RED "Behind the Look" on NOBODY: ISO implications

[M Harvey]

Found the recent "Behind the Look" write-up on the use of the Helium by DP Pawel Pogorzelski on the feature film NOBODY very interesting:

https://www.red.com/nobody?utm_source=bm23&utm_medium=email&utm_term=Behind+The+Look+Nobody&utm_content=Behind+The+Look:+Nobody+%26+The+New+RED+Arsenal+Reel&utm_campaign=Nobody+%2B+RED+Arsenal&_bta_tid=14745566075476436801364332164293974615218171691665668663284470502676978671921465732562347044937279067944

One thing that jumped out at me is that in camera testing they chose the look of the Helium rated at ISO 1600, but then shot with a custom LUT that made the image "at least two-thirds to a full stop darker." So their LUT, which they used for lighting as well, essentially brought the image back to the "normal" ISO of 800.

I've only seen the trailer as of yet, but it looks good-- so they did something right. But can anyone explain to me the point of rating at 1600 and then using a guide LUT to light to 800? It apparently wasn't calculated, the DP described the choice in intuitive terms--, but did they actually gain anything technically via this two-step process?

 

[Nick Morrison]

Sounds like their are protecting their highljghts by shooting 1600 ISO (and rating to that - meaning the gaffer and lighting team also use this as a guide). But then the LUT maybe brings down the mids and shadows to add some mood and texture.

Im just guessing here.

 

[Mike P.]

Does the histogram/false colour take into account the LUT? I know the camera uses ISO for their non-raw exposure tool representations, but not sure if the LUT would be represented.

If that's the case (the exposure tools *do not* take the LUT into account), setting to ISO1600 would give a more even over/under for exposure, while the LUT would wrangle noise by crushing it out.

Edit: Yeah, like what Nick said... Expanded a little further, if they were using ISO1600 on light meters/externally measuring exposure, that would do the same thing (protect highs, while giving a more finished look on the LUTed monitors).

Edit 2: Just read it; very weird. He'd disable the LUT to check that shadows were being captured and for a particularly dark scene, he for lit for T4, but shot at t2.8 (which would nullify the ISO bump). Also weird that they shot at 6k on Helium.

 

[Aaron Lochert]

My theory is that 800 --> 1600 via RAW controls was not the same but opposite -1 decrease as going 1600 --> ~800 via the LUT was. Meaning, there was an unintended but desirable change when pushing in RAW and then pulling in the grading suite. Either something with the order of operations or the tool used to pull the image down doing something non-linear to the image.

I'm sure they could have gotten there with an 800 base starting point and a LUT that instead pushed 1/3rd of a stop rather than 1600 base and a LUT that pulls 2/3rds, but that's not how these things happen. In testing, ISO 1600 elicited a certain emotional response in so they locked that in and stuck with that. And then further down the line in pre-pro, the colorist realized that 6K at 1600 ISO was probably getting too thin and he needed to crush the image back down to increase the density a bit. And then during production, since the monitors were showing such a dark image, they lit a little hotter than intended.

So it's a roundabout way to get back to an ~800 base, but that's what happened. Does it make much sense? In my opinion, no, not really. But as long as they were consistent with this workflow, then they got a consistent result in the end. In the same way, one could theoretically set the camera to ISO 100, design a LUT that pushes 3 stops, and then light for 800 because the LUT was so bright.

Many ways to skin a cat.

 

[M Harvey]

I think each of you guys are onto something. The missing information: we know they were lighting to the LUT, and occasionally checking the log signal to make sure they were getting detail in the shadows, but we don't actually know what they were metering to, or if they were doing it consistently. So they may have been in effect protecting the highlights (by backing off the extra stop), or they may not (by just lighting to the monitor-- as @Mike P. mentions re: lighting to T4 and then opening up a half stop).

My theory is that 800 --> 1600 via RAW controls was not the same but opposite -1 decrease as going 1600 --> ~800 via the LUT was. Meaning, there was an unintended but desirable change when pushing in RAW and then pulling in the grading suite. Either something with the order of operations or the tool used to pull the image down doing something non-linear to the image.

This is a good point. While the DP says the LUT pulled down exposure 2/3-1 stop, he doesn't say it did so in a global way-- that is, it probably was a unique contrast curve rather than the type of global reduction you'd get from decreasing your ISO rating to 800/1000. It might be that what they responded to in the ISO 1600 test had a lot to do with the quality of the image at a particular place in the curve--so they liked the new middle grey--, and that their custom LUT crushed stuff below that for the kind of rich blacks they liked from Korean action movies.

Part of the reason I was interested in the method to the seeming madness is that this DP also shot MIDSOMMAR, which has an interesting approach to white. (In that it is a bright, daylight horror film where the top end of the contrast curve is associated with horror, rather than the shadows.) So he's clearly someone who thinks about his contrast curve, even if he maybe doesn't approach it the way an engineer would.