RED Color Science - Using reference look or itself for colors?

[Robino_J]

When crafting the color science for RED Cameras, are the color decisions based on analogue film stock(s) or is the goal to be as neutral as possible using using metering hardware / charts?

I have heard that the Panavision DXL color science is based on Kodak 5219.

Wondering what is the balance between creative liberty and technical accuracy in the process.

 

[Jake Wilganowski]

It's less filmic pretty and more technically accurate

Alexa is the opposite

DXL is more like Alexa color rather than technically accurate

 

[Jake Wilganowski]

That's why I use a custom LUT to get prettier color response

 

[Phil Holland]

Teehee.

 

[Jake Wilganowski]

Cmon Phil spill it

 

[Marc Wielage]

When crafting the color science for RED Cameras, are the color decisions based on analogue film stock(s) or is the goal to be as neutral as possible using using metering hardware / charts? I have heard that the Panavision DXL color science is based on Kodak 5219. Wondering what is the balance between creative liberty and technical accuracy in the process.

It's fair to say a lot of this boils down to "secret sauce" in all camera invention. Glenn Kennel's Color & Mastering for Digital cinema provides some clues, and David Stumpf's Digital Cinematography provides a lot of other thoughtful background theory.

It's fair to say there's a degree of trial & error as well as math. I think it's not an exact science because there's a lot of interpretation going on, but 10 years after the release of the first Red camera, it's fair to say that 2017's top cameras look more similar than they are different. To me, they still don't look like film, but they're all accurate and can perform well in the right hands.

 

[Stacey Spears]

Sometimes a color science is tuned for objective accuracy and sometimes its tuned for subjectively pleasing results. Some are going for what they call a cinematic look.

When you serve multiple markets, this becomes more difficult. Panavision has a more focused audience with the DXL while Red has a much wider audience with their cameras. When shooting nature footage, for example, you may want a more accurate looking image compared to shooting the next Marvel movie, where you are going for something more fantasy looking.

 

[Hrvoje Simic]

It's less filmic pretty and more technically accurate

Alexa is the opposite

Yes and no.

In some image aspects one is more accurate, in some the other.

 

[Marc Wielage]

Sometimes a color science is tuned for objective accuracy and sometimes its tuned for subjectively pleasing results. Some are going for what they call a cinematic look.

I never know what that means when somebody says "a cinematic look," but I would say a lot of what goes into adding depth and quality to the image is just the lighting, not the color science. There's a huge amount we can do in post, but we can never fix bad lighting problems, especially when it's uneven or an actor is missing an eyelight or a backlight. All those little things add up to "cinematic" to me, not what color science it is.

There are some interesting videos up on Panavision's DXL site that provide more clues as to what they're doing with this camera:

http://dxl.panavision.com/

 

[David Battistella]

I think the RED Color Science is tuned to reproducing accurate colors across the spectrum. With an accurate base, you can then tweak for whatever you need. IE, drama, doc, landscapes, features, etc. The Alexa color science seems to be built more on their film stock and scanner knowledge, delivering an image that DP's who saw and shot a lot of film understand, so it pushes more toward the skintones from film stocks than "totally accurate" color reproduction, more like "pleasing".

The "video" colors people complain about sort of originate out of SONY HD camera's and their specific colors. Then Panasonic has it's own color palette, etc.

But I believe RED aims to be technically accurate because there are many situations where you need to be, more like a the way a DSLR reproduces color, than a Nikon F3 with a roll of specific film in it.

Of course, Graeme is the only person who can completely answer this question.

David

 

[Aaron Lochert]

I think the RED Color Science is tuned to reproducing accurate colors across the spectrum. With an accurate base, you can then tweak for whatever you need. IE, drama, doc, landscapes, features, etc. The Alexa color science seems to be built more on their film stock and scanner knowledge, delivering an image that DP's who saw and shot a lot of film understand, so it pushes more toward the skintones from film stocks than "totally accurate" color reproduction, more like "pleasing".

I wonder if RED would consider providing both pipelines. Something technically accurate and full spectrum and something more Kodak-ish/filmic/cinematic/whateverwordyouwannause.

 

[Robino_J]

I wonder if RED would consider providing both pipelines. Something technically accurate and full spectrum and something more Kodak-ish/filmic/cinematic/whateverwordyouwannause.

I think it would great to have the option as well - using secret sauces is cool but most of the time you need many sauces to make it work for all shooting scenarios. In addition to the new pipeline (which is absolutely fantastic compared to the older color options) it would be cool to be able to dial in a more stylized look out of the box that would work for most shooting scenarios.

 

[Aaron Lochert]

I think it would great to have the option as well - using secret sauces is cool but most of the time you need many sauces to make it work for all shooting scenarios. In addition to the new pipeline (which is absolutely fantastic compared to the older color options) it would be cool to be able to dial in a more stylized look out of the box that would work for most shooting scenarios.

I mostly want them to do it so they could drop the mic on the whole "ARRI has better colors" argument.

 

[Hrvoje Simic]

I think it would great to have the option as well - using secret sauces is cool but most of the time you need many sauces to make it work for all shooting scenarios. In addition to the new pipeline (which is absolutely fantastic compared to the older color options) it would be cool to be able to dial in a more stylized look out of the box that would work for most shooting scenarios.

More stylized look from the start implemented in base color science limits more creative choices as the signal is already specifically predetermined in behavior and conditions the final outcome. If you want digital to seem as stock A (it never will fully, it can just imitate some image aspects) and you target it as a basis of a digital camera's signal, before it reaches the grading suite, less creative choices are possible with that digital signal, the signal which would otherwise be more shapeable.

So let's say you had a camera CS shaped to mimic some image aspects of particular Kodak stock. And you'd prefer for one project going closer towards some Fuji stock. You won't be able to come to that outcome as close as you can with more neutral base signal, since your colorimetry has already been pre-defined.

Digital image acquisition in its higher-end segment allows getting that base image more accurate and referent then ever before, due to a) higher performance electronics, b) high precision quantization, c) high precision and density recording and d) post production-oriented signal mapping, so mimicking specific emulsion colorimetry for a base color response (not final creative outcome) may be considered as a step back in a creative palette, otherwise possible with today's digital imaging technology.

Digital's challenges are DR, color properties coming from the sensor capabilities and consequences of quantization and signal mapping. DR and main color properties are what they are and predetermined by properties of electronics. What is not there cannot be re-created. What is deviating due to sensor's capabilities determines how it can be mapped.

Stuff happening in digital domain with what is quantized can be improved in various stages: 1) base signal mapping, 2) signal optimization prior to color grading & 3) color grading process. Each relying on different approaches and having particular limitations and potential. Where the results of latter step are always pre-determined by the results of preceding steps, but latter steps can achieve some things which preceding ones cannot.

Those many sauces you mention are possible when the digital signal is close to reality as much as possible and natural hues are represented more accurately. If you have some tonalities "choked", accentuated, grouped or distorted, as it is the case with any digital camera with more "stylized" look, it is never going to allow the same creative range as if you have less biased and more healthy signal distribution with all hues more accurate, harmonized and separated.

Base signal mapping is limited by properties of processes used and camera's imaging potential does NOT end at that stage. Some image aspects can be further improved, further in the signal chain and using different principles. This part allows prepping the signal through particular compensations for sensor + quantization + mapping caused signal deviations, which happen with every digital camera.

This is possible due to today's very good sensor technology and base color science combined with very high density data recording, together allowing re-shaping of the signal particulars without visible image deteriorations. When that is achieved, harvesting the potential of every part of high-end digital imaging chain, prior to creative look creation (acquisition + processing + base mapping + base signal optimization) the image does NOT look sterile and unappealing, it is just less biased and predetermined at the factory, allowing more aesthetic outcomes crafted, due to more adequate signal basis.

That is already available today and works on existing Graeme's color science.
If one knows where to look.