Dynamic Range of the 5D

[Kharak]

So I've doing some thinking and reading.

5D mk iii has a dynamic range of 11.8. (This is h264 i believe).
I know Alex is working on some sub 200 iso's where the dynamic range on the mk ii and iii is increased with some decimals. I believe I read that the mk ii has more dynamic range than the mk iii at these 90 iso's or something. Think the mk ii hit 12.4 and mk iii, 12.1. Dont quote me on these numbers, read it somewhere deep in a developers thread.

This aint the pooint about this thread either, just mentioning them.

When I record Raw on my mk iii and ettr and I pull down the highlights, add cineon convert and all the tricks of the trade to pull as much information out of the highlights as possible and raise detail in shadows. Am I still only within those 11.8 of dynamic range?

Is post work on raw files included in these DR tests?

[jimmyD30]

Dynamic range is rated at the camera without regard to post. The ability to process dual-ISO/HDR with greater DR from post processing routines isn't included in these test.

[a1ex]

When you use ETTR, you don't increase your dynamic range, you simply use the available range better.

The DR on these tests was per-pixel and not very exact (it estimated the noise levels from the optical black areas). Recently I've found a more accurate method that estimates DR from the noise curve, but the only cameras tested with this method were 5D3, 6D and 50D.

5D3 results, movie and photo mode, stock firmware

6D results (no graph yet)

50D results (older, without DR estimation)

I'm still waiting for test results for all other cameras. The testing procedure is here, and there's no need to compile anything.

[Kharak]

I asked this to get a little clarification. Working with RAW In post, I sense that the DR is so much more than 11.8. I mean when I really get in to highlight recovery with Acr+cineon and what not , I manage to pull out detail from the Sun itself (not actually), but exposing for indoor without any ND filtration on windows and still being able to pull an incredible amount of information from the outside, is astonishing. With some good masking and cc the shots look surreal.

I also own the mk ii and would like to help with your tests, but I am out sailing and my mk ii is at home. I will most likely return home in a week or two... Hope for one.

What is the goal of the test you are conducting?

[a1ex]

When you take a picture of the sky, the red channel is going to be a bit darker than the other two (around 1 or 2 stops), and a good highlight recovery algorithm will get quite a bit of extra detail.

Canon does not do any kind of highlight recovery in camera, so where the JPEG preview is white (and the regular zebras indicate clipping), there's usually some more detail captured. Even if you have only one channel not clipped, it's enough for recovering details in the sky or in the clouds.

If you use RAW zebras, only the black areas are totally clipped. However, these are slow in LiveView, so ML defaults to fast (approximate) zebras. If you rely on highlight recovery, you may want to change this.

With the tests, I want to tell exactly how much DR improvement resulted from the tweaks, and the first step is to have an accurate method for measuring it, and to know the figures with stock firmware. Next step will be to check the results with the tweaks applied, and compare them.