h a l f b a k e r yFree set of rusty screwdrivers if you order now.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Current display cards in 16-bits-per-pixel mode use five bits each for red, green, and blue (or else 6 for green and five for red and blue). With some fairly simple hardware the display could instead interpret 16-bit-data as YUV format (Intensity, reddishness, and bluishness). This would allow for
better color rendition than current 16bpp modes.
Note, btw, that JPEG converts pictures to YUV format as a first-step to compressing them; the color-space of a YUV display would thus more closely match that of a JPEG than that of a 16bpp RGB display.
color
http://www.handprin.../HP/WCL/color1.html [egnor, Oct 05 2004, last modified Oct 17 2004]
[link]
|
|
Great idea, but I don't think the
correct ratio is 6:5:5. Digital video signals often use YUV 8:4:4. |
|
|
mmm... but there are really no excuses (nowadays) for using 16bit mode anywhere instead of full-colour 24bit (32bit mode just for the sake of the convenience and speed). or am i missing something here? |
|
|
In theory, 24-bit color should be something like 10:7:7 YUV. Also, intensity should be logarithmic. But with current display technology and the limitations of our eyes, nobody notices anyway; a bigger problem is just standardizing what we've got (gamma, &c). |
|
|
The 8:4:4 ratio commonly given for YUV refers to spacial resolution rather than bit depth (bit depth is typically 8 or 10 for video; don't know about other purposes). And you're quite correct that standardizing on a logarithmic display curve would almost certainly be a good thing. Too bad "the wonderful thing about standards is that there are so many of them." |
|
|
If you're talking about spatial resolution, why would you say "8:4:4" and not "2:1:1"? |
|
|
Can I get YUV on my HUD for my SUV? |
|
|
<i>If you're talking about spatial resolution, why would you say "8:4:4" and not "2:1:1"?</i><p> |
|
|
Fair question to which I don't know the answer. The "8:4:4" digital video standard I looked at used a pixel rate of 3.5 times Fsc, with one Y byte for every pixel and a U and V byte for every other. So I don't know what the numbers referred to. |
|
|
if YUV or YCrCb is intensity, reddishness, and blueishness,
don't we need greenishness or yellowishness to get all
the colors? |
|
|
No, [roborarmy], you can calculate greenishness given intensity and redishness and blueishness. The wonderful thing about color (as humans perceive it) is that it can be represented as any 3 parameters so long as:
1) Each parameter can be expressed as an invertible function of some subset of {R, G, B} only.
2) No parameter can be expressed as a function of the other two only. AND
3) Each of {R, G, B} must influence at least one of the parameters, although not necessarily the same parameter. |
|
|
Hence, we have a multitude of available color spaces, all of which are equally capable of describing the full range of colors (given infinite numerical range and precision for each parameter). Sometimes one color space is more convenient than another. For example, RGB is most closely related to the way color monitors actually generate images and the way human eyes see them. YUV is more efficient for digital transmission where both numerical precision and range are limited to just a few bits per pixel. CMY(K) is most closely related to ink and dye transfer involved in printing. HSV (sometimes called HSI) offers compactness and direct backward compatibility with black-and-white formats. |
|
|
Thank you BigBrother. A couple of slight nits, though otherwise you're 99.44% on the money. |
|
|
First of all, HSV more closely matches people's perception than RGB, but is computationally cumbersome to convert to RGB (which is what a monitor, electrically, has to display). YUV is not as close to people's perception as HSV, but is closer than RGB. |
|
|
Secondly, color may be represented by any set of 3 parameters provided there exist functions to convert any RGB value into a parameter value which may be uniquely converted back to RGB. Note that the RGB need not translate into a unique parameter value if all of the parameter values into which it could translate map back to the original RGB value. |
|
|
For example, an RGB value of (0,0,0) corresponds to HSV values of (180deg,0,0), (120deg,5,0), or (anything, anything, 0). An RGB value of (,5,.5,.5) corresponds to HSV values of (180deg,0,0.5), (20deg, 0, 5), or (anything, 0, 0.5). |
|
|
Otherwise, though, you got stuff pretty much right. |
|
|
Thank you, [supercat]. I had forgotten about the non-uniqueness of certain mappings between color spaces. |
|
|
As for human perception being closer to RGB or HSV, I know that reality lies somewhere in the fuzzy area between between the two modes, and I will leave it up to the physicists, neurologists, and philosophers to debate precisely where. Cheers. |
|
|
you cant really know exactly which is closest. all eyes are slightly different, even the same eye at different times, depending on how many of the color representing chemicals are present in the retina's neurons, and on how many cones and rods are in the retina and currently active. |
|
|
They don't vary all that much. The question is undefined for another reason: "perception" depends on what point in the neurological stack you're talking about. Actual cones are (sorta) RGB, but it gets quickly turned into something much like YUV. See link (and other sources). |
|
|
// The 8:4:4 ratio commonly given for YUV refers to spacial
resolution rather than bit depth //
// If you're talking about spatial resolution, why would you
say "8:4:4" and not "2:1:1"? // |
|
|
Can somebody confirm or deny this, and (if true) explain
what it means? It makes no sense to me for those numbers
to refer to spatial resolution (pixels of the screen?) instead
of color resolution. |
|
|
Also, I had always thought 24-bit "true color" was 8:8:8 RGB
(because image editing software, when using RGB color, usually
gives R, G, and B sliders that each go from 0 to 255, and because that
seems a convenient format for a monitor to receive and display). Is it
actually
8:8:8 or 10:7:7 YUV? |
|
| |