@ronadair
Your insight is very helpful -- you've got a great brain for this stuff and you're very well informed. I want to be clear about the fact that I'm not aiming to be on the opposite side of the table from you on this one... I'm merely trying to weed through the onslaught of quasi-mis-information on the web about how to make the very most of the differences the d800 offers.
The output from the d800(e) HDMI is a 10-bit 4:2:2 stream if left untouched. I'm seeing some websites claiming the output is only 8bit -- not sure where they are getting this info. I'll chalk it up to the confusion around te d800's specs -- much like the initial stories that it couldn't output 1080p24... we know it can.
I should also point out, again, that I am coming from a VFX and post background -- my first concern is getting the best data possible; and keeping it as clean as possible throughout the entire production process. My workflow spec is simply based on proper color theory and the VFX process -- I don't expect this will be relevant to the way many work if they aren't concerned about keeping as much flexibility possible through out the process. Going directly to ProRes will probably be the best solution for many that really aren't concerned that much about post/grading/vfx... but if you are a film maker, this stuff should definitely concern you. I am amazed when I speak with a director that has no clue about the differences between 4:2:0, 4:2:2, 4:2:2+0:2:2, and 4:4:4... yet they expect to be able to "fix it in post"... And they will sit there and tell me how awesome their 4k footage is... until I ask them if they know what debayering is... and that their RED isn't really shooting a true 4k... and how I can take 2K rgb 4:4:4 footage, process it faster with more accuracy, and it will look better, in a shorter time, than their "4k" 4:2:2 footage run through the same process... I'm simply getting into the finer points of the output for those that might find it helpful. The ability to save the "untouched" output stream from the camera directly to disk without a codec is the very reason many opted for this camera.
Those coming from the world of VFX and post most likely have the horsepower and storage space to support a linear floating point workflow already... so "uncompressed" 10bit vs. ProRes HQ will make very little difference to a system that's setup to fly with 2k muli-channel floating point exr frame sequences.
I'm betting a lot of members on this forum saw the DP review write-up of the d800 and d800e -- they do a nice job of touching on the advantages of uncompressed in the video section:
http://www.dpreview.com/reviews/nikon-d800-d800e/30
Another interesting article that breaks down different recording/caputure solutions available for the d800 and mentions the many strengths the Ninja2 has going for it... and ranks it at #2 due to the fact that it is unable to capture the "uncompressed" stream:
http://nanofunk.net/raw-video-for-dslr-with-magic-lantern-and-black-magic-hyperdeck-shuttle-vs-atomos-ninja-5d-mkii-550d-rebels/
Regarding visible differences -- that's not really a valid standard of measurement. Water and hydrochloric acid "look the same" but they sure behave differently in the lab! People see things differently. What is an obvious difference to some may go completely unnoticed to others... If you are planning to post process/grade your footage then you are operating on a level much more accurate than the human eye... and assuming you are using decent software, the computer will always see the difference and it will work better with the uncompressed data. Why would you leave it up to your eyes to prove the difference? To recap -- your eyes may not see a difference -- the software definitely will.
Again, I do not blame the Ninja 2, I blame the codec. I think the Ninja2 could easily do what I'm asking with a firmware upgrade.
Either way, after buying a few fast memory cards at around $100 a pop, the better and better an external capture solution starts to look!