ETTR and Consistent Noise Levels

[Charles Angus]

I've been thinking more about ETTR, exposure practices in general, and keeping noise levels consistent.

With film, consistent noise levels are easy to maintain, as they are locked in by the film stock / processing technique, so long as you are consistent with your exposure.

With digital, exposed by the ETTR method, I am finding that the the change in noise level from shot to shot can be objectionable. A good example of this was Slumdog Millionaire - noise levels were very inconsistent, even from one shot to the next within a scene. I found it very distracting.

I was wondering how other people felt about ETTR and consistent noise levels. Are people concerned about this at all? Are people using a "scene-ETTR" approach, picking an exposure that should work approximately for the whole scene while keeping information in the shoulder as much as possible? Are people lighting film-style and not using ETTR at all?

I personally think I will go back to exposing film-style for digital, trying to keep a consistent light level and exposure for a whole scene, rather than using ETTR for each shot.

I'd like to get a discussion going on the topic.

 

[David Wyatt]

I don't know for certain but a lot of the really noisy shots in Slumdog appeared as if they'd been cranked up in the grade to see a bit more (especially some of the night-time shots), so I'd assume these shots were originally intended to be darker and moodier but then maybe a decision was made in the grade to "see a little bit more" (as can often happen).

In the past I've been told similar stories by DPs who light their night-time scenes all realistic and moody and then the producer complains that he can't see the lovely location he spent a small fortune on so the levels get cranked up in the grade leading to exaggerated grain/noise :crying:.

IMHO it makes sense to avoid noise wherever possible (expose to the right), but if you want "grain" to add it uniformly and controllably in post [my $0.02]

 

[gbalaji]

I would like to get clarification regarding the Slum Dog Millionaire capturing approach. Whether they capture uncompressed raw format like r3d or cineform codec which writes a RGB 10-bit codec?

Is there any possibility of noise being because of this?

Just out of curiosity to know their workflow and method of approach and try to compare the quality of image consistency?

 

[Dominic Jones]

They captured the SI-2K stuff using Cineform, iirc. Virtually no chance of noise introduction because of that...

As previously mentioned, it's going to have been pulling the images up (either in camera or in post), either due to a change of decision or, possibly, due to low light conditions on the shoot. Could also have been an artistic choice, of course, but you would expect it to be consistent in that case.

As for ETTR practices, I personally tend to favour more of a "film" style of exposure, but I do keep my fill levels higher than I would on film, then crush the lower end of the register in the grade to minimise noise, wherever possible. It's often not a bad idea to go half-way house and expose a little hotter than you might otherwise do for low-con scenes (scene-ETTR as mentioned above, essentially). I am absolutely against the "stills" style of using ETTR to the max for each individual shot, for my 0.02.

Cheers,
Dom.