Hello everyone. I had searched quite extensively when looking into a solution for my color banding on an ancient Apple 30" cinema display which was on it's last leg. That search led me to 10bit and it's surprisingly obscure implementation in PS. Allow me to share my experiences so others doing google searches can potentially be helped, and also pose a few questions that Nvidia could not answer for me.
First, I am running Windows 7 x64, latest drivers/patches on everything. Everything includes:
PNY Quadro K4000
Dell U3011 via Displayport
Windows 7 x64
Photoshop CS6 x64
2560 x 1600 Desktop resolution
So the long story here is that I did not think my combination was working. I tried the ramp.psd from AMDs website, and saw absolutely no change. Frustrated, I hit the internet, and unfortunately, found a plethora of AMD/ATI based help, but very very little Nvidia Quadro support, msot likely due to their "it just works" mentality (http://nvidia.custhelp.com/app/answers/detail/a_id/3049/~/how-to-enabl e-30-bit-color-on-windows-platforms). The help files say that the Quadro and PS will just detect the ability to use 10bit color depth and use it, but there is no way to confirm that that combination is in fact compatible without actually seeing the result. No screen overlays, no checkboxes in the driver (as far as confirmation goes). You can force 10bit in the quadro driver, but that doesn't tell you it's actually working.
So I got on the live support with a helpful tech from Nvidia and eventually got to the bottom of it. Things I learned:
- with Photoshop maximized at 2560 x 1600, 10bit did NOT work with any of the flyout panels open. This was the number one hair-puller, and why I thought this was not working at all the entire time. I had my history panel out. At some point while chatting with the Nvidia rep I decided to take photoshop off of full screen so I could see both PS and the chat window at the same time. The PS window then occupied maybe 70% of the screenspace and poof.. the bands disappeared. I slowly drag the corner of the screen out to make it larger and at some point the bands pop back in. Odd. So I maximize but close the flyout history panel and poof, bands disappear. If I'm fullscreen and hit tab to hide shelves and panels, smooth as butter as well. The Nvidia rep summized that it had to do with OpenGL overlays interfering, but had nothing to offer in the way of the resolution/screen size issue.
So I either have to view 10bit as a toggle, or not use my pallets as flyouts in the main PS window, or not maximize PS.
So, I'm sure there are other people running this combo, so I'd love to hear your results. ATI/AMD cards are clearly a different driver set, and thus, probably totally different results. Nvidia claims there is NO resolution limit to 10bit written into their driver, so the resolution being the problem did not make sense to them. That laid the blame purely on Photoshop. I'd love to hear from Adobe on the matter.
In any case, hopefully this helps someone struggling as to why they see no difference on that gradient image. It is literally night and day, so if you aren't seeing that "pop," it's just not working.
I must say that, as I've read in countless other posts, in actual photos, this 10bit can be a very minimal improvement, but it is noticeable in some of my shots where I have color grads on skintones. The red burn is gone, and it's a much smoother transition. This is insanely evident in purely CG created gradients or gausian blurred overlays, however. I have a few CG-assisted background hotspots with the addition of a gausian blurred screen of white, and it has helped that quite a bit.
I, for one, welcome out 10bit overlords. Hope this helps someone.
Great post, thanks for sharing your experience.
I'm not a nVidia user, but I am looking at getting a Dell U3013 when they are released, and I have been interested in the 10 bit saga, because I am hoping to be able to use it.
Have you tried disabling desktop composition (Aero Glass)? I have heard there can be conflicts there that prevent drivers from enabling full 30 bit color.
Something I've been curious about, since you DO have the ability to get 10 bit color under some conditions... If you create a 16 bit dark gray gradient in the ProPhoto RGB color space, do you see any color inaccuracies (reddish and cyannish banding) in the display? With 24 bit color it can be noticeable. Here's a test gradient file you can check it with... Click on it to make it full-sized, save it to disk, and open it in Photoshop.
"Have you tried disabling desktop composition (Aero Glass)? I have heard there can be conflicts there that prevent drivers from enabling full 30 bit color."
This is disabled automatically by the Quadro driver the second you turn on the Deep Color option. And no, as someone who uses his machine for production moreso than pleasure, I immediately turn all that stuff off anyways as it's useless.
I will test that file after work tonight when I get home. No 10bit here. Irony.
Thanks. I'll also try to make a video/animated gif of sorts to illustrate the difference to your naked eye. I was also quite clueless as to the "real world" implications when I started looking into this, as all you really find are generated gradients, which seem doctored at times. Seeing the pop happen with your own eyes is quite interesting, and a definite selling feature.
Oh, also, I will add that I previously ordered the new U2713H from Dell, and I was 100% disssatisfied with the backlight bleed, as well as the color representation and quality control. Mine had a very terribly noticeable yellow color bleed in the lower left. I found this to be a regular occurance for these newer LED backlit monitors, and ultimately returned it and went with the older CCFL U3011. It is far superior IMO and only slightly more expensive. Highly recommended. This forum posting swayed my ultimate decision away from the U3014: http://hardforum.com/showthread.php?t=1751499
Hm, I guess I'm out of touch - I didn't realize the new monitor was out already. Thanks for letting me know.
I had heard it might be possible the U3014 would have positional calibration (i.e., be able to be calibrated so the display response was more flat). Guess I need to do another round of looking at review sites.