RED - examples of natural, "filmic" color?

[M Most]

Excellent examples, Florian. A good illustration of some of the things colorists and others have been talking about, but a bit more scientific.

 

[Jake Bastian]

Excellent examples, Florian. A good illustration of some of the things colorists and others have been talking about, but a bit more scientific.

Except not really because it's not comparing Rec. 709 to Rec. 709. That's Redcolor2/Redgamma2 to Rec.709 - better would be to apply a LUT to RedLogFilm/Redcolor2 so you can really compare apples to apples. The bottom two pics are interesting to show what comes out of each camera RAW though.

 

[Stephen Matthews]

I guess this shows how subjective things like skin tones are as I like them in the epic shot much better than the Alexa shot. Thing is to really know if it's accurate I'd have to be looking at the guys hand in real life as well as on a properly calibrated monitor to see if that is actually what his skin tone looked like. So to me this whole proper skin tones to reality is very subjective, especially considering to me a proper comparison is Alexa out of the box and Epic after a first pass in redcinex(though could be seen as not fair to some it is to me) I'll take flexibility in my grade over a good look out of the camera, two different mentalities to me. And that's great to each their own, some want digital to act just like film, I just want the most flexibility and universal camera I can get that is not restricted by how things have been done but is writing how things will be done.

 

[RivaiC]

Hi Jake

It is not going to make any closer to Alexa in terms of skintone. The color dictates by the sensor, filter usage, method of debayering and color space.

Alexa recording needs to have matrix apply, any transformation need to have LUT. But RED footage is different, it happens during the debayering process, and you can select it on your own preferences. Which what you are trying to suggest to have LUT and have it transform outside RED realm.
In Davinci, it exists 3D logC to rec709, but let's take a look at Graeme opinion in the previous post


-----
Graeme's post :
Just think back to earlier in the thread where Deanan mentions how Sony sets their red response filter characteristics wide to allow a lot of red through, Arri set their's narrow to let through less far red and RED set their's inbetween. If all three camera types record some deep far red colour - the Arri might not see it at all, the Sony sees it all and RED sees it inbetween the two. Now there are very valid reasons for why each camera manufacturer sets their response where it's at, but no LUT, IDT or matrix on this planet will make the Arri see the red it's native spectral response cannot see to match the output of the Sony which can see it. That is why, in one particular example, why you'll never get the same spectral response via ACES or any other colour space. Of course, three tristimulus values of a colour in a colour space will only ever approximate the actual spectral response of a colour under a real light with a real spectrum anyway....
-----

It involves other thing rather than mere color space. Characteristic of the sensor comes into play too...

my 2 cent

PS. Florian, if you can help to upload the data for testing purpose both RED and Alexa original footage, it will be great. I can help you to host the files if needed.

Rivai

 

[Florian Stadler]

I'm not going to publish my RAW files. They are my own tests used to calibrate my own imaging systems. Everyone that is serious about image calibration needs to conduct their own tests. All I wanted to do is to contribute some actual comparable examples to look at in this discussion. I set the LUT in the upper example to match each manufacturers recommended view setting as that's what most people get to see as what the camera "looks like". On the lower tests I set the LUT/debayer setting to what can achieve the maximum information out of the "digital negative" - Redcolor2/RedlogFilm/320ASA and LogC.
The point of the illustration was that both systems resolve the color spectrum quite beautifully when exposed normally. When you push things around to the limits (I overexposed the charts below 4 stops!) you start seeing some significant differences in color rendition.

 

[Alexander Ibrahim]

There are some very very minor differences ... because cameras are different.

What a lot of people seem to be saying is that Red doesn't look they way they want it to ... that it has a signature look. Well ... so does Alexa.

Much earlier in this thread I pointed out that most cinematographers experienced with film can tell Kodak from Fuji readily, many can tell which "brand" stock (Vision 3 versus Vision 2 etc.) some claim to be able to identify particular stocks (5219 vs 7219, both vision 3)

That's the "gold standard" film. If film has a signature, then of course different companies digital cameras will have a signature ... We've known that for years with Sony versus Panasonic versus whatever else.

The thing is that I wouldn't count on being able to make an Alexa match a Red all the time or vice versa. You can get close enough so audiences won't care, but a true match? Can you get a true match between A cam on 5219 and B cam on some Fuji stock? Nope.

Part of what irritates me is that both Alexa and Epic represent images beautifully ... no seriously. My only gripes with Alexa are resolution, price and size, and I'm willing (but not eager) to forgive the size. So ... I am a Red guy.

Part of all this though is that there is a wildly different attitude about what is the camera's job and what is the DP/colorists job between Arri and Red. (And Sony, Panasonic et. al but they aren't in the ballpark today.)

Arri thinks they should make it easy to get a great look out of the camera so you can shoot it as a Rec709 camera and be happy. The flexibility of RAW is a distant second to them.

Red thinks you want to have absolute control over the image and focuses on giving you that control, even at the expense of ease of use or a canned look. RAW is the absolute center of what they do.

The problem is that they are both right, and they are both wrong. I want the camera to give me dead easy pictures that look gorgeous like Alexa does, then tomorrow I want every bit of control I can possibly get from a Red workflow.

So, if you really love film ... go shoot film! Alexa doesn't look like film, and Epic doesn't either. They can get you close, but they are not the same thing.

For me, Epic is the camera I always wanted. I really like Alexa - Arri gave it a great name after all. It's not what I really want. I'll use it and enjoy it when I do ... but it will always be a rental choice for me.

In the end ... we are splitting hairs, when we should be rejoicing in the embarrassment of riches. 5 years ago we would have killed for such wonderful choices in digital cinema.

 

[Jake Bastian]

Hi Jake

It is not going to make any closer to Alexa in terms of skintone. The color dictates by the sensor, filter usage, method of debayering and color space.

Alexa recording needs to have matrix apply, any transformation need to have LUT. But RED footage is different, it happens during the debayering process, and you can select it on your own preferences. Which what you are trying to suggest to have LUT and have it transform outside RED realm.
In Davinci, it exists 3D logC to rec709, but let's take a look at Graeme opinion in the previous post


-----
Graeme's post :
Just think back to earlier in the thread where Deanan mentions how Sony sets their red response filter characteristics wide to allow a lot of red through, Arri set their's narrow to let through less far red and RED set their's inbetween. If all three camera types record some deep far red colour - the Arri might not see it at all, the Sony sees it all and RED sees it inbetween the two. Now there are very valid reasons for why each camera manufacturer sets their response where it's at, but no LUT, IDT or matrix on this planet will make the Arri see the red it's native spectral response cannot see to match the output of the Sony which can see it. That is why, in one particular example, why you'll never get the same spectral response via ACES or any other colour space. Of course, three tristimulus values of a colour in a colour space will only ever approximate the actual spectral response of a colour under a real light with a real spectrum anyway....
-----

It involves other thing rather than mere color space. Characteristic of the sensor comes into play too...

my 2 cent

PS. Florian, if you can help to upload the data for testing purpose both RED and Alexa original footage, it will be great. I can help you to host the files if needed.

Rivai

Thanks Rivai. Doesn't the LogC to Rec.709 LUT take the Alexa footage out of the Alexa realm in that case? Or are you referring to the fact that ARRI supplies the LUT for the conversion whereas with RED you're kind of left on your own to produce a proper Rec. 709 LUT?

 

[RivaiC]

Hi Jake

Yes, the Davinci logC to rec709 i believe comes from arri themselves which implemented inside davinci. So if you have logc footage that u need to transform to rec709, you are free to use it in the 3D LUT section.

BUT this doesnt restrict you to apply to other footage ALTHO it is not meant to be. Like what you pointed out. You can use it with RED RAW which has RLF and RC2 setting. Apply logc to rec709 from the grading node. RLF is a cineon standard log develop by RED, it will be a bit different from logc. but RLF is still log

The proper way to view RED footage in rec709 is of course to use rec709 color space from the raw color space setting itself, so you are not doing it outside red realm.

I believe the skintone will not make any much difference, i wont be surprised if it look worse. What people here seem to miss is, the fact this discussion is about skintone in particular. It diverts to overall look of the camera, which cant be help, but it's really about skintone from the beginning.

What i try to understand now is, looking from the color chart, which section of the color that represent skintone ? I dont think red or yellow alone represent the skintone color. That is why skintone subject is very complex

Rivai

 

[Pietro Impagliazzo]

I also remember a discussion on David Mullen's thread where he was talking about how skin tones usually also play a large part in the lighting temperature (as mentioned here), make-up and grading.

I'm linking some images made by Gunleik Groven that he posted on his Epic test thread:

Add grain, pretend it's from film?

Oversharpen and say it's a Sony camera? The green hue is flat and invades every gradation, faces look too flat and the overall feeling is that it's too digital.

Everyone will believe, but both shot on Epic. :biggrin5:

And I think the main question is: How to handle RED footage in order to look more organic and not flat? That asks for a case study, REDucation, tutorials and whatever else one thinks of.

 

[Graeme Nattress]

Actually Rivai, for REC709 output it's good to select REDColor2 and then add saturation "to taste".

In terms of skin tone - it is indeed rather complex. It's a combination of type of light, skin, make-up (if used), how the light is used, filtration, and then camera and it's development of the image. And most important of all, it's down to the viewer's taste in how they like things to look. With my girls who I often use as test models, I like the way the RED camera differentiates their particular skin types - English Rose for Colette, and the different, more olive tones of Wendy's and Megan's skin. I did try to use my skin type for testing, but it was getting expensive with all the lenses I was breaking....

Graeme

 

[Jake Bastian]

Hi Jake

Yes, the Davinci logC to rec709 i believe comes from arri themselves which implemented inside davinci. So if you have logc footage that u need to transform to rec709, you are free to use it in the 3D LUT section.

BUT this doesnt restrict you to apply to other footage ALTHO it is not meant to be. Like what you pointed out. You can use it with RED RAW which has RLF and RC2 setting. Apply logc to rec709 from the grading node. RLF is a cineon standard log develop by RED, it will be a bit different from logc. but RLF is still log

The proper way to view RED footage in rec709 is of course to use rec709 color space from the raw color space setting itself, so you are not doing it outside red realm.

I believe the skintone will not make any much difference, i wont be surprised if it look worse. What people here seem to miss is, the fact this discussion is about skintone in particular. It diverts to overall look of the camera, which cant be help, but it's really about skintone from the beginning.

What i try to understand now is, looking from the color chart, which section of the color that represent skintone ? I dont think red or yellow alone represent the skintone color. That is why skintone subject is very complex

Rivai

I know paint mixes different than light, but to mix skin tone that way you need red, yellow, and blue which would mean skin tones cross all RGB values. I think the mistake most people make is trying to pull an HSL key on the skin by grabbing the reds and yellows rather than adjusting the entire image to make the skin tone look "right".

That's part of the reason threads like this bother me is because saying that RED produces worse skin tones makes your first instinct to correct the skin rather than the image as a whole. I get way better results when I ignore skin tone comments all together.

 

[RivaiC]

I think what Florian shows is the perfect example out of the box picture without any adjustment. it just that RC2 has a lower saturation comparing to original RC. But bare in mind, looking at the other colors of the color chart, it seems to match closely between RED and Alexa, without any adjustment. The only thing that stands out is the skintone difference and that is without make up ! I dont know if it has light, if there is anylight, the most standard is 5600K light, which in this case, it's an identical situation, doesnt matter. In this shot, i do prefer on how Alexa produce the skintone, it does look more pleasant, i wont comment any further than that, risking this thread to be close.

BUT i have to say, we have seen and grade out of the box Alexa picture which has a plastic skintone and greenish. That is make up and lighting condition which makes it look like that. It's a little hype to say Alexa doesnt produce plastic skintone, it does !

What pietro shows, it seems it has already been graded, so a bit hard to judge and both picture has different color temperature, lighting setup and subjects

Graeme, is it possible IF you can shoot your girl with same condition but with both camera EPIC and Alexa at the same time. What kind of skin color difference will you get and what preferences do you like.

PS. I dont know if it's possible for you to do that, but it will be a good case study.

Rivai

 

[Blair S. Paulsen]

IRL

IRL

How about a skin tone shootout at RSH?

Point an Epic and an Alexa at a couple models with different skin types. Put the images on the big screen side by side. Keep the actual models on set directly adjacent to the screening (lots of duvetine to prevent spill washng out the image) so we can see the way their skin looks in real life and compare it to what's on the screen. Would be nice to see both the "default" image and what a proper colorist can do to match each camera's output to the models...

Cheers - #19

 

[Graeme Nattress]

Blair, I think there was, sortof for a skincare commerical. As far as I know it went Epic. Jarred mentioned it a while back.

Graeme

 

[Bruce Allen]

I also remember a discussion on David Mullen's thread where he was talking about how skin tones usually also play a large part in the lighting temperature (as mentioned here), make-up and grading.

I'm linking some images made by Gunleik Groven that he posted on his Epic test thread:

I like this one! Yeah this is more like it. Actual play of different subtle colors across the face. Not afraid to have saturation... maybe a little much for "natural tones" but I think you might be able to make this one work with some grading. Encouraging! Will check out Gunleik's thread now Thanks!

How about a skin tone shootout at RSH?

Point an Epic and an Alexa at a couple models with different skin types. Put the images on the big screen side by side. Keep the actual models on set directly adjacent to the screening (lots of duvetine to prevent spill washng out the image) so we can see the way their skin looks in real life and compare it to what's on the screen. Would be nice to see both the "default" image and what a proper colorist can do to match each camera's output to the models...

Cheers - #19

Gotta include film too. Still beats the pants off Alexa for color IMHO. Epic shouldn't be trying to duplicate Alexa, they should be trying to surpass film Maybe I am crazy though.

Bruce Allen
www.boacinema.com

 

[Deanan]

Florian - if you look at the skin of your hand holding the chart, it looks a better on the Alexa. Look at the subsurface scattering and subtle red shades on the fingers!

Except that alexa cuts off most of the red response so it's doing the exact opposite of what you're saying.

 

[Ricardo Moreno]

I know a lot of people who "don't like RED color" or say "Alexa rocks" to which I respond...
1. You're an idiot
2. You just don't like RedColor or RedColor2 or whatever default processing they used for the dailies.
3. Wait till they grade it
4. RED does seem to be trickiest when working with the red tones though. Trick is not to clip anything. Expose to the left. And give the poor ol' blue channel a chance. Shoot daylight or try a blue filter on the camera if you shoot tungsten.

AWESOME!!

 

[RivaiC]

Except that alexa cuts off most of the red response so it's doing the exact opposite of what you're saying.

So what makes the skintone stands out ?

EDIT : Looking back at the chart. The red color from alexa seem to match EPIC red color.

 

[Deanan]

So what makes the skintone stands out ?

EDIT : Looking back at the chart. The red color from alexa seem to match EPIC red color.

I think that depends on your perspectives and prejudices. I've seen enough epic vs alexa tests (they just won't stop with the tests) where epic has been as good or better than alexa.

There one particular test where people insisted that they didn't need to test epic because alexa had better skintones. They tested and epic won the skintone test. This was for Victoria's Secret, Every skin tone and shade you could imagine on real life models in real shooting environments under different lighting, daylight, strobe, tungsten and HMI.

 

[CJ Roy]

I think that depends on your perspectives and prejudices. I've seen enough epic vs alexa tests (they just won't stop with the tests) where epic has been as good or better than alexa.

There one particular test where people insisted that they didn't need to test epic because alexa had better skintones. They tested and epic won the skintone test. This was for Victoria's Secret, Every skin tone and shade you could imagine on real life models in real shooting environments under different lighting, daylight, strobe, tungsten and HMI.

Deanan- It'd probably be good for you guys to publish some of those tests. Worth a shot and might save some time.