C300: a short test C300

[Pave1]

Some quick test from rented C300. The best pictures are at the end.
I try a bad exposure and color corection at post, the first part is canon-log and shutter 1/50, the second part is normal gamma and shutter 1/25 (with 50mm/f1,4 - I used F2,2).



I suggest to download the orig. file.
www.vimeo.com/59308641

 

[Pave1]

With this test a I have a one question - can somebody tell me any details about C300's A/Dc (is it 10bit, 12bit or more)? Is the Digic DV III DSP working at 10bit?

If I shot with canon-log (4000 : 1) then I need the 12bit A/Dc (with 4096 samples). If I have only the 10bit A/Dc then it is only 1024 samples, so I can't get 4000:1, it is 1024 : 1.


I think, that canon-log with 4000:1 needs 12bit A/Dc and 12bit Digic procesor.
I want to buy C100 but I need to know a real limitation of that camera.
Sorry for my english.

The pictures is from the beginning of my short test video (with canon-log gamma), it well illustrates my question (skipping of samples may result in banding). For higher quality look at video, the banding is realy visible (time 00:48).

http://www.videoproduce.cz/images/C300_canon_log_banding.tif

 

[Pave1]

This banding on a smooth gradient with canon-log is realy big, I found only this material with some sort of the answer:
http://thebrownings.name/WHP034/pdf/WHP034-ADD70_Canon_EOS_C300.pdf

page 14 (figure 9):
Normally, the distribution would follow the slope of the gamma curve, which would produce 14dB more noise near black than near white. Although these results follow the general trend of the gamma slope, they do not fit it well.
This could be the results of attempts to optimise the use of digital processing by using non-linear ADCs
(i.e. performing some form of pre-gammacorrection on the analogue signals from the sensor).
Since the camera’s digital processor handles only 8-bit signals at the output, it may not be handling
more than 10-bit data at the ADCs, and so non-linear ADCs make a lot of sense.

It is clear that a better digital processor would result in significantly lower noise values in this camera.

---

With normal gamma or cine1 gamma it can be ok (I think, but I don't have a camera now for the other testing).