What are the best way to examine "test" the dynamic range of a camera?

[Abdul Rahman Jamous]

Good day to you awesome people and hope that you are doing FANTASTIC!

 

I do have the opportunity to get access to a rental house that got a lot of camera so I am starting to do some testing and compare between them so I would like to ask:

What are the best way to examine "test" the dynamic range of a camera?

 

[David Mullen ASC]

This is probably the best way but quite expensive:

https://dsclabs.com/product/xyla-21/

But some of what makes these cameras different from each other isn't the range but how things look as they reach the clip point, does the image burn-out in a natural-looking way (i.e. ARRI) or does it look odd (not naming names.)

So you could shoot a face and a grey scale / color chart and do a simple under-normal-over test in 1-stop increments, or whatever increments you want. Technically the ISO setting affects the clip point but once you see the overall range you can decide whether you should shift the ISO.

[Max Field]

An extremely cheap way of doing it which may not lead to scientifically perfect results is to point the camera at a white bath towel and keep it in focus and filling the entire frame. Then put a 1K tungsten or some other very bright source on it so you have tons of light to play with. Put on false color (and make sure no curve LUTs are applied) and reduce the camera exposure to the absolute darkest point that still has a bit slightly registering beyond 0 IRE. Then using triangle of exposure, count the stops you open step-by-step going brighter and brighter and see how many stages you can raise before you see any highlight clipping on the white towel.

However many stops you can count before clipping is your dynamic range (including your initial stop).

Have done this with a few cameras and it's matched up with the professional tests every time.

[Jon O'Brien]

10 hours ago, David Mullen ASC said:

... But some of what makes these cameras different from each other isn't the range but how things look as they reach the clip point, does the image burn-out in a natural-looking way (i.e. ARRI) or does it look odd (not naming names.) ...

Thank you David Mullen ASC, I think that's a good policy of not naming names of the ones that don't look good. However could you possibly name two or three other makes of camera, some models of which brand similarly have a suitably natural image burn-out look at clipped areas in the image? I'm sure others would be interested to know, too. I ask only out of genuine curiosity as you are very knowledgeable. For instance the Canon C300 MkIII camera is said by some cinematographers to have a dynamic range similar to the Arri range due to its dual gain sensor, but of course you may not have used this camera, so I understand if you can't comment on it. Thank you Sir for your sharing your excellent knowledge and experience here at cimematography.com.

[Jon O'Brien]

Typing too fast. Make that cinematography.com. ?

[Jon O'Brien]

Having used the Canon EOS C300 Mark III camera I can say that as far as I can see it has excellent dynamic range and its images can, in post, be made to look pretty good, comparable to film. In my opinion.

[Karim D. Ghantous]

Perhaps I can't answer your question directly But here's how I look at it: ever since the Red Dragon sensor, digital cameras have more DR than film does. Modern cameras like the Alexa 35 and the V-Raptor, separated by perhaps a half a stop, are better again. That includes the Venice II, which is perhaps only a stop behind the other two.

There are publicly available tests out there, but I don't know exactly where to find them.

Even the old Alexa came pretty close to 5219. This test by Zacuto is at least ten years old, IIRC:

Notice how both the Kodak stocks are arguably over-rated, even though their DR here is more than any of the digital cameras. But that's a whole other subject matter.

In any case, with modern tools like scopes and goalposts and false colour, I'm not sure it's as big a deal as it once was. Just IMHO. It's also worth mentioning that people are still using the HD BMPCC. 

[Ray Lenore]

I think David summarized pretty well. Numbers don’t tell everything. Xyla charts are good for scientific comparative purposes between cameras but they don’t tell me how a sensor “see”. For testing cameras I always use models and color/grey charts. Underexposing and overexposing. Then we put those images trough specific pipelines to determine which are the “better-to-us” ways of dealing with those images in post. How you treat them in post can alter quite vastly what and how a sensor see.
After that I have a perfect idea what’s the dynamic range of the sensor for my purposes, taking into account the workflow and the final look needed. To me you can’t separate that for the equation, it’s not only the sensor it’s the whole process what matters.