How to send client/director monitor feed that approximates final output when exposing to the right?

[Johanan Pandone]

Guy's I'm not sure if this is more of a DIT question, but it has to do with exposure so I thought I might get some input from the DP's

I'm using Canon DSLR's a lot shooting 14-bit raw, and I am able to get much cleaner chadows by exposing to the right and pulling the exposure down in post. I can load custom picture profiles in the camera that do not affect the recorded raw image (basically an LUT in camera) as well as regular .cube LUT's on my monitor and the downstream feed. I'm not talking about shooting log and previewing with a rec.709 LUT, I 'm talking about exposing to the right (usually the preview looks overexposed by 2-3 stops) and using an LUT or a picture profile to pull the exposure back down for the preview monitor so the client and director can see something that looks like it's exposed normally (approximating a 1-2stop pull in post) Does anyone have any tips, any recommended LUT's, or practices when using this technique? Additionally, how do you feel about exposing to the right in general? I know it's a controversial subject, but in my own tests I'm getting much better shadows when I do this. The 14-bit MLV files produced by magic lantern have a ton of headroom in the highlights and after converting to CDNG for resolve, highlight recovery is actually able to bring back 2 stops of highlight detail that you can't even see when you're shooting, but the shadows break up very easily in color correction if they are exposed properly on the day. Thanks in advance for any opinions or advice

[Frank Wylie]

Not an expert on this but this video might give you some ideas...

https://www.premiumbeat.com/blog/how-to-use-the-lut-box/

 

[David Mullen ASC]

While ETTR probably makes sense for what you are doing with a DSLR recording video, in general, it only makes sense in cinematography as a general concept or for one-shot sequences or shooting VFX plates, establishing shots, etc. 

Continuity of look, mood, contrast, noise, etc. between coverage so that they intercut with the master is important in cinematography.  All the shots within a scene ideally would be exposed so that they could be intercut with minimal color-correction adjustments in post, so if a character is standing in a darker part of the set in the wide shot, then they are similarly darker in their coverage with minimal cheating of exposure -- they won't get noisier by being darker because you won't be lifting the signal in post to make it brighter. So exposing each set-up differently based on ETTR principles creates a lot of visual discontinuity that has to be fixed with color-correction - making the image on set inaccurate and requiring color-corrected dailies.

Now if what you are actually talking about is a consistent extra amount of exposure on everything, sure a monitor LUT could correct for that... but then, also simply lowering the ISO to compensate would do that same thing, either way, you're giving the sensor more light and getting less noise in the signal. Of course, the difference between the two approaches is the recording -- you've lowered the noise by giving the image more exposure either way, but if you correct it with a viewing LUT, then your recording is also overexposed and maybe you prefer that in post because of your recording codec.

[Johanan Pandone]

David,

To be clear I would expose every shot 2 stops over with the intention to pull them all back down by an equal amount in post. Similar to rating film slower for a cleaner negative. The MLV format is not a codec it's 14-bit uncompressed raw bayer data recorded directly to disk and it's debayered to 16bit CDNG after dumping the files to the computer. There is no native ISO on a canon DSLR, well, there is, I guess we could say it's 100 iso, but increasing the iso in camera applies analog gain to the signal and it gets baked into the recorded file. I have worked with cinema cameras that will let you rate the camera at different ISO but the native ISO is what gets baked in, I don't have that luxury with a hacked DSLR, but i think I have foud a solution in case anyone else runs into the same situation:

Magic lantern has a digital gain feature that will allow you to push/pull from the analog gain (full stops in canon menu) when recording to h.264 8-bit mov. the digital gain is applied before the encoder, but when shooting MLV Raw the digital gain only affects the preview and the recorded raw file retains the analog gain set from canon menu. applying negative gain in the magic lantern menu allows me to pull the preview back down to normal exposure without affecting the recorded file!