Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Arri Alexa color and Redcode

Bruce Schultz

Bruce Schultz

Well-known member
Joined
Jul 27, 2010
Messages
161
Reaction score
0
Points
16
Location
Malibu, CA
Art Adams has written a brilliantly insightful and ultimately intriguing article about how Arri Alexa cameras handle color and why they are the generally recognized as having the most “filmic” look.


“Alexa limits color saturation in highlights: when the luma portion of the video signal passes 40% or so (middle gray) the color saturation freezes in place: color only gets brighter, not more saturated. This is a big part of why film looks like film, and colorists have known for years that one of the tricks to making video “filmic” is to desaturate highlights.”


(full article here: http://www.dvinfo.net/article/acquisition/sonyxdcam/sony_sgamut_vs_sgamut3.html )


Art goes on to describe how he posed this to Sony engineers and asked if they could replicate this process in a gamma profile, and apparently they scratched their heads and came up with 4 gamma profiles that are currently available for the Sony F55/5 cameras.


It got me to thinking about whether this is possible to do in Redcode. In another article he shows how he hand-built essentially a macro in Davinci Resolve to mimic this saturation freeze for F55 footage, apparently with good results.


(full article here: http://www.dvinfo.net/article/post/making-the-sony-f55-look-filmic-with-resolve-9.html )


The question/discussion I would love to have here is to see if this would be something possible to either implement in RedCine X, Redcode itself, or as a plug in for Resolve, Scratch or other color system software.


I have always noticed a profound difference in Arri Alexa footage color matrix processing and now with Art’s excellent and detailed articles I know why that is so.
 
I believe Alexa also has a dual pixel readout where proper gain is mapped on pixel level to achieve optimized exposure. Much like WDR sensors.
anyway that was the initial rumor upon Alexa release.
 
Just to clarify why I posted this, I am an Epic X owner and a DP/DIT so anything that I can do to promote Red Epic over Arri is a good thing in my world. One of the arguments I hear a lot about is the "filmic" look of Arri vs Red footage, so I am trying to suss out if there is a way to neutralize this particular viewpoint amongst producers and directors.
 
Bruce Schultz said:
It got me to thinking about whether this is possible to do in Redcode.
Click to expand...

I have not read the article(s) yet, but I don't think you would need to touch the redcode itself. Instead this could most likely be done as part of the color science during debayer. e.g. DRAGONcolor.
 
Desaturated highlights are a natural function of how the application of a typical gamma curve works. There's no magic there. Where it becomes hard is if you want to do the opposite and retain full saturation on really bright colours. A simple analysis of an RGB cube will show you why this is the case. Imagine the RGB cube sitting on the black corner, with the white corner vertically above the black. Where are the most saturated colours now? Well, they're half-way up along the curve and as you get brighter beyond that half-way point, saturation is naturally curtailed as the brightest colour is the completely un-saturated white.

Graeme
 
Graeme Nattress said:
Desaturated highlights are a natural function of how the application of a typical gamma curve works. There's no magic there. Where it becomes hard is if you want to do the opposite and retain full saturation on really bright colours. A simple analysis of an RGB cube will show you why this is the case. Imagine the RGB cube sitting on the black corner, with the white corner vertically above the black. Where are the most saturated colours now? Well, they're half-way up along the curve and as you get brighter beyond that half-way point, saturation is naturally curtailed as the brightest colour is the completely un-saturated white.

Graeme
Click to expand...

Presumably preserving color saturation in the shadows is also difficult for the exact same reason - white and black have no saturation.
 
Black indeed has no saturation which again limits the gamut. However I don't find this as perceptually obvious.

Graeme
 
Yeah, but the article implies that saturation is limited/remains the same from 35% onward (in Alexa and in Film). Not just that the highlights (which implies the top 33% of the image -- 66.7% onward) are being desaturated.

When I desaturate the highlights, I'm usually doing it to the top 5% and only in order to balance out wonkiness or potentially to stretch contrast in that top 5% to get more even detail out of the least overexposed channel (kind of like DRX on MX footage, but without clipping the non-clipped channel(s) to get even colour.)
 
Mike, if you look at what a gamma or log curve is doing on typical linear light image data with decent dynamic range, you'll see how the brighter the colour, the more it gets squashed down to fit into the display range. Squashing a colour down like this to fit necessarily lowers saturation (just as an increase of contrast increases saturation by making the colour components spread further apart in value). Again, nothing magical going on, just the normal application of a simple LUT.

Graeme
 
As I understand it the one very evident advantage of the ARRI approach is that the over/ exposure can be easily corected byt pulling down the brightness and keeping the saturation intact.
That's why, unlike other digital systems the image doesn't fall apart with more that 1.5 stop of over exposure. Are you saying that the same can be done with RED images simply by applying a curve to them? Even though the raw image acquired is linear(both saturation and brightness equally boosted with over exposure) ? Are there any tests that show that?
I may be completely wrong so please, correct me...
 
From what I can see there is no "ARRI approach". The observed appearance is just what you'd expect with a standard LUT. Sony try to be really clever and make their particular super-saturated TV look that's popular with their cultural base.

Graeme
 
I gotcha... I'm just trying to figure out if keeping saturation the same strength through the top two-thirds of the image (via colour correction, since I don't have a LUT that does this specifically) is worthwhile or not. Art's examples (a DuMonde chart) are great for comparison purposes, but don't exactly give real-world results. And again, without a LUT that does this, I have to do it by hand. So while a gamma/log curve may lower saturation, we have no way of knowing by how much (or how it compares to the saturation level at 35% brightness) without doing it in a suite or with scopes attached.

All I'm saying is, this is the first time I've heard of this, and clearly not all gamma curves do the same things (and I happen to think Alexa looks incredibly film-like out of the box, so in my opinion, it's doing something(s) right -- maybe this is one of those things). After all, we're shooting raw, experimenting with this kind of metadata is part of the fun. I'm in no way implying that r3ds can't be processed this way (quite the contrary actually). I'd just like to get other peoples thoughts on the validity of Art's claims...

For example, does film actually retain a similar saturation level above 35% brightness? If a gamma/log curve inherently desaturates the highs, shouldn't Art be *adding* saturation to keep it the same above 35% brightness?
 
Graeme Nattress said:
Sony try to be really clever and make their particular super-saturated TV look that's popular with their cultural base.
Click to expand...

I find this comment interesting and in my experience sadly true. Been looking at the quality of images from cell phones recently and learning that companies are changing the saturation and color based on region. A given cell phone will produce an image more red in Asia than in the US for example. Makes me want to hit my head against the wall at times. :-) Some are optimizing the pictures so they look good on the handset they were taken with, which makes them look over processed on a handset with a good display.
 
Mike P. said:
(and I happen to think Alexa looks incredibly film-like out of the box, so in my opinion, it's doing something(s) right -- maybe this is one of those things). After all, we're shooting raw, experimenting with this kind of metadata is part of the fun. I'm in no way implying that r3ds can't be processed this way (quite the contrary actually). I'd just like to get other peoples thoughts on the validity of Art's claims...

For example, does film actually retain a similar saturation level above 35% brightness? If a gamma/log curve inherently desaturates the highs, shouldn't Art be *adding* saturation to keep it the same above 35% brightness?
Click to expand...

Exactly. Like you I am trying to discern whether or how R3D files can accomplish this chroma desaturation process, or as Graeme suggests, it isn't really necessary.
 
You must log in or register to reply here.
Back
Top