3

Before I make a significant investment in hardware to get to 10-bit per color (10bpc) nirvana I want to make sure I understand the complete path from shutter release to enjoying my prints. I feel like I understand the PC hardware part of the equation:

10bpc capable:

  • OS (Windows 7)
  • Display (HP zr30w)
  • Video Card (Quadro FX 580)
  • Drivers (Quadro FX 580 10bpc drivers exist)
  • Cable (DisplayPort)

I am a little less unsure about the application software but I think both Lightroom and Photoshop CS5 can handle 10bpc but there might be some initial configuration needed. Can someone confirm this for me? The part that I am completely lost on is the printing. I have googled my way around a bit but I cannot seem to find any info on equivalent bpc for printers or for print shops. So here are my questions:

  • Are there any printshops you can provide 10bpc files?
  • Can you print 10bpc files to your printer and if so is there anything special you need to do to take advantage of 10 bits?

Thanks.

stephenmm
  • 133
  • 5
  • 2
    "A device that is able to reproduce the entire visible color space is an unrealized goal within the engineering of color displays and printing processes. While modern techniques allow increasingly good approximations, the complexity of these systems often makes them impractical." - Wikipedia – RedGrittyBrick Feb 01 '11 at 00:23
  • 1
    I'm certainly not trying to reproduce the entire visible spectrum but the 10-bpc up to the display seems to be practical with pro-sumer products. I am going to buy a nice display anyway so I just want to take full advantage of it for printing. Also trying to improve my understanding. It may be that 10-bit makes no difference in printing and that gamut is the primary deciding factor in print quality. I doubt it, but maybe. – stephenmm Feb 01 '11 at 00:50
  • I am still not clear on a couple points: 1) Is there any photo lab out there that supports 16-bpc files? 2) Photoshop CS5 does not display 10-bpc to the monitor (even with all the correct drivers and hardware) but Lightroom does, correct? – stephenmm Feb 02 '11 at 17:07
  • I was able to pick up a refurbished Dell u3011 for a little over $800 and am VERY happy with it so far! My next step is to get the 10-bpc images to paper. I am considering a dye sublimation printer to have better gradients and generally better quality prints (plus I HATE inkjets and there clogged nozzles). I would like to purchase the Mitsubishi CP-3020AU but I cannot find it for sale anywhere. May start a new thread about my hunt for a dye sub printer. – stephenmm May 09 '11 at 04:49

2 Answers2

3

It is possible to print with full 16-bpc color to printers that support XPS (Microsoft XML Paper Specification) drivers. I know the Canon PIXMA Pro9500 II has a Windows 7 64-bit XPS driver that supports 16-bpc printing (vs. the current 8-bpc printing).

It is important to note that 16-bpc printing is only supported with XPS capable RIP software. At this time, I only know that the Canon Easy-Photo Print Pro software is capable of full 16-bpc printing. I've heard rumors of other RIP's that support 16-bpc XPS, but I have not found any specifics.

Additionally, it should be clear that 10-bpc displays and 16-bpc printers are not directly linked in any way. You can process a 16bit per channel (48bit) RAW or TIFF file on a normal display, and still print it to a 16-bpc printer. You can also obviously use 10-bpc displays with 8-bpc or 16-bpc printers.

jrista
  • 70,728
  • 15
  • 163
  • 313
  • Do you ever/would you ever go to a lab to print your > 8-bpc files? – stephenmm Feb 01 '11 at 01:32
  • The only lab I have really used is BigPhotoHelp.com, and they use HP printers that I am pretty certain do not do any more than 8-bpc printing. They accept 16-bit TIFF files in the Adobe RGB gamut, however it is still best to convert to sRGB and 8-bit TIFF files for lab printing. Printers and RIP's that support XPS 16-bpc printing are VERY rare, as the technology is pretty new and fairly "unvetted". I happen to have a Canon PIXMA Pro9500 II, and I happily stumbled upon the new driver a couple months ago, and been experimenting ever since. – jrista Feb 01 '11 at 02:07
  • With bright, pure white photo rag paper (no optical brighteners), there is a definite, if maybe still subjective, feel of much finer tonal gradations with the 16-bpc driver and EPP Pro. As far as detail goes, the Canon EPP Pro software really isn't the best RIP, and hard edges don't really look as good as when I print from say qimage (although again, that is a subjective comparison of my own.) I think there is a lot of potential in 16-bpc printing, but software is lagging farther behind than hardware. – jrista Feb 01 '11 at 02:10
3

I want to make sure I understand the complete path from shutter release to enjoying my prints.

An important part of that path is the bit between print and enjoyment.

Assuming you are using a three-colour representation (RGB or HSV for example), a crude estimate would suggest that 10-bits per channel provides (2^10)^3 distinct color values. That is 1,073,741,824 values - over a billion. However some say that human eyes can perceive no more than 1,000,000 to 7,000,000 distinct colors under optimal laboratory conditions.

Current display and printer technologies are only capable of presenting a small subset of the gamut of colors that a human can perceive.

This suggests that even if your software allows you to send 10 bits per channel of color information to a printer or display, most of that information does not end up being seen.

As a measure of perceived quality of color reproduction, I suspect that gamut is more important than number of bits per color-channel.

RedGrittyBrick
  • 2,648
  • 1
  • 16
  • 21
  • 1
    Human perception is both color and monochrome. While studies show that humans can generally see 2-3 million colors (one study stated 10 million), we can see far finer gradations of tone. The total number of "colors" reproducible by 10bit or 16bit images includes both degree of chromaticity (hue and saturation) as well as luminance. The total number of "colors" perceived by the eye is pretty much the same for any bit depth, however the fineness of tonal gradations DOES improve as bit depth is increased. Increased bit depth is more about levels of brightness than volume of distinct colors. – jrista Feb 01 '11 at 00:45
  • @jrista: The source I quoted included luminance or brightness in the calculation - "1,000,000: the number of colors (combinations of hue, saturation and brightness) that the eye can discriminate under optimal laboratory conditions." If increased bit depth was more about levels of brightness then couldn't we adequately represent perceivable colors with (say) 6-bits of hue, 6-bits of saturation and 12-bits of value? So far as I know, no cameras, software or output device follows an asymmetric model. – RedGrittyBrick Feb 01 '11 at 00:57
  • Great comments guys, thanks. My anecdotal comment would be that people who use 10-bpc displays are reporting MUCH better image quality with all other factors being equal. I don't have any statistical or scientific studies to point to but the anecdotal evidence is pretty convincing that moving from 8-bpc (~16 Million) to 10-bpc (~Billion) displays with the same gamut makes a big difference. In fact using the same 10-bpc display and providing it 8-bpc data seems to vastly different quality. I would argue this means that we are definitely capable of perceiving > ~16 Million colors. – stephenmm Feb 01 '11 at 01:03
  • 2
    @Red: "If increased bit depth was more about levels of brightness then couldn't we adequately represent perceivable colors with (say) 6-bits of hue, 6-bits of saturation and 12-bits of value?" EXCELLENT QUESTION! Assuming we had a display that was capable of generating pixel colors based on those specific values, yes. Currently, displays and digital sensors work off of a 3-channel RGB model...so, no, we can't do that with current technology. – jrista Feb 01 '11 at 01:05
  • The RGB model is very poor for representing color, HSL is a better model, as it appropriately separates hue (360 degrees), saturation (color purity), and luminance (level of intensity). The human eye has two separate devices that sense these things. Cones are our color-sensitive "photosites", however there are only about 6-7 million of them. Rods are far more numerous, numbering about 120 million, and are responsible for sensing luminance over a monstrous dynamic range. Because of this dichotomy, we are far more sensitive to changes in luminance than to changes in chromaticity. – jrista Feb 01 '11 at 01:09
  • 2
    As someone who spends thousands of hours of my free time screwing around with color calibration of both my computer screens and my printers, and my recent experimentations with Canon's 16-bpc XPS driver, I can personally vouch that higher bit depth produces considerable improvement in tonal gradations in print. Comparing my own Apple CinemaDisplay 30" screen to a LaCie 730 (14-bpc display), the same difference is very apparent. Posterization and aliasing in smooth gradients are non-existent when rendered with higher bit depth. – jrista Feb 01 '11 at 01:12
  • @jrista: so are you saying that many of those 10 bits expended on hue and saturation are wasted and 10 bits may be insufficient for luminance? – RedGrittyBrick Feb 01 '11 at 01:12
  • @Red: It is kind of difficult to explain how those bits are used to produce hue, saturation, and luminance. The problem with the RGB model is that you are only blending different luminance values for three specific color channels. Certain mixes of red, green, and blue, at varying intensities, produce certain colors. With an RGB model, there are times when you have to increase the intensity of a single channel far more than others (i.e. magenta is a blend of maximum red and blue, and zero green.) – jrista Feb 01 '11 at 01:17
  • @jrista, LaCie 730!!! I now bow before you, my king. – stephenmm Feb 01 '11 at 01:19
  • 2
    With 8-bit color, you only have 255 distinct levels of luminance for each channel, which limits both color and tonal fineness. To create specific colors with such limited tonal range, you have to make certain compromises on which distinct colors are reproducible, and in doing so, you also limit distinct levels of tone for any given color (since you are not blending hue, sat. and luminence...you are blending distinct levels of luminance of red, green and blue.) RGB is, as you mentioned, rather wasteful. – jrista Feb 01 '11 at 01:20
  • With 10-bpc color, you have 1024 levels, which expands the range of accurately reproducible color (probably to the maximum of 2-3 million), while also expanding the range of usable tones. Again, there is some waste, since all you are really blending to form distinct colors is the luminance of red, green, and blue. If we hypothesize about an HSL display...luminance would indeed be INDEPENDENT of chromaticity, and that expands your range of distinct tones to a vast number. I honestly can't speak to the plausibility of such a display, but if possible, I think it would be better. – jrista Feb 01 '11 at 01:23
  • @stephenmm: HA! I totally wish it was mine, but alas, no. The day my CinemaDisplay dies, however, there is no question that I'll buy whatever top of the line LaCie is available at the time. There is no question that screen is the best on the planet, and I totally swear by RGB LED now. – jrista Feb 01 '11 at 01:25
  • @jrista If you were to spend < $1500 what display would you buy right now? Would you go down to a 24" just to get the RGB LED? – stephenmm Feb 01 '11 at 01:29
  • If you are talking about Apple's new LED display, it is not actually an RGB LED display...it still uses blue led with yellow phosphor. I actually don't know if Eizo has any RGB LED displays right now, and the LaCie 730 is about $3200. I really want the LaCie, but I am currently saving my money for a 5D III, which I think will serve my interests better. I would also need to get a video card with a DisplayPort to support the 14-bpc output of that screen. If I had to buy a new screen today...I'd get the Apple 27" LED CinemaDisplay. Its only $900, and produces better color than I have now. – jrista Feb 01 '11 at 01:33