In this video, the rolling shuter is composed of 4 components that add together:
1. The A7S rolling shutter (duh)
2. The raster scan display (equal to ~1/24 or 1/30 s for the 4K monitor)
3. The unknown rolling shutter speed of the camera filming the monitor
4. The raster scan display you are watching (e.g. 1/60 s, with rare exceptions like Lightboost)
explanation for display skew:
The rolling shutter of the A7S might still be the dominant factor, but I'd wait and see for some real tests.
âŒ˜7 opens the videoscopes window. In the dropdown menu choose the >Waveform >Luma (this could be your standard videoscope, FCP always displays your latest choice on next start). This shows you a "scientific" representation of what's there. The legal parts live between 0 and 100, what is called broadcast range. Being the uppity DSLR freshmen we are, we dismiss this shit, and "bring the values into the range". You should not believe in any 8% rule, this is just changing one ancestral wisdom for the next. You should color correct using the exposure window and the balls for highlights and shadows whilst watching the lines in the waveforms.
What I guess from the questions about any ominous workflow, there seems to be some insecurity of what acutally should be done.
Don't be too scientific. You do this for a better looking image, so please trust your eyes!
What is suspected by the lovers of 5D2RGB is that Quicktime cuts of values because it misinterprets the ranges. With 5D2RGB you can manually override the broadcast flags and choose full range for any footage.
FCP X can edit all your footage natively. Of course, you really should have Lion (wait a few weeks before upgrading to 10.8, Mountain Lion, there seem to be sum bugs still) to complete the AV-Foundation framework. There you can easily judge for yourself, if a file from your own camera is treated better with 5D2RGB (download the lite-version, it's free) than as original or transcoded by FCP X to optimized media (ProRes as well). With material from 7D and GH2 I found there was nothing lost (judged by the waveform) with any of the methods.
If you feel better then, buy the batch-ability of 5D2RGB and check [i]full range[/i] before you hit [i]convert. [/i]Note, that the values now do not fit into 0-100 in the other direction: The lowest values don't touch zero, the highest (even if clearly flatline-clipped) don't touch one hundred. Left to my own devices, I would interpret this as an actually [i]narrowed[/i] range, as baked-in false values.