RedWideGamutRGB: discuss!

[Michael Tiemann]

There are a few different threads that started off as Helium or 8K VV threads, but which morphed into the topic of color science. Awesomely, the RAW R3D format means that even those of us with conventional WEAPONs (and DRAGONs) get the new color science. Yay!

I submitted a badly behaved R3D frame to RED some months ago. When viewed in linear light, it is clear that nothing is clipping. But when converted to DC2/RG4, multiple bizarre color artifacts emerged. Architectural lights (meaning colored glass that let light pass through, not lights aimed at architectural elements) that were blue started fringing purple in the hot spots. Changing the color temperature from warm to cold reduced that, but then architectural lights that were purple would fringe blue. And the colors were way, way too saturated.

With RedWideGamutRGB, the colors stay in the box, but they are very much desaturated from the real-life vibrant colors of the diamond lights. If you know what light looks like when it goes through cobalt-blue glass, that is the color blue we are aiming for. It's blue. It never turns purple. In any case, I think RedWideGamutRGB will be a very helpful starting place for me to build a LUT that ultimately handles my (architectural) glass.

Here is the R3D file, along with some JPG stills showing how things change across temperature (with DC2) compared to RedWideGamutRGB:

https://www.dropbox.com/sh/id1lhjnss1ot8uw/AAAZh3JrW_lGmQrhSw4Awb_6a?dl=0

 

[Graeme Nattress]

Thanks for investigating RWG Michael!

Graeme

 

[David Battistella]

I'm looking at some differences in LUT performance with the RWG selected. Interesting.

David

 

[Graeme Nattress]

LUTs should perform very well on RWG data. I've done extensive testing here with that use case and it's working very well.

Graeme

 

[David Battistella]

I'm noticing less of a need to dail back saturation and just an overall crispness to the layers within the image

David

 

[Graeme Nattress]

That's great feedback David. Thanks!

Graeme

 

[David Battistella]

Here is a quick and dirty using RCX to import and apply the LUT. 5300K seemed to be the sweet spot for a onelight. Obviously lots more could be done.

DC/RG3

LUT applied with RWG/Reglogfilm

 

[Graeme Nattress]

That's looking sweet David!

Graeme

 

[David Battistella]

Here it is with same settings as a night look. 2000ISO

 

[Michael Tiemann]

Here it is with same settings as a night look. 2000ISO

ZOMG!

 

[Stacey Spears]

Nice images David!

 

[Mario Gonsalves]

Love the colors on the second image in the night time shots David.

 

[M Most]

With RedWideGamutRGB, the colors stay in the box, but they are very much desaturated from the real-life vibrant colors of the diamond lights.

Well, yes. That is the nature of a wide gamut color space viewed on a display with a much narrower gamut. "Proper" saturation would be achieved via use of a color matrix in the processing path. In Arri wide gamut color pipelines, this is often implemented as part of the LUT. Up to this point in time, Red has always done this as part of the debayer process, offering different matrices for the purpose (Redcolor 1-3, Dragoncolor, etc.), allowing a RedlogFilm output to be "normalized" by a 1D LUT and achieve "normal" saturation. With the use of the wide gamut color space, you're basically back to what in much earlier Red days was referred to as "Camera RGB," albeit with more sophistication and much better accuracy. But it also means that to yield "normal" saturation is going to require a color matrix somewhere in the post path.

 

[Michael Tiemann]

Well, yes. That is the nature of a wide gamut color space viewed on a display with a much narrower gamut. "Proper" saturation would be achieved via use of a color matrix in the processing path. In Arri wide gamut color pipelines, this is often implemented as part of the LUT. Up to this point in time, Red has always done this as part of the debayer process, offering different matrices for the purpose (Redcolor 1-3, Dragoncolor, etc.), allowing a RedlogFilm output to be "normalized" by a 1D LUT and achieve "normal" saturation. With the use of the wide gamut color space, you're basically back to what in much earlier Red days was referred to as "Camera RGB," albeit with more sophistication and much better accuracy. But it also means that to yield "normal" saturation is going to require a color matrix somewhere in the post path.

Very good point!

 

[Graeme Nattress]

Yes Mike - RWG is quite similar to CameraRGB, but it's standardized so that it is one colour space rather than an almost infinite array of spaces that differed upon the nuances of the camera, sensor and how the raw development was set up. RWG gives plenty of the benefit of CameraRGB and some new abilities, and as all the cameras get normalized into it, workflow can be greatly simplified.

Graeme

 

[Daniel John]

Forgive me if it's obvious but I'm not seeing the option for RWG in the drop-down menu in RCX. I went into preferences and checked "show advanced controls" and still not seeing it.

I'm on build 41.0.0

Thanks!
Daniel

 

[Graeme Nattress]

You need to get the beta for version 42, and then make sure advanced controls are turned on.

Graeme

 

[Daniel John]

Ah, gotcha! Didn't realize it was on the Beta version.

Thanks for the reply Graeme. Much appreciated.

 

[Stacey Spears]

"Proper" saturation would be achieved via use of a color matrix in the processing path.

One of Poynton's FXPHD talks goes into a lot of detail on this very subject. He shows an example of a camera's native gamut and how you need to apply a 3x3 matrix for it to be really useful.

 

[Sabovic Adis]

Is this normally saturated, oversaturated, undersaturated?

In RCXP beta 42:
- REDWideGamutRGB
- Log3G12
- Kelvin 5600
- Tint 0
- ISO 800...

...everything default.

"Full CTB" added, and graded for brightness in NLE...by eye-ballin. :biggrin:

Second shoot with A.D.D.