REDCINE-X Saturation ???

[Rami Mustakim]

Hi everyone,

4 little things before I start :

- First of all I'm a huge RED's fan and I'm just looking for an answer here.
- Second, I'm not a colorist so my statement may be dumb. If it is I'm sorry.
- Third I used Mercedes M-X Reel R3D sample on which I applied Jim's "RCX 10" latest Look in REDCINE-X only because I want to avoid being said that my RED one camera doesn't work well or I don't know how to shoot or I don't know how to use REDCINE-X.
- And finally my english is not so good so try to read between the lines ;-)

Ok so let's start :

We all agree on the fact that REDCINE-X is a powerful tool as Jim said. I really do find it is !
But I can never color saturate my R3D footage as I expect to in REDCINE-X. Why?
I always leave Saturation level to 1 in REDCINE-X and add saturation to my footage after (in Color or FCP for example).
I do get better results that way.

Here is a good example of my comparison test between saturation in REDCINE-X and CS5 Photoshop :
I took the same frame with 3 different saturation levels (default 1, 2 and extreme 4) Then I took the default 1 and added saturation in Photoshop (set to 50 to match REDCINE-X level 2, then added 50 more to match REDCINE-X level 4)
What do I see :
Results in Photoshop look much better to me. The sky and the hill's colors seems more "natural". Also green color is greener than in REDCINE-X.
I made a 100% crop of the car light in the frame so you can see how the blue channel is overexposed when saturated in REDCINE-X.

So am I missing something?
Shouldn't I get better results in REDCINE-X working directly with my RAW file ?
I'd like to learn more about it.

Thanks.

 

[Rami Mustakim]

I also noticed that it is much easier to handle noise if your footage is not very color saturated.
That would mean it's better not using saturation level to much in REDCINE-X if you want to handle noise correctly. And leave saturation operation for post REDCINE-X.
Again I did a little test using another M-X Reel's sample whit dark areas to show you what I mean.

 

[M Most]

Using Redcolor rather than Redcolor2 will get you more saturation as part of the debayer process. Many colorists (myself and others) have settled on a combination of RedlogFilm and Redcolor (not Redcolor2) for just that reason.

 

[Michael Lindsay]

Hi Mike

We have often globally increasedy saturation within red's post tools at the pull stage (redline, rocketcine now redcineX). We do this for, what I believe, is the very same reason you suggest using red colour 1 over 2. It is easier to take out than put in and the cleanest place to saturate is within rthe debayering phase.

My question is: have you tested the differences between a different colour matrix decode and a simple saturation increase within redtools (either RCX or the sdk)? Is the colour matrix decode cleaner than say pushing saturation by the equivalent amount?

Thanks

Michael L

Ps sorry to hijack Rami but your post worried me as we have been working of a few assumptions and probably should have tested more.

 

[M Most]

My question is: have you tested the differences between a different colour matrix decode and a simple saturation increase within redtools (either RCX or the sdk)? Is the colour matrix decode cleaner than say pushing saturation by the equivalent amount?

I would say yes, based on experience. Graeme and Deanan would probably say no based on numbers. You should form your own opinion based on tests using your own material.

 

[Michael Lindsay]

Thanks for the responce.

Unfortuantly for me I'm too aware of the merits of rigorous testing (rather than fiddleing about with fast reductive thinking and quick conclusions) and will pop that onto the long list of nonurgent thngs to look at. We have somthing that works well at the moment so we are not under any real pressure.

I had just thought you might have been the kind of guy that took the time to conduct a rigorous test and new for definite (and didn't mind shareing). After all it is not like the results of lens test (which are very dificult to share and probably need to be conducted fresh again by each interested party).

Will post back here when we do look at it...

thanks again

Michael L

 

[Deanan]

I would say yes, based on experience. Graeme and Deanan would probably say no based on numbers. You should form your own opinion based on tests using your own material.

I would say I don't understand the question...

 

[Michael Lindsay]

Hi Deanan

Basically if you want to add a little extra saturation within RCX so your grading app dosn't need to add saturation (When working from 10bit 422 or 444 files rather then r3d). Would it be cleaner to process (when doing a pull in RCX) with Redcolor 1 and saturation at defualt or to process with Redcolor2 but with a saturation lift(with RCX)??

thank you

Michael

 

[Deanan]

Hi Deanan

Basically if you want to add a little extra saturation within RCX so your grading app dosn't need to add saturation (When working from 10bit 422 or 444 files rather then r3d). Would it be cleaner to process (when doing a pull in RCX) with Redcolor 1 and saturation at defualt or to process with Redcolor2 but with a saturation lift(with RCX)??

thank you

Michael

If you're going to a 10bit format, there will be some difference so do it before.
For a format like 16bit EXR, it matters less.

Additionally, there is a significant difference as to where the saturation is done.
In RCX, saturation is done in linear (light) space, where as if you're output
RLF, RG2, etc. and then adjust saturation, you're doing it a different gamma space
to the results will be different (unless you linearize first).

 

[M Most]

I would say I don't understand the question...

In previous conversations we've talked about the efficiency of bit usage, and I recall you (or maybe it was Graeme, probably was) pointing out that increasing saturation uses more of the "bit budget" (my term, not either of yours) and limits the amount of detail information that can be included by a small amount. I don't recall Graeme claiming that was significant, but at least in theory, he seemed to be saying that less is better in terms of retaining information as opposed to noise.

 

[Dan Hudgins]

Gamma 1.0 vs. Gamma 2.2

Gamma 1.0 vs. Gamma 2.2

You can get some shifts between the chroma and luma depending on if the image is at Gamma 1.0 or Gamma 2.2 or 1/1.7 for Cineon.

You can try adjusting the gamma before you make color saturation adjustments, and then use the reciprocal gamma after color saturation.

In my DI system I use a 9 value color matrix so that its easier to adjust the color masking to bring out various colors and not to over saturate and clip other colors, you can look at the chroma vectorgram to see the maximum color saturation for all the colors in the shot as I have adjusted the chroma vector gram to be a hexagon with red, yellow, green, cyan, blue, and magenta about equal distant from the equal white point. You don't want to "flat top" the chroma.

Its also possible to apply an S-curve to the chroma to avoid clipping, I don't have that as a separate feature but if you split the luma and chroma you can process them as you like then put them back together with various software. In general its best not to have the luma and chroma processed with separate curves as that can lead to "detachment" and excess chroma in the shadows.

In REDCINE-X you see the default is to have a slight lowering of the luma in the lower tones, if they do not also lower the chroma, which seems to the the case from the examples, then you get the chroma being brighter than the luma in the shadow areas. To avoid that put the luma curve in REDCINE-X back to straight line and use a program with 9 value color matrix to adjust the saturation as the mixing will adjust the luma as well as the chroma.

You can also try adjusting the chroma at gamma 2.2, gamma 1.0, and gamma 0.45, to see how that changes the mixing.

 

[Rami Mustakim]

Ok…

When I started this thread I thought maybe someone would help me understand why I can't achieve the color saturation I want in REDCINE-X.
As I thought it's better working on a RAW file.

All began as I was trying to achieve a good colorful skin tone within REDCINE-X. Problem is when I increase saturation level in REDCINE-X, skin turns orange, lips are very pale. When I do the same within Photoshop skin looks more natural and lips are red. So I started to look at RGB channels to see what the difference was.

What I realized is that no matter what your saturation level is in REDCINE-X Green channel won't move. As opposed to Photoshop for example where all RGB channels move when you increase saturation. Again I'm not a colorist and I'm just looking for an answer here. Could someone help me with this ?

So here is an example of what I'm talking about.
I took several saturation levels (0.5, 1, 2, 3, 4) of the same frame in REDCINE-X and guess what, green channel is the same all way through.

I took the default saturation 1 level frame to Photoshop CS5 and applied saturation to match REDCINE-X and then all RGB look different according to the level of saturation.

Ok it's not obvious but you can see green channel is not the same all way through.

 

[Jannard]

Checking... which version of RCX?

Jim

 

[Rami Mustakim]

REDCINE-X_build_470 the last one.
But I think it has always been the same for me.
Fell free to tell me if I do something wrong.

 

[Jannard]

We'll let Graeme wake up, get some coffee and answer this one. He is in Eastern Canada now where it is 2:15am. He likes detective work.

Jim

REDCINE-X_build_470 the last one.
But I think it has always been the same for me.
Fell free to tell me if I do something wrong.

 

[Rami Mustakim]

Jim, this Green channel "thing" only seems to appear within Camera RGB Color Space.
As soon as I go with Redcolor2 Color Space the green channel is changing normally with level saturation.
But then with Redcolor2 Color Space there comes another problem, the Blue channel over saturates when I increase Saturation level in REDCINE-X like it's burnt.

For Example here is what I see in REDCINE-X with saturation set to 2

And then this is what I see when I take the same frame from REDCINE-X with saturation level set to 1 and I increase saturation in Photohsop to match REDCINE-X level 2. Red and Green channels look almost the same but the Blue channel is lot less burnt. Colors seem more natural to me with Photoshop saturation. Look at the guitar behind his shoulder. It seems very yellow in REDCINE-X and more natural in Photoshop.

I start thinking it's better leaving saturation level to 1 in REDCINE-X and saturate in post software.
Question is why ? Why is it easier to achieve color saturation outside REDCINE-X which is the tool for RAW R3D footage ?
Wouldn't it be better if REDCINE-X could handle saturation more "friendly"?

 

[Graeme Nattress]

I'm going to do a bit of code tweaking.

Graeme

 

[Christopher Probst]

This thread is interesting to me... I'm currently in the process of trying to make sense of ALL of the different workflow scenarios, parameters, consequences and pitfalls for all of the various post tools and digital camera systems out there. I have had way too many instances on commercials and to a lesser degree on music videos where the post work is completely handled by agency or client and I have no input on the decisions made.

I would like to get as much understanding on what options are being used and try to come up with a few recommendations based on the parameters at hand... (such as, are they coloring on a system that can handle .r3d or will the be coloring off transcodes?) I fear one arbitrary misclick of a setting can drastically influence the quality of the colored outcome. And if I have no presence in the matter, anything goes.

As with all of my obsessions, if I delve in deep enough, then I feel I have something to offer up, but if I don't feel completely solid in my understanding, I'm screwed.

Here is where I'd love to sit down and talk with Graeme, Mike Most Michael Cioni and the link and get into the nitty and the gritty.

Great topic.

 

[Jannard]

Our color science has improved significantly since the beginning. We are always looking for a problem to solve... the only way to get better. The science is now incredibly good but can always be tweaked to make it better. Feedback is good. People try stuff we don't test for and we are all better for it.

That is the purpose of this forum... (well, one of them).

Jim

 

[Luis Ortiz]

Excellent thread. Thanks, Red, for chiming in so quickly. Can't wait to hear back from Graeme on this topic.