C500: AG DOK 2020 Camera Tests: C500ii

[Grug]

The AG DOK in Germany have just posted up their 2020 camera tests (which are great), and I was particularly interested to see that a little theory I had about exposure with the C500ii, was proven correct.

The earlier exposure tests we've seen from the C500ii, suggested appalling noise performance when you underexpose the camera. However, we'd heard it mentioned that the camera re-clocks its ISO at each stop of exposure (like dual-ISO architecture does for a just two levels of ISO in most other new cameras).

The initial tests had suggested a rather strange (and quite unique) situation where you had a camera that somehow exhibited both a very wide dynamic range, but somehow quite limiting latitude at the same time ("latitude" being how recoverable an image's highlights and shadows are). I was very intrigued by this, and wondered how the performance would compare if you used the camera's constantly-reclocking ISO (instead of pulling the image up in post), in order to expose in low-light situations.

Well, these tests seem to answer that quite unequivocally - if you don't have the exposure you need, ALWAYS bump up your exposure via ISO (in one-stop increments from 800 ISO though, the intermediate 1/3 stops are digital gain not analog gain) and NEVER leave it to post. This difference in performance is really quite extreme. On the lefthand side of frame you have the base exposure underexposed and then pulled back up in post, and on the righthand side you have the drop in exposure compensated for by raising the ISO in the camera:


XF-AVC UNDEREXPOSURE:
https://vimeo.com/423444510

RAW UNDEREXPOSURE:
https://vimeo.com/423410789

LATITUDE TEST:
https://vimeo.com/403190248


This same behaviour seems to be born out by quite a few other cameras in the test. So is clearly an important thing to test for.

 

[cyvideo]

Interesting. I can see the difference for sure. This could be a real trap for shooters who are hiring a C500 II and who aren't aware of the fact that they are better off gaining up in the camera than leaving it to post. A few could come quite unstuck leaving it to post gain. Apart from the obvious noise differences and color shifts at various gain steps the other thing that would fairly limit a post gain workflow is that in all three tests where the gain has been applied in post there are obvious sensor banding issues appearing. Even on web delivery off Vimeo I am starting to see horizontal banding appearing on the girl's chest at -2, and at -3 and -4. The banding to me at -3 and -4 are totally unacceptable in all three video examples. Not much room for error correction in post then. You really do have to rely on you cameraperson / DP really being across how to use this camera to get the best out of it prior to post getting their hands on the footage.

Another interesting comparison for me from AG DOK is this one. Quite surprising how the oldest camera of the bunch, over five years old now, still holds up surprisingly well after all this time. Bar of course the one exception which is that of a newer camera that's using a much lower resolution older sensor. Still the yard stick in so many tests, the ARRI Alev sensor in the Amira.

Chris Young

https://vimeo.com/402727266