Same thing with the video, it can probably record at 1080 4:2:2 ..but, since it can't output that (like a D800 can, or like many other pro camcorders can), it relies on what Canon's codec outputs. The results are videos with fewer lines that look a little softer. Sometimes that is a good thing, for the right effect (interviews, things with people)...sometimes its not (architecture, nature, etc)
Codec has almost nothing to do with sharpness. Panasonic GH2 is sharp despite it has very poor bit rate codec. The better codec and higher bitrate only reduces artifacting, and has not really much to do with recorded sharpness.
Simple experiment:
Shoot a video with RED Scarlet/Epic.
Shoot the same video with the 5D mark II or mark III
and shoot the same video with the 7D and finally 1D mark IV.
Now scale all videos to the same resolution, 1920x1080 (in other words you need to only scale the RED from 4K to 1080p). Then drop the bit rate of the resulting video lower than the lowest denominator of these, e.g. lower than the Canon's bit rate. Now compare the actual resolution of the images.
What you can see very easily is that:
1. RED is very sharp
2. 5D mark III is next
3. 5D mark II is next
4. 7D is second last
5. 1D mark IV is the very last of this group.
When you compare 1, 2 already you get a feeling that anything less than the mentioned 1 is crap. If you look sharpness a bit through fingers, then you can add 2 and 3 to the acceptable category. If you are irritated with moire, you can drop everything below 2 (in other words, mk II, 7D and 1D are unacceptable).
I am excited for the Zacuto 2012 camera test to prove this this year. I already knew the results, but want to see it done "properly" so that it is easier to point out to people who do not understand that the number of pixels in the file do not equal to the resolved row & line resolution of the camera. It is like comparing Canon L glass with some cheap plastic lens and say that both have same magnification. Indeed yes if you don't care about quality.
