If sRGB is so limited why not use DirectX?

[angie_nelson]

<p>Maybe this is a theoretical question but it came up when I was trying to explain what I'd been learning about color management with my "gamer" husband... We have such difficulties with the sRGB color space on the monitor (it's limited and doesn't reflect what the printer can do). But if you start an online game, it uses DirectX, OpenGL and heaven knows what else to produce many more colors than sRGB (at least it seems that way to me). Why not use that system for Photoshop, gimp, and similar software? Would it give a better gamut on the monitor so we can "take the gloves off" the printer?</p>

<p> </p>

[angie_nelson]

<p>Would directX solve this? If I could see the colors in a larger gamut on the screen, then I'd be able to tell that the skin tone is off, wouldn't I? Best of both worlds would be keeping the vibrancy of the eye color while neutralizing the skin tone.</p><div></div>

[Wouter Willemse]

<p>Angie, to be honest, you're mixing up things that are completely unrelated. DirectX and OpenGL are programming interfaces to allow software to talk to the graphics card, both mostly used for rendering 3D constructed scenes quickly (at least, a lot faster than a generic CPU can). Colour management has got nothing to do with these interfaces. Neither DirectX nor OpenGL make more colours available in any way.</p>

<p>There are monitors that can show more than the rRGB colour gamut, but they do come at an extra cost, and typically they're not optimised at all for games (their response times are slower, which is very annoying for games, but rather irrelevant for editing photos). Some of them can reach the Adobe RGB colour space, which brings you very close to what good printers can deliver. Either way, good calibration is still key - calibrated screen, good colour profiles for printer with the paper used, and the colour profile used in the image itself. Without a calibrated screen (whether it's only capable of sRGB or Adobe RGB or ...) matching the printer with the screen is always going to remain a guessing game.<br>

But the vital point to understand here is that the limiting factor typically is the monitor, and the amount of different (shades of) colour it can display. The problem is not the signal the PC sends to it. So in short, DirectX will not solve a single thing.</p>

 

[digitaldog]

<p>The better wide gamut displays can be set for an sRGB emulation mode quite easily so when you do wish to play a game, or work outside color managed app's that assume sRGB, you're all set. Press a button in the software, you're back to wide gamut behavior and one that you've calibrated for a screen to print match. As many as you wish! You'll want to look into something like an NEC SpectraView with their host software to do this. You can have the best of both worlds but it's going to cost a bit:</p>

<p>http://www.bhphotovideo.com/c/product/1006644-REG/nec_pa272w_bk_sv_multisync_27_gb_r_lcd_display_spectraview.html?N=0&gclid=CjwKEAjwkMWgBRCJ1L_wypbX0wkSJAC3Xio2TkVUwvt3_rliqAedt_Bp0pZpKlm6wIvYY0xU4ADW-hoCsdrw_wcB&InitialSearch=yes</p>

[angie_nelson]

<p>Thank you for the clarifications about how to color manage. In an area so complex, all advice is welcome.</p>

<p>I do understand that working with more bits will produce better results sometimes and that if you work with RAW files, and then go to sRGB, you're discarding data. I was just wondering if DirectX and a good video card could produce billions of colors or something like that, why not use it?</p>

<p>To explain the wide gamut idea more clearly: <a href="http://www.dpreview.com/forums/post/50881703">http://www.dpreview.com/forums/post/50881703</a> </p>

<p>Here are some gamers discussing that with the advent of Windows 7, DirectX may be able to produce more than 16 million colors. Something about a 1.07 billion colors mode? I'm no expert in this by any means, it just occurred to me as something to wonder about. <br>

<a href="http://en.community.dell.com/support-forums/desktop/f/3515/t/19342590">http://en.community.dell.com/support-forums/desktop/f/3515/t/19342590</a></p>

<p>excerpt: "DirectX 9 was released for Windows XP and it wasn't until Windows 7 that the O/S supports >16M colors. I haven't investigated, but I doubt that even DirectX 11 supports this." (age of comment, Aug 2010)</p>

<p>It made me go "hmm..", yknow?</p>

 

[digitaldog]

<blockquote>

<p>I was just wondering if DirectX and a good video card could produce billions of colors or something like that, why not use it?</p>

</blockquote>

<p>You can't see 16.7 million let alone billions of colors. Don't confuse the encoding of data (24-bit, 48-bit) with anything to do with what<strong> you can actually see</strong> or gamut. Gamut and number of colors have nothing to do with each other. Gamut is range not number of colors. A further explanation:<br /> <img src="http://www.digitaldog.net/files/ColorNumbersNotColors.jpg" alt="" /></p>

[angie_nelson]

<p>I see what you're saying. What is the gamut possible in video games? Does direct-x and open gl work from RGB? If so, how can they define so many more "colors"? </p>

<p>I remember those parabolas of human visible spectrum, and the gamuts of rgb, and various printers superimposed... I'm not convinced that "you can't see millions/billions of colors" What you've shown is a mapping of some numbers from one system to another (a conversion). I'm not sure what L, a, b has to do with it unless you're saying my screen can only work in Lab values? Or according to the Lab theory of human vision, this mapping takes place? I'm not sure where to apply what you've said. My eyes? My monitor? </p>

[digitaldog]

<blockquote>

<p>What is the gamut possible in video games?</p>

</blockquote>

<p>What's the gamut of the display? There's your answer. </p>

<blockquote>

<p>I'm not convinced that "you can't see millions/billions of colors"<br /></p>

</blockquote>

<p>Find a source that says you can, then lets talk (you can't). </p>

<blockquote>

<p> I'm not sure what L, a, b has to do with it unless you're saying my screen can only work in Lab values?<br /></p>

</blockquote>

<p>Lab defines human vision (as best as the current technology allows). Two RGB values that produce the same Lab value are the same color! </p>

[digitaldog]

<p>http://hypertextbook.com/facts/2006/JenniferLeong.shtml<br>

Take your pick, 100,00 or 10 million. 16.7 million, billions? No way. </p>

[angie_nelson]

<blockquote>

<p><a href="http://hypertextbook.com/facts/2006/JenniferLeong.shtml" rel="nofollow" target="_blank">http://hypertextbook.com/facts/2006/JenniferLeong.shtml</a><br /> Take your pick, 100,00 or 10 million. 16.7 million, billions? No way.</p>

</blockquote>

<p>Okay, then I suppose the real question is, are the RGB colors the correct ones for our eyes to react to? Or do most of them just map to some variation of white or gray or neutral and are not helpful to express what is seen in the real world?</p>

<p>Basically, how can we get the screens and printers to just print and show the whole gamut of human vision like this?:<br>

<a href="http://www.photocanvasprint.net/rgb-vs-cmyk/images/gamut.jpg">http://www.photocanvasprint.net/rgb-vs-cmyk/images/gamut.jpg</a></p>

 

[digitaldog]

<blockquote>

<p>Okay, then I suppose the real question is, are the RGB colors the correct ones for our eyes to react to?</p>

</blockquote>

<p>The question makes no sense to me, sorry. </p>

<blockquote>

<p>Basically, how can we get the screens and printers to just print and show the whole gamut of human vision like this?</p>

</blockquote>

<p>You can't. Impossible. It's like asking how can you get your dog to see colors like you can. <br>

Color isn’t a wavelength or property of light. Color, is a perceptual property, something that occurs deep inside our brains. So if you can't see it, it's not a color. As such, colors are defined based on perceptual experiments. Color is not a particular wavelength of light, It is a cognitive perception. Another term is Color Values which refer to human perception and specifically to colorimetry. Lab, Luv, XYZ, Yxy, etc are all color values. That's what you are seeing in that gamut plot. </p>

[angie_nelson]

<p>Then why do people complain about it? Why is it said that RGB "isn't enough"? Why is it said that "CMYK isn't enough"? etc? Are people complaining about nothing?</p>

[angie_nelson]

<blockquote>

<p> Color is not a particular wavelength of light</p>

</blockquote>

<p>NASA disagrees:<br>

<a href="http://science-edu.larc.nasa.gov/EDDOCS/Wavelengths_for_Colors.html">http://science-edu.larc.nasa.gov/EDDOCS/Wavelengths_for_Colors.html</a></p>

[angie_nelson]

<p>Getting back to the point, this discussion should probably be closed soon unless someone has some information about the gamut available to gaming programmers while using DirectX.</p>

<p>I appreciate everyone's replies. </p>

[digitaldog]

<blockquote>

<p>NASA disagrees:</p>

</blockquote>

<p>No it doesn't, you have failed like much of this series of posts, to understand basic color theory! The article discussed a wavelength of light, not the perception of color that result within our brain due to that wavelength. </p>

<blockquote>

<p>Why is it said that RGB "isn't enough"? Why is it said that "CMYK isn't enough"?<br /></p>

</blockquote>

<p>Another statement that illustrates you don't <strong>as yet</strong> have the basic concepts of color and color management. RGB what? CMYK what? You want to understand color spaces and gamut? Start here:</p>

<blockquote>

<p>Everything you thought you wanted to know about color gamut<br>

A pretty exhaustive 37 minute video examining the color gamut of RGB working spaces, images and output color spaces. All plotted in 2D and 3D to illustrate color gamut.<br>

High resolution: http://digitaldog.net/files/ColorGamut.mov<br />Low Res (YouTube):

</blockquote>

<p>Maybe that will start you on the path of asking the appropriate questions that can be answered.</p>

<blockquote>

<p>Getting back to the point, this discussion should probably be closed soon unless someone has some information about the gamut available to gaming programmers while using DirectX.<br /></p>

</blockquote>

<p>You're rather confused. That gamut is an attribute of the output device, <strong>the display.</strong> I already answered the question: What's the gamut of the display? There's your answer.</p>

[angie_nelson]

<blockquote>

<p>you have failed like much of this series of posts, to understand basic color theory</p>

</blockquote>

<p>Is there some other way to interpret this other than a personal and professional attack? <br /> Again, I believe it is time to close this thread as it has become too fractious. I asked a simple, "I wonder why not" type of question, and you jumped on me trying to prove that don't understand this or that (ignoring the original question in the process). Color theory is indeed difficult to understand but I'm farther along than someone who claims that color is not wavelengths of light. That much I understand.</p>

<p>OK, the display can only display what, 16 million+ colors? But my question was couldn't DirectX do a better job than RGB at displaying even more colors than that (more wavelengths within the visible spectrum)? Y, N, IDK? It wasn't a discussion on whether or not all displays can show it to you. A wide gamut display may still be needed, or a redesigned one, or maybe not, I don't know personally, which is WHY I ASKED.</p>

<p>Did I fail to say thank you for people's replies? I'm pretty sure I did that, even the replies that I disagreed with. If not, I'll thank you now. I think the videos especially are very good, but of course, I already saw digital dog videos several times a while back when I was relearning it. Having returned to photography after a hiatus, this is the second time I had to understand gamut, color spaces and all of that. This question was a product of my thought process. </p>

<p>FYI - there are no "appropriate" questions, just questions. While I think you make really good videos, I hope you don't teach college classes. It would be like closing minds instead of opening them to tell people there are "appropriate" questions. </p>

<p> </p>

[digitaldog]

<blockquote>

<p>Is there some other way to interpret this other than a personal and professional attack?</p>

</blockquote>

<p>Nope, it is a suggestion to tackle basic color theory before asking qustions that make no sense and can't be answered. Like <em>what does the state of California weight? How many pins can dance on the head of an angel? </em>etc<em>. </em> As you were told in the first post, <em>Angie, to be honest, you're mixing up things that are completely unrelated.</em><br /> <em><br /></em>And again, Color, is a perceptual property. So if you can't see it it's not a color. Color is not a particular wavelength of light. It is a cognitive perception, the excitation of photoreceptors followed by retinal processing and ending in the our visual cortex, within our brains. As such, colors are defined based on perceptual experiments.</p>

<blockquote>

<p>OK, the display can only display what, 16 million+ colors?</p>

</blockquote>

<p><strong>No</strong>. You're confusing encoding values with colors. You can't see 16 million colors. So a display can't produce 16 million colors. You were given an example above with two values that produce the same color. </p>

[angie_nelson]

<p>If color is so "perceptual" then how can we be sure we are all seeing the same thing? Don't you remember being young in art class and asking the teacher if the color red they see is the same one you see? And discussing color blind people? Color is independent of what happens in the brain or it will have to be full of millions of personal exceptions. </p>

<blockquote>

<p>the excitation of photoreceptors followed by retinal processing and ending in the our visual cortex</p>

</blockquote>

<p>No, it's the potential excitation, and the unique excitation.... which is why I am not bothering with it and focusing on the wavelengths of light instead, which have the potential to excite, but might not. </p>

<p>Anyway you're getting lost in the minutiae. Neither you or I know what the gamut diagram of DirectX is in comparison to RGB, when mapped over the visual spectrum, so this argument is going nowhere fast. </p>

[q.g._de_bakker]

Angie, Andrew is right (and so was Wouter).<br>It would really be a good thing to read up on the subject matter (some links were provided. You can find much yourself too). That you continue to get told "you are mixing up things" is not because people want you to say thank you, but because, well... you continue to mix up things. ;-)<br>A few 'concrete' answers: colour is related to wavelenght of visible light, but colour vision turns out to be much more complicated, involving much more than that.<br>DirectX is an API. A collection of preprogrammed functions that make programming graphics easier. It is not a colour producing method nor colour management system.<br>You give three options as answer to your question, Yes, No, or I Don't Know. The answer is No. And that was the answer you got straight away.

[Tim_Lookingbill]

<blockquote>

<p>Then why do people complain about it? Why is it said that RGB "isn't enough"? Why is it said that "CMYK isn't enough"? etc? Are people complaining about nothing?</p>

</blockquote>

<p>Perceived gamut of any one particular color is directly related to luminance level either through a transmissive display or reflective media such as a print.</p>

<p>How bright can a saturated color be pushed to the limits of saturation by increasing it and surrounding brightness until it stops changing its appearance. Gamut plots don't show this relationship making it difficult to understand color spaces in relation to color volume from a practical sense.</p>

<p>What makes it even more confusing is no one can test whether these colors exist due to the fact most colored objects in the real world captured by a digital camera fit within sRGB just fine. There are current synthetic dyes, pigments, fabrics and plastics whose colors expand beyond sRGB, two of which-an intense thalo yellow/green and a deep, intense cyan-I've actually seen with my own eyes on a print off a Fuji Frontier drylab (inkjet) printer at Walgreens, the cyan is even beyond AdobeRGB. I wouldn't have known this until I saw the print.</p>

<p>Does a color not exist just because we haven't seen it? Will it exist if we view it in brighter light? On more advanced imaging systems? Current technology is showing the value of larger gamut models for encoding color that didn't use to exist until today. The green leaf shot I took of the Fuji inkjet inks on glossy paper fluorescing under the light posted below was not possible 20 years ago, maybe even 10 years ago.</p>

<p>This would've helped a fine art gallery owner I was working with back in 2006 who complained about digital captures and inkjet reproductions on canvas of his client's fine art paintings not capturing certain colors of paint that fluoresce similarly viewed under museum halogen lights similar to what's happening with the green leaf example below.</p>

<p>He was using an old HP wide format printer. I couldn't tell at the time if it was his camera or printer preventing him from achieving his goals because I've never seen any inkjet or commercial CMYK press fluoresce under light like that. The display and print couldn't reproduce it just as a current AdobeRGB display couldn't reproduce the cyan on the Fuji inkjet print.</p>

<div></div>

[digitaldog]

<p>Angie, your question makes NO SENSE because you don't understand enough of the fundaments to ask the question that can be answered. Sorry. <strong>DirectX</strong> is a set of low-level Application Programming Interfaces (APIs), it has NO color gamut. Neither does your motherboard or video card or pigs or a plate of pork chops!<br>

Your question is like someone asking in the photo forum:<br /><em>I want to buy either a Nikon or Canon, which has less calories? Which can hold more milk? </em><br>

See, the question can't be answered, it makes no sense. <br>

An output device can have a color gamut once we measure it's RGB primaries and plot that within human vision like the plot you asked about. A display and it's video system can all contribute to this but asking about the gamut of an API makes zero sense. It's a question that shouldn't be asked because it can't be answered. It has nothing to do with color gamut in a single context (and <em>maybe</em> not within the video path). </p>

<blockquote>

<p>If color is so "perceptual" then how can we be sure we are all seeing the same thing?</p>

</blockquote>

<p>I'll answer but it's way out of line based on all the other items about color theory you need to get up to speed on <strong>first</strong>. Colors are defined based on perceptual experiments such as those done in the 1930's by the Commission Internationale de L’Éclairage:<br>

http://en.wikipedia.org/wiki/CIE_1931_color_space<br>

I don't think this is going to help you at this point, but at least you can attempt to get your head around how color is defined using these perceptual experiments such we can <em>attempt</em> to plot the gamut of human vision and from there, a display which may or may not have anything to do with DirectX (which it self has no color gamut)! </p>

<blockquote>

<p>Anyway you're getting lost in the minutiae.<br /></p>

</blockquote>

<p>No, you simply don't understand the basics of the topic of which you're trying to get an answer to question that doesn't make a lick of sense. Three of us have tried to point this out to you already. </p>

<blockquote>

<p>Neither you or I know what the gamut diagram of DirectX is in comparison to RGB<br /></p>

</blockquote>

<p>That statement again makes no sense. Comparison to WHAT FLAVOR of RGB? Until you have a color space, the term <em>RGB</em> alone is meaningless. </p>

[johne37179]

<p>As an aside, try this test: Look at an image with each eye separately -- close one, then the other. If your eyes are like mine and most people I've tried this on, each eye has subtly different color perception. Extrapolating that there are probably no two people whose color perception is absolutely identical. Probably, no way to test that either. This may be the ultimate futile effort when you get way down in the weeds.</p>

[Wouter Willemse]

<p>Angie, sorry I didn't exactly check back. Things apperantly got a bit side-tracked, but to come back to your point on DirectX a bit more elaborate, hoping I'll succeed better in showing why I believe you mix up things. I am going to simplify things, so probably Andrew is going to disagree with parts I write for "scientific" correctness sake. Doesn't matter, we all have our pet peeves.</p>

<p>Consider at this point only the PC and the screen. A program running on your PC creates an image that needs to be shown on the screen. Such program could be Word, a game or Photoshop. All of them do their own internal calculations and work, and in the end create an image (a frame) that the monitor needs to display. This image is sent via the videodriver and videocard down the wire to your screen, which then does its best to show this image to the best of its abilities.</p>

<p>Breaking down this chain of events, DirectX and OpenGL sit exactly at the beginning of this chain. They're "engines" that generate images that eventually need to be sent to the videocard.<br /> For the construction of that image, bit depth for each colour does come into play. Consider that DirectX (and/or OpenGL) performs a lot of the calculations needed to 'construct' each individual frame in a game - calculating the shape and position of 2D or 3D models, their textures, colour, transparency, brilliance etc, and applies the effect of the lights in the scene (colour, shadow, highlights etc.). The processing of all this is done with a fixed amount of bits per channel too, and a fixed amount of channels as well. The more bits are used for each channel, the more accurate the final image; but it also means more processing. (<em>Note: to make matters more complicated, the videocard does already a lot of this work, but in its "capacity" as a highly specialised processor</em>.)<br /> So, all this just discusses how internally the 'engine' of a game constructs and calculates the image that it will send to your screen later on. This stuff exists only inside the RAM memory of your PC, and is *not* the final image you see on screen. The accuracy that a PC can have internally to calculate on images can be a magnitude bigger than what any commercially available screen can display at this moment, or in fact a magnitude more than we manage to send over copper wiring. Usually, however, it is kept down to not make the CPU work a lot more on accuracy we do not need. But also consider HDR Imaging software, which can use up to 32 bits per colour channel to do its calculations to create high-bit-depth images with a extremely wide dynamic range and or a superb accuracy. But all this is still just something that will be sent down to the videodriver/videocard to be next sent to the monitor.</p>

<p>The videodriver and videocard scale the image to the resolution set, and scale the colour to the bit-depth you have set in your OS (typical: 8 bits per channel). Theoretically, it can be scaling up or down the image that came in from the software. (<em>Note: most applications know from the operating system what the resolution is, and hence can adapt; the level of interaction is a bit simplified here, just to make the concept more clear that these are distinct different steps</em>)<br /> Consider colour management as a final step in all this: the image is ready to be sent to your screen; each pixel with RGB values between 0 and 255 per channel. The colour management maps each value to the calibrated value on your screen, ensuring 255,0,0 is vibrant pure red, and 128,128,128 a pure mid-grey and so on. So, it makes sure the monitor receives the exact right signal for its capabilities.</p>

<p>So, this is why DirectX or OpenGL cannot and do not influence colour management, nor why they cannot be used to enhance the gamut of your screen. They're simply different steps, steps that sit well before any type of colour management is applied.<br /> I hope that helps a bit, I'm very obviously not a teacher so my explanation may be fuzzy in points and certainly for minutiae will be disputable, but in broad lines it's where the difference between DirectX and colour management is.<br /> ____<br /> As a sidenote; the thread on the Dell site discusses the ability of some selected monitors and videocards to use 10 bits per channel, instead of the usual 8. The DirectX related part is just a diversion in the answer, and apparently confused (the answer is poorly written, so not too strange). It's just an example where having a monitor display capable of handling more bits per colour channel is just a waste - because the input is limited to 8 bits anyway - that is: the program discussed doesn't generate more than a 8-bits per channel image.</p>

[angie_nelson]

<p>Soo... what I'm gathering from this discussion is that the DirectX game might be capable of any bit depth of color it chooses, and may advertise as such (billions of colors!), but when it is presented on (nearly?) any monitor, the limits of the monitor apply and therefore we're back to 16 million colors possibly displayed.</p>

<p>Wounter - For not being a teacher, I think you explained it quite well. I'm still not sure I understand what you mean by "It's just an example where having a monitor display capable of handling more bits per colour channel is just a waste - because the input is limited to 8 bits anyway" -- I'd be of the opinion that, (like colors you can't see if you're color blind) it'd be better to have the bits there and provide an opportunity for programs to generate more than 8 bits pre channel. Without that possibility, it's a certainty that the program won't progress.</p>

[digitaldog]

Bit depth has nothing to do with color gamut! They are different attributes.

 

If the entire preview path is not high bit, more than 8 bits per color, the net result is NOT high bit. The OS, application,

video card, etc must ALL support high bit.