So I've been scouring the internets with the utmost diligence lately, waiting for someone to post some examples on Vimeo of D800 footage shot to an uncompressed source. In the dearth of such postings that has ensued since the very recent launch of the camera I've decided to give it a shot. Also included are some quick tests comparing the D800 and 5d mkII.
I'm not the type to shoot professional comparison tests. Wait a little and you'll see people with more intelligence and resources doing proper testing. I only have one lens which I have to switch between cameras and this was set up as a quick test at work with no time. That said, there is still much to be gained from this little venture. It represents real world situations and shows how the camera holds up. If you're the type to complain about how others test things, do your own tests which inevitably will be more useful as they'll be suited to your needs. I did this for myself and just wanted to offer it as it is. Tried to include both skin tones and something with high frequency detail enough to alias.
The D800 was set to record at 23.98 progressive internally and spit out the same (see note). Files were captured through a Decklink card to Prores 422. Standard shutter, aperture at 2.8 on a 55mm Nikon lens, ISO 800. You'll see noise in the test, but you'll see it in real life too so I figure ISO 800 is reasonable for a general test.
I may post these if there's any interest, but I did a quick few tests of different picture settings and found these settings to provide the best image:
Sharpness: "0" (far left, NOT centered in this case)
Contrast: "-" (far left)
On the 5d comparisons the picture profiles are matched to a reasonable degree and the same lens is used with an adaptor on the Canon (obviously).
No grading or manipulation (besides the inevitable recompression) was done on the footage.
Note: While the settings in the HDMI out menu only go up to 1080i, the D800 does seem to output 23.98 if you set the HDMI 'Output Resolution' to 'Auto.' This will need to be tested further, but if all is well, this is a welcome alternative to shooting interlaced dealing with pulldown removal afterward.
I tried posting the originals to ftp but it is under maintenance at the moment. If there is interest in this please notify. I look forward to someone posting some more proper tests with the cameras lensed and profiled as close as possible and shooting some outdoor stuff. I'd love to see trees and foliage compared as I've often wondered if how bad they look on Canon's is solely due to aliasing or in part the compression. Post away professional testers.
Loading more stuff…
Hmm…it looks like things are taking a while to load. Try again?