F5/55 Raw exposure techniques

[Bruce Schultz]

There have been a few extended threads concerning the proper and recommended exposure techniques for Sony/Arri/Canon S/C Log shots, but I am interested and curious to hear from Red shooters about their Raw exposure experiences and techniques vs the S/C Log exposure techniques - specifically how they might differ or be similar. It's been my experience with Red cameras that it's more of a normal linear exposure process than the logarithmic S/C Log exposure routine is, but I'd love to hear some more opinions about it since it's very germaine to the F5/55 upcoming raw recording experience.

[Mark Kenfield]

Good question Bruce. I'd be keen to hear about this too - would certainly be 'interesting' if you're switching between log exposure and linear exposure whenever you switch from video to raw recording.

[Nate Weaver]

I owned a R1 from 2008 to 2010, it was almost all I shot in that period...

[edit: Bruce, I just read all this and apologize if you understood the real differences, and just wanted to hear about exposure philosophies. As I recall, you're pretty deep into most of this]

The principles are the same. The camera is recording a very wide latitude, to the eye, and viewed without assistance, looks very flat and desaturated. The nomenclature between the cameras is quite different though.

The difference is that by default, the Red has a viewing LUT on at all times, active on all monitoring outputs. If you want to see a view of what's really being recorded, you assign a user switch to toggle "raw" view on and off. They don't even really call it a viewing LUT though...users tend to say things like "I'm monitoring in Redcolor/Redgamma, I checked the raw view to see where the highlights are really falling"

The F3 is the opposite. By default there is no viewing LUT, and you have to go out of your way to turn it on. Like the Red, I have LUT on/off assigned to a user switch. I flip it back and forth to gauge exposure. The difference between LUT on and off on the F3 is pretty minimal to doing the same thing on a Red.

One final difference, the viewing LUT you see on a Red is a 3D LUT, and it's much contrastier and saturated than any of the viewing LUTs for S-Log in the F3. I wouldn't be surprised if another Red shooter joins this conversation and wonders what I'm talking about...Red doesn't talk in LUTs...but from a S-Log point of view, that's exactly what their doing.

As long as you understand that both cameras would like to show you one thing and record another, you're on the right path. The details are quite different, but the underlying principles are the same.

One more thing, if you're shooting at the "native" ISO of a given Red camera, the exposure between the "Redcolor" view and the raw view will be the same. Sony however has built the S-Log LUTs to encourage effectively underexposing and therefore retaining more highlight information. Which they can get away with because the F3 is very clean.

I would have been lost on S-Log if it hadn't been for my time shooting Red and really getting a firm grip on raw workflow.

I've always felt that S-log workflow achieves just as much as raw, but raw seems more magical because it applies LUTs on both camera and post ends automatically without intervention, making it more accessible and netting better images for people just jumping in. Thus the large fan base espousing the benefits of raw. S-log is more "we'll get you there, but first you're going to have to learn how it works from the inside out and do everything yourself"

[Alister Chapman]

The key difference between raw and log is that raw is linear and as a result much more forgiving of over-exposure. The exposure principles are the same. You will be able to use a light meter with the F5 or F55 just as with any other camera or use a grey card etc. I believe the normal viewing LUT will be S-Log2 when shooting raw, but this needs confirmation. Optimum middle grey will be 32% I believe.

The issue with log is that when you over expose you are placing more and more of the scene in the compressed part of the curve and this can have a negative effect on the all important mid range and things like skin tones. With raw, which is almost always linear, if you over expose your mid tones it has no real negative effect on the mid tones unless the over exposure is extreme, but obviously your highlights will clip earlier. So while the basic principles and exposure levels for log and raw will be very similar, raw will be far more forgiving of over exposure and you will have more flexibility when it comes to raising or lowering your mid point to balance noise against over exposure. I've done many F65 demos where we have deliberately over exposed by several stops to show how the mid range still looks very good and I see no reason why the F5/F55 in raw should be any different.

Where the F5/F55 may be a little more interesting will be when you dual record using a conventional gamma or s-Log2 and raw at the same time. In this case you will have to work out which is the most important. Your conventional footage or the raw. In most cases I suspect you would favour the conventional material as the extra raw latitude will mean the raw should always be within a decent exposure window.

[Nate Weaver]

Alister, that sounds like the most important part of all, I forgot all about that, the linear vs. log part of it.

If everything is perfect in front of the lens, it doesn't change how you expose so much, but if you're doing splits...

[Chris Medico]

With the linear RAW where will mid grey fall? Will it be closer to 40-45%?

Under the definition of linear I would expect it would be 50% if you were giving equal exposure range above and below mid but would you still expose down slightly to give more headroom in the highlights?

[Leonard Levy]

Thanks for a great explication of the difference between shooting RED and Slog Alister.

I'm not a Red fan-boy by any means but i must say that on the set being able to view a decent looking "LUT" with red color, red gamma or 709 is a tremendous advantage.
is there an inherent reason why Sony has not provided a similar saturated viewing LUT with SLog like Red provides?. I would have expected something like that with the F5/55 rollout. Does it have something to do with the difference between the linear RAW and the compressed curve in Slog?

[Bruce Schultz]

Super good explanations from Nate and Alister, thanks for your inputs. Yes, I was only inquiring about various RAW exposure techniques but your explanations enlightened a few details I was unclear about. For one, the fact that the Red "lut" is a 3D one explains why the monitoring with their cameras has more color saturation than Sony's LUT's do. Can't figure out why Sony only have 1D LUT's in the F5/55 cameras, but maybe they will add it later.

Second, I have always found middle grey crossover on a logarithmic chart to be at 45% for linear exposure and I don't think that changes with RAW recordings. S-Log2 middle grey is rated by Sony at 32% vs 38% for S-Log1.

[Nate Weaver]

Originally Posted by Leonard Levy

For one, the fact that the Red "lut" is a 3D one explains why the monitoring with their cameras has more color saturation than Sony's LUT's do. Can't figure out why Sony only have 1D LUT's in the F5/55 cameras, but maybe they will add it later.

Since Red doesn't think, talk, or document in LUTs, I suspect Graeme if he read this would say it's not a 3D LUT at all.

But if we want that same functionality, that's what Sony needs to add. Support for 3D LUTs.

At the Sony event tonight, I hope to remember to ask about 3D LUT support in the new cams. I suspect however, the answer will be 1D.

Historically for Sony, they seem to associate comprehensive LUT support with their highest tier products...my PVM OLED won't take a LUT, but the BVM OLED will. So I don't hold out much hope until the next stage of Sony's newfound competitiveness is underway.

[Leonard Levy]

Nate , Did you get a chance to ask about 3D LUT support?

It's curious with all the work they did on these new cameras (and all the SDI outputs) that they would leave out such an obvious feature.

[Nate Weaver]

I did, and the answer is 1D. BUT:

It appears the new camera *might* have a framework to get around it, for your purposes (because by the very definition of raw recording, there has to be another method):

Here's the deal. Both cameras can record 2 "streams", so to speak. An offline and an online. Both to SxS cards, if the codecs chosen are suitable.

The "online" can either be raw via the external recorder, or SLog2 to XAVC/SxS (or later, SR)

The "offline", if also recording raw, is XDCAM. Which can be any camera paint setting you choose. Saturation to +99, etc etc. Not necessarily a 1D LUT. Follow? (by the way, I have yet to see any evidence of "Picture Profiles" with this camera. Which is fine by me)

The question of course then becomes, can you do the same with a dual stream setup to both XDCAM (offline)+XAVC(online)? Or, somehow play the menus so the offline doesn't get recorded at all but the monitoring chain still shows camera paint settings, and the XAVC gets untampered SLog2?

This makes my head spin just to type it, but the short of it is is that 1D LUTs are non-applicable when your online is raw, so there has to be additional options to see a viewable image (i.e., the camera paint menu). The tiny details of how the menus work is going to dictate just how far you can go with this method.

By the way, the reasoning was from the rep on 1D vs. 3D LUT was that in a 1D LUT, the LUT has enough information to precisely define mathematically every possible change in output. In a 3D LUT, the data structure cannot be large enough to define every possible result, so there is interpolation that must be performed. The rep implied that this interpolation would not be acceptable. Reading between the lines, I'm hearing that interpolation is computationally too expensive for this particular camera.

[edit: I don't have any special knowledge or experience with the camera other than 3 minutes with the menus and 2 conversations with reps about LUTs and camera metadata that gets passed to raw files. There's certainly a chance some important details are unknown to me. But just the same, a raw+XDCAM recording scheme would not have LUTs as any part of it, and that fact gives you hope to work the way you wish]