compositing multiple debayers for one image: when is it appropriate/desirable?

[Brett Harrison]

Would anyone be willing to fill me in on the reasons for comping together several debayered r3d images of the same shot. Is this a way to get more dynamic range? Is it a preliminary means of secondary grading? Any clues would be great.

 

[Shiblon Wixom]

Some people like to render the HDRX A and X tracks separately and mix them later in another program, or they are trying to get more range out of the highlights and/or shadows by exposing for each and comping the two together.

 

[Brett Harrison]

Thanks for you reply Shiblon. I should clarify that I am referring to single tracks, i.e. no HDRX, where a compositor/colorist debayers 2 or 3 times with varying settings on the r3d, and combines them.

 

[Angelo Lorenzo]

Your last reply confuses me, are you talking debayer quality like full/half/ or are you talking about different RAW adjustments being stacked. The latter can help in some situations but it doesn't gain dynamic range so much as it does offer a "different flavor" of adjustment rather than slapping curves on it in Resolve. The issue with blending multiple exposures of the same image is that it also increases noise in the final image, sometimes severely; it's something that needs to be looked at and accounted for.

It's common in the Photoshop world where people think stacking different exposure adjustments of the same image leads to more detail, there is no "something for nothing" in either case.

 

[Brett Harrison]

Thanks Lorenzo, yes I am referring to different RAW settings, i.e. ISO/FLUT, kelvin, tint, etc.

So if I decide to choose a lower ASA, qualify a highlights luma range on that, and blend that with the normal ASA exposure (i.e. the base grade), will this bring added detail to the highlights (or conversely, shadows)?

 

[Jeff Kilgroe]

Stacking differently processed images, of the same shot is something people do to overcome shortcomings in their workflows or in the RAW processing tools of their desired application. Where we see it most is people will use REDCINE-X or similarly capable tool to build different exposures and save them out to a non-RAW format that is easier to work with in another application, layer or blend them there. In something like Resolve, Quantel, Speedgrade, Mistika, etc.. software that fully supports R3D RAW, this serves no purpose. There is full access to the RAW data for manipulation.

That said, even in applications that provide full access to the RAW data and a good set of tools to manipulate it, it's not uncommon for the artist or colorist to duplicate the frame, apply changes, blend it back. There's not much more to say here, It's an artistic process.

 

[Axel Mertes]

Hi Jeff,

what we did sometimes was to convert RED R3D RAW to CineForm RAW and use different deBayer algorithms to create e.g. a sharper key from a greenscreen shot. Unfortunately the sharper deBayer don't stop at the women skin, making it undesireable sharp. So we used a smoother deBayer, like the original RED (software only) deBayer to actually fill the image, while the key was pulled from another deBayer.

Thats something that works pretty well on occasion.

Cheers,
Axel

 

[Dermot Shane]

i use mutiple debayer at times, HDR for sure, and greenscreen as noted above.

a few days ago i was gradeing a hair product tvc and the hair was debayered seperatly from the rest of the shot, i would rather do as much as i can in the red sdk, and not trade off OLPF, nr and sharpening settings

there is valid reasons to use this option, but it works best if you already have a toolset that makes sense of the workflow, ie; Pablo/Rio, Flame Premium, DS or Mystika imho

d

 

[paulherrin]

yep, can be useful for tracking, key-pulling, or certain look creation techniques to have an alternatively processed stream.

 

[Marc Wielage]

Would anyone be willing to fill me in on the reasons for comping together several debayered r3d images of the same shot. Is this a way to get more dynamic range? Is it a preliminary means of secondary grading? Any clues would be great.

I think a wiser move is to use an actual color-correction program like Resolve, Speedgrade, or Baselight and then use nodes or layers as a way to create more dynamic range and enhance secondaries.

In particular, pulling parallel nodes from the original balanced image in Resolve is a very powerful feature. For example, this allows having a very cold, desaturated image that still has "normal" skintones with minimal effort; or a darkened foreground with very lush, bright skies -- or vice-versa -- could be done fairly quickly and easily, using keys, windows, multiple nodes, curves, or a combination of each.

I think trying to use RedCineX-Pro for something like this would be a bad idea. And I think this is very doable without multiple image streams per se. Don't forget it's always possible to do simultaneous different color corrections on a single image. Enhancing dynamic range is a very standard part of the job of a colorist.