces mentioned in the summary/article) are defined by three primary colors. Nothing new there.
The tricky bit is that the specifications define these three primary colors in terms of a precise frequency of light. The only light source that comes close are tuned lasers. Consequently, that LCD monitor sitting on your desk (or lap), probably backlit by a fluorescent light, can only reach something like 80-90% of the specified color gamut.
LEDs are pretty close to lasers in terms of color purity, and monitors backlit by LEDs can often reach an astonishing 98% or more of the color gamut. This wide gamut often allows them to cover more than one color space adequately, as exemplified by the monitor mentioned in this article.
Better than CRT, actually. At least under certain conditions.
Matrix-style displays have some big inherent advantages over scanning phosphor technology, such as crisp, precise, flicker-free display.
Meanwhile, there have been "deep color" displays like this capable of more than 24-bit color for a while. Use of LED backlights give them a much wider color gamut than phosphors are capable of.
The main failings of current LCD technology fall into two categories:
First, LCDs block light imperfectly, so you get potentially poorer black levels. (CRTs aren't as good at this as their boosters would like you to believe, though.)
Secondly, you have the color shift problem, where the angle of viewing distorts the color accuracy. The degree depends on the technology, but it can't ever be completely eliminated.
Under proper viewing conditions, LCDs can do a good job on both fronts; a major movie studio is certainly an example of an absolutely color-critical user. However, it comes at a big cost.
The future is probably OLED, or maybe e-ink. Unlike LCD, OLED is a light emissive technology, so it has absolute blacks and no color shifts. However, who knows how long it'll take OLED technology to reach commercialization, due to the problem with blue OLED lifetimes; the closest thing right now is a tiny 11" Sony TV that costs a small fortune, and minuscule cell phone screens.
Ha-ha. My DSL connection actually has 50% more upstream bandwidth than your cable connection. (Admittedly, I mainly use it for gaming rather than streaming video, so this and consistent low latency may be more important to me than it is to you.)
Incidentally, if you think Netflix streaming is near-DVD quality, or that it takes anywhere near 3-6 Mbps to deliver, let alone 9+, I have a bridge in Manhattan I'd like to sell you...
Actual Netflix bitrates range from 0.5 to 2.2 Mbps, depending on the negotiated stream quality. You don't really think they're going to alienate the majority of their (US-only) marketplace, which certainly isn't paying for a 6+ Mbps connection, do you? There's some talk of providing HD streams at 3+ Mbps, but right now it's just that, talk. (Even then, it's going to be pretty sucky HD; Blu-ray provides up to 36 Mbps of H.264, and ATSC 19.4 Mbps of MPEG-2.)
If you have a Netflix subscription, go ahead and watch it sometime, and check your bandwidth meter. (You can do this using iftop under Linux, or the Task Manager in Windows.) It's not going to get anywhere near pegging your available downstream bandwidth.
Slashdot Mirror
I'll back this up, in case anyone doesn't believe him. After I bought a colorimeter and calibrated my display, gradients have almost no stepping (even though the calibration process actually removes colors, because it maps to a subset of the available colors). And I don't even have particularly nice monitors, like the 2405FPW.
I find it amusing how most people don't even realize how poorly calibrated their monitors are. If they don't come out poorly calibrated from the factory or the store, someone fiddles with the picture settings and skews everything way off.
Try this little experiment: Post a picture of something online, then ask a few different people to describe it. It's amusing how many widely different descriptions you get of the same colors.
Frankly, most daily tasks don't require good eyesight. I don't even bother wearing my glasses unless I'm reading signs or driving or something. And my level of eyesight actually requires correction; a lot of people have less-than-perfect eyesight that's still legal to drive with.
When I go to the movie theater or watch a DVD on a big screen or something (if I'm watching on my laptop, I can already see every pixel at a comfortable viewing distance), I do put on my glasses so I can enjoy the sharpness (if it's that sort of movie; some movies are better without being pixel-perfect sharp).
However, for everyday life, it provides marginal benefit. And corrective lenses inevitably introduce other kinds of distortion, which I find give me a headache. Certainly if I want to make sure something is straight and level, I take off my glasses, because I can't trust my lenses to match what my brain has been wired over the years to perceive as straight.
The human eye isn't linear in color response, but neither are monitors. While the actual technologies have various response curves, they're calibrated to take, say, an 8-bit input, and map it to a power curve (gamma) which closely models human eye response.
This means, essentially, that every step between colors on the RGB scale looks like the same difference in intensity, even though it's not. That's why color gradients don't look like crap; RGB in the computer world is perceptually linear, rather than physically linear.
This raises an interesting wrinkle in 3D rendering, because lighting calculations need to be performed in physical, linear RGB (for obvious reasons; lighting is a physical process), but the output needs to be in gamma-corrected perceptual RGB for output to the monitor. Get this wrong and everything will look too dark, and gradients won't look right. (This also applies to texture maps, which are generally stored in perceptual RGB.)
The same thing also applies in the case of antialiasing, which is based on the linear intensity produced by subpixel rendering...
The real customer for a display like this, people who can actually spend that $3000 and call it an investment, are creative types. You know, graphic artists, filmmakers, photographers, etc., both professional and highly motivated amateur. Probably a large fraction of those are female. And in any case, probably have good color perception if they're in the market for something like this to begin with.
There are some interesting comments about whether or not the human eye can actually distinguish all these colors, but I think they miss the point about the true purpose of the extra bits.
It's so you can throw them away.
Achieving color accuracy requires a lot more than just having a lot of precision. If any given display can output 2^30 different shades, that still doesn't get you accuracy, because you want any given 3x8-bit color to map to a precise one of those 2^30 shades.
The extra bits give you room to make minor adjustments to get exactly the color you want. You'll notice how they mention a laundry list of color spaces that they support, each with a slightly different mapping from 24-bit color to what this monitor outputs.
Dynamic range is a red herring; these displays aren't designed to produce high dynamic range (check out the BrightSide monitor if you want to see where that tech is going). They're designed to be perfect, idealized versions of what you've got in your living room. It doesn't do you much good to proof on a supermonitor which doesn't resemble the final output device. (And yes, output to film stock does provide plenty of opportunity for dynamic range, but that's still not the point of these monitors.)
Incidentally, for those who don't understand the bit about the "wide color gamut" enabled by LEDs, color spaces (such as the Adobe RGB, sRGB, NTSC, and so on spaces mentioned in the summary/article) are defined by three primary colors. Nothing new there.
The tricky bit is that the specifications define these three primary colors in terms of a precise frequency of light. The only light source that comes close are tuned lasers. Consequently, that LCD monitor sitting on your desk (or lap), probably backlit by a fluorescent light, can only reach something like 80-90% of the specified color gamut.
LEDs are pretty close to lasers in terms of color purity, and monitors backlit by LEDs can often reach an astonishing 98% or more of the color gamut. This wide gamut often allows them to cover more than one color space adequately, as exemplified by the monitor mentioned in this article.
Better than CRT, actually. At least under certain conditions.
Matrix-style displays have some big inherent advantages over scanning phosphor technology, such as crisp, precise, flicker-free display.
Meanwhile, there have been "deep color" displays like this capable of more than 24-bit color for a while. Use of LED backlights give them a much wider color gamut than phosphors are capable of.
The main failings of current LCD technology fall into two categories:
First, LCDs block light imperfectly, so you get potentially poorer black levels. (CRTs aren't as good at this as their boosters would like you to believe, though.)
Secondly, you have the color shift problem, where the angle of viewing distorts the color accuracy. The degree depends on the technology, but it can't ever be completely eliminated.
Under proper viewing conditions, LCDs can do a good job on both fronts; a major movie studio is certainly an example of an absolutely color-critical user. However, it comes at a big cost.
The future is probably OLED, or maybe e-ink. Unlike LCD, OLED is a light emissive technology, so it has absolute blacks and no color shifts. However, who knows how long it'll take OLED technology to reach commercialization, due to the problem with blue OLED lifetimes; the closest thing right now is a tiny 11" Sony TV that costs a small fortune, and minuscule cell phone screens.
Ha-ha. My DSL connection actually has 50% more upstream bandwidth than your cable connection. (Admittedly, I mainly use it for gaming rather than streaming video, so this and consistent low latency may be more important to me than it is to you.) Incidentally, if you think Netflix streaming is near-DVD quality, or that it takes anywhere near 3-6 Mbps to deliver, let alone 9+, I have a bridge in Manhattan I'd like to sell you... Actual Netflix bitrates range from 0.5 to 2.2 Mbps, depending on the negotiated stream quality. You don't really think they're going to alienate the majority of their (US-only) marketplace, which certainly isn't paying for a 6+ Mbps connection, do you? There's some talk of providing HD streams at 3+ Mbps, but right now it's just that, talk. (Even then, it's going to be pretty sucky HD; Blu-ray provides up to 36 Mbps of H.264, and ATSC 19.4 Mbps of MPEG-2.) If you have a Netflix subscription, go ahead and watch it sometime, and check your bandwidth meter. (You can do this using iftop under Linux, or the Task Manager in Windows.) It's not going to get anywhere near pegging your available downstream bandwidth.