10 bit colour display path for lightroom 3??

Question

I have been following an interesting forum topic on the PS forum here about 10 bit monitor support in PS CS5.
For some background:
10 bits/pixel is becoming more prevalent with high-end displays these days (it provides 1.07 billion color possibilities instead of the 16.7 million colors available with the regular 8 bit display path). Getting a fully working 10 bit path from the app to the display appears to be more of a challenge though, but a challenge which appears to have been overcome now in PS CS5. I'm wondering if moves are afoot for Lightroom to also support the 10 bit color path??
This is my take for what you need for PS CS5 to display 10 bit colour in windows (sorry, I'm not a Mac person so couldn't comment):
Suitable graphics adaptor - Some ATI FirePro models and nVidia Quadro models support 10 bit color (cheapest is nVidia Quadro FX 580) - the availability/use of the displayport interface is most often the tell tale pointer as to whether the adaptor supports this. Only more recent driver packs provide 10bit API support for apps to utilise, so latest drivers are often required.
Displayport connection between graphics adaptor and monitor (DVI only supports 8 bit color)
A monitor which supports 10bit color - (High end NEC models (PA241W/PA271W models and new Spectraview Reference models in UK) now support this as does the HP DreamColor plus some Eizo and Lacie models to mention a few)
Enable 10bit color support in the Graphics adaptor control panel - this option generally only appears when all of the above is in place, and when enabled, provides the display path via OpenGL, for apps to utilise if they are coded for such (this option is known as 'Enable 10-bit pixel format support' for ATI, and 'Deep color for 3D applications' for nVidia)
Enable OpenGL hardware support (under Preferences, Performance,  'Enable OpenGL Drawing' in PS CS5)
So will this be supported in Lightroom 3 before long? Or perhaps it is already??
Anyone know this?
Thanks, Paul
p.s. I realise the above chain of requirements sounds complex, but I got it working this week for a new PC build and it was actually quite straightforward.

Answer 1

martin-s wrote:
What would be the advantage of this?
If you google around for how many colours the human eye is able to distinguish, you'll find scientific research quoting numbers around 10.000.000.
Horribly false.
The prevalent sRGB color space is covered fairly well by the 8 bits per pixel depth most commonly used ("32 bit" or "True Color" color in Windows), and that is about 16.7 million colors.  While bit depth and color space are different things, once you go to the wider gamuts in other color spaces such as Adobe RGB without using more bits you will start to see banding in similarly colored (blue skies is a good example) areas because there are not enough digital steps to make smooth transitions as the color gradually changes.  Take my word for it (I've seen it in many of my own pictures.) or do more reading for more details if it's not clear, but the key points follow.
First go to:
http://en.wikipedia.org/wiki/XvYCC
and read about the xvYCC color space.  In their reference number 2 down toward the bottom of the web page you find a link to:
http://data.memberclicks.com/site/hopa/2007_xvYCC_forHPA_FINAL.pdf
which is a pdf of a presentation discussing extended gamut color spaces.  Check out slide 3 which is at the bottom left of page 1 to see examples of many things in the real world that are quite visible to humans but outside of sRGB.
Next check out slide 19 at the bottom left of the 5th page to see just how pathetic sRGB is at covering human vision.  The small pictures on the right in this slide show the coverage of the various color spaces of the "Munsell Color Cascade".  The gray areas indicate colors that each space cannot cover, and you can see sRGB has huge blotches of gray.
Munsell did some of the very early definitive lab testing of human color perception.  You can start reading about him and his color system here:
http://en.wikipedia.org/wiki/Munsell_color_system
and continue reading other links found on that page for more info about the extensive lab testing of human color perception.
Looking back at slide 4 on page 1 of the presentation linked above you will find an odd looking colored region that is curved around the top.  It looks like this:
That colored region is the estimated range of standard human vision.  There is a version of that which has the sRGB color space shown on it at:
http://en.wikipedia.org/wiki/File:Cie_Chart_with_sRGB_gamut_by_spigget.png
And that large view also gives you a much better idea of just how pathetically inadequate the sRGB color space is at covering human vision.  Again, bit depth is not color space, but 8 bits per pixel is pretty much maxed out trying to represent the sRGB color space without having to fake colors within it's triangle.
There are probably cases where banding can be seen even in purely sRGB images with only 8 bits per pixel.  And every color that you see in that large region with the curve at the top (in the picture shown in the last link above) that lies outside the sRGB triangle will have to be "clipped", which is converted into some color that does lie within the space.  That means you never see any actual colors that fall outside the triangle even if 8 bits per pixel could have covered them somehow using some other color space.  They're simply lost.
If you do try to use 8 bits for wider gamut color spaces you end up with the banding I mention above because there are too many shades of color to be represented by 256 steps for each of red, green, and blue.  Thus the need for 10 bits or more per pixel.
I hope this helps to dispel any more myths about the limitations of human vision falling within what 8 bits per pixel or sRGB can provide.