Dragon Non-Linear Color Response?

[Björn Benckert]

There is one scenario where changing makeup is not an option - an advert for a specific brand of makeup!

I do agree makeup tests for every other scenario would be useful - maybe Red should run some more comprehensive tests themselves, including comparable Alexa, Sony, Canon cameras as comparison, and with a range of skin tones.

These probably should NOT be done or announced publicly as they will want to fix any issues without feeding FUD !!

You think they actually use the gue thats for sale?

 

[Björn Benckert]

Ok, so if we are done calling people "DP"'s if they did not do a make-up test can we talk about what is important.

If this is still an issue on Dragon i really don't care about the "how" or the "why?"

I am looking at this with a producers eyes for a minute, I want to know if this can be fixed by RED.

So i guess its a question of if this is an issue inherent in the hardware or if firmware/colour science can deal with it.

If it can be fixed by firmware or colour science then the questions are 1. when can we expect this fix, 2. how RAW is RAW if processing etc in camera is used to alleviate it , and 3. why was this not done for MX if it is fixable.

My gut is that this is not something that can be fixed on the software/processing side (in camera), mainly due to how RED does RAW.

Maybe we need a new slider in Redcine, "greenaway"....

I dont really see what you are talking about, this is straight of redcine no setting exept for whats in camera... no curves no nothing. whats wrong with these skintones? I do not see it?

http://www.flickr.com/photos/107237384@N04/12863353275/

even in unbalanced underexposure dragon beats the shit out of any camera I tested...

hardly any light tungsten shot balanced.

http://www.flickr.com/photos/107237384@N04/12546637984/in/photostream/

the same shot with sliders to native.. i.e 800iso 5600k:

http://www.flickr.com/photos/107237384@N04/12546448695/in/photostream/

To me this shows that Dragon is pretty freaking linear. No doubt or the color shift from my red/yellow face to balanced would cause oddites, but it does not it just looks right when kelvins are changed. Some low level blue noise sure but thats a given when there is hardly any blue light in the shot. So not saying this is a way to shoot makup but it for sure is a workable picture for most, and most other cameras would have been toast for the same capture.

 

[Ivan McCullough]

I dont really see what you are talking about, this is straight of redcine no setting exept for whats in camera... no curves no nothing. whats wrong with these skintones? I do not see it?

http://www.flickr.com/photos/107237384@N04/12863353275/

I don't follow Björn, is this one of Josh's examples which we are discussing? Doesn't look NSFW to me

 

[Björn Benckert]

I don't follow Björn, is this one of Josh's examples which we are discussing? Doesn't look NSFW to me

No thats just a dragon frame, I though the discussion was if dragon was good enough for skin tones and handled colors linear etc. Then, to me at least, those pictures shows that, well it handles skin tones pretty good and linearly. But just as any camera only if you expose somewhat with in range. Josh shots are way under expo when you bring the levels up you get poor color resolution blue is gone and red and green does not have much to offer so the result is poor.

 

[Ivan McCullough]

No thats just a dragon frame, I though the discussion was if dragon was good enough for skin tones and handled colors linear etc. Then, to me at least, those pictures shows that, well it handles skin tones pretty good and linearly. But just as any camera only if you expose somewhat with in range. Josh shots are way under expo when you bring the levels up you get poor color resolution blue is gone and red and green does not have much to offer so the result is poor.

Maybe have a read over the thread then.

It's moved on, the infamous green skin with certain make-up seems to be present on dragon.
And yes I am aware tungsten makes it worse but I would have thought dragon would have fixed what other cameras have no issue with.

 

[Scott Crawley]

Ok, so if we are done calling people "DP"'s if they did not do a make-up test can we talk about what is important.

If this is still an issue on Dragon i really don't care about the "how" or the "why?"

I am looking at this with a producers eyes for a minute, I want to know if this can be fixed by RED.

So i guess its a question of if this is an issue inherent in the hardware or if firmware/colour science can deal with it.

If it can be fixed by firmware or colour science then the questions are 1. when can we expect this fix, 2. how RAW is RAW if processing etc in camera is used to alleviate it , and 3. why was this not done for MX if it is fixable.

My gut is that this is not something that can be fixed on the software/processing side (in camera), mainly due to how RED does RAW.

Maybe we need a new slider in Redcine, "greenaway"....

Understanding the how and why is key to problem solving and DPs are problem solvers. It's in the job description. That is what we were talking about. This community is frequently part of the solution, but we can't be if we ignore the how's and why's an just point to Red and wait.

 

[Ivan McCullough]

Understanding the how and why is key to problem solving and DPs are problem solvers. It's in the job description. That is what we were talking about. This community is frequently part of the solution, but we can't be if we ignore the how's and why's an just point to Red and wait.

Scott, if I ever need tips or info on what the key to being a DP is I'll be sure to get in touch. Right now I am trying to have a dialog with a camera manufacturer about a problem I and others have witnessed on their cameras.

If you have something to add to that discussion from your experience of this issue then please do. Otherwise there are plenty of people over on NFS discussing what being a DP is.

 

[Scott Crawley]

<Passive aggressive insult deleted.> "Right now I am trying to have a dialog with a camera manufacturer about a problem I and others have witnessed on their cameras."

Graeme isn't airing his dirty laundry here if he has any, but he is here reading. So in the interest of helping in whatever way I can, I ask myself what does he take from this discussion? How would I use RedUser if I were the architect of Red's color science?

Well I have to back up, he is here so let's ask him. Graeme, how can we best help you?

If I were a scientist I would not be looking for the theories of a non scientist like Scott Crawley who has admitted to having no direct experience with this. (If that is your point, I get it.) I would be using this site as a research tool, a repository of data contributed by users who's expertise is different from my own. I would want to know the details of the how and why.

That is why I have attempted to direct the discussion to those things you choose to ignore, not to have a side argument or lecture you.

I tend to use "you" in the general sense while discussing these things. I can only ask that You and Eric not take those remarks personally but I'm not sugar coating it.

 

[Graeme Nattress]

Well I have to back up, he is here so let's ask him. Graeme, how can we best help you?

Two issues here - one is raw development which is always under constant R&D for improvements, 2nd is makeup. Best advice for makeup is to always ensure the light used to apply the makeup is the kind of light used to illuminate the talent. I did some testing on makeup and found that it would appear different to the human eye depending on lighting conditions, and that difference was also seen by all the cameras I have here. I also discussed makeup with some top makeup people, and they refused to blame any camera for makeup issues, even with lots of prodding from me.

Graeme

 

[Ivan McCullough]

Two issues here - one is raw development which is always under constant R&D for improvements, 2nd is makeup. Best advice for makeup is to always ensure the light used to apply the makeup is the kind of light used to illuminate the talent. I did some testing on makeup and found that it would appear different to the human eye depending on lighting conditions, and that difference was also seen by all the cameras I have here. I also discussed makeup with some top makeup people, and they refused to blame any camera for makeup issues, even with lots of prodding from me.

Graeme

You know your stuff Graeme but you cannot deny that this is not the first time you have heard these things.

Eric has talked to you about this before, I remember reading the thread. His experience is that other cameras on set at the same time did not exhibit the issue.

From my experience the issue is there, do I think RED should do something about it? Yup, for the sake of RED, cause I like their cameras ,and would like to stop having to make excuses for them.

Am I gonna get a bunch of different complexion models, various different lights and every brand of make-up I can think of to test this and find the common denominator/s... Hell no, unless someone wants to pay me for a week+ of work to do it.

I already know the common denominator that can rid me of this problem, I don't like that option, other people could care less about not even considering a RED camera and that is likely what happens when you have a doubt about performance.

 

[Scott Crawley]

To be clear, this is what he posted.

I have shot side by side with a 5D, and the 5D does not render the same problem as the MX sensor. I have not encountered the problem with the Alexa.

It does not say that he shot Epic and Alexa side by side and he does not expressly say here that the Alexa does not exhibit the issue, though he apparently leans that way.

It would be great if we could reproduce the problem and shoot Epic and Alexa side by side. Then we would be on a track that could help clarify that point. To do that though we need to know the specifics of makeups and lights that will produce the problem. I have followed this discussion since the MX release and have not seen much specific info on those, particularly the makeup.

Does anyone know a makeup artist or two who is up on this and might want to join in?

 

[Ivan McCullough]

To be clear, this is what he posted.



It does not say that he shot Epic and Alexa side by side and he does not expressly say here that the Alexa does not exhibit the issue, though I suspect that is how he leans.

It would be great if we could reproduce the problem and shoot Epic and Alexa side by side. Then we would be on a track that could help clarify that point.

Ok so from Eric we have 5D does not exhibit the problem.

In my case the make-up artist told me she had used the make-up on D-21 jobs and on Alexa with no issue. I believe her but thats third hand at best, accepted, we are not the only ones who have commented in the past with regards to this issue.

Anyway, what is your deal? do you deny that there is an issue on the face and neck of the model in Josh's grabs?

If you accept there is an issue do you not want to find out (years later) what it is? Do you not think it should be a priority for RED (considering how long and how many generations of sensor this has been a problem for)

I remember tungsten on build 16, Jesus that was ugly, we have come a long way, can we just catch up to where the other guys were 10 years ago in this regard?

 

[Scott Crawley]

Ok so from Eric we have 5D does not exhibit the problem.

In my case the make-up artist told me she had used the make-up on D-21 jobs and on Alexa with no issue. I believe her but thats third hand at best, accepted, we are not the only ones who have commented in the past with regards to this issue.

Anyway, what is your deal? do you deny that there is an issue on the face and neck of the model in Josh's grabs?

If you accept there is an issue do you not want to find out (years later) what it is? Do you not think it should be a priority for RED (considering how long and how many generations of sensor this has been a problem for)

I remember tungsten on build 16, Jesus that was ugly, we have come a long way, can we just catch up to where the other guys were 10 years ago in this regard?

FWIW I also have not seen it in my older 5D stills.

It would be helpful to know what lamps you were shooting under when you encountered the problem, what was the metered exposure and color temp, what were your exposure settings, and what was the temperature on the set?

I would tap your makeup artist for more info. Info like what specific makeups she knows to be involved and if they were applied under set lighting. What kinds of complexions has she seen it on, and does she believe that specific complexions are more or less prone to present a problem. If so, why does she think that is the case? Just what might be the cause from a makeup artist's perspective?

I would be interested to know those things.

 

[Ivan McCullough]

FWIW I also have not seen it in my older 5D stills.

It would be helpful to know what lamps you were shooting under when you encountered the problem, and tap your makeup artist for more info. Info like what specific makeups she knows to be involved and if they were applied under set lighting. What kinds of complexions has she seen it on, and does she believe that specific complexions are more or less prone to present a problem. If so, why does she think that is the case? Just what might be the cause from a makeup artist's perspective?

I would be interested to know those things.

But you just ignore my questions.

You are obviously very young with ego issues Scott. You do not like to admit ignorance. This is borderline trolling at this point.

I hope Graeme takes it more seriously than you do.

 

[Hrvoje Simic]

Ok so from Eric we have 5D does not exhibit the problem.

In my case the make-up artist told me she had used the make-up on D-21 jobs and on Alexa with no issue. I believe her but thats third hand at best, accepted, we are not the only ones who have commented in the past with regards to this issue.

5D does its own thing with color casts and conceals a lot in 8 bit mush, which is substantially less precise than 16 bit image acquisition. Some 16 million times less precise.
D21 and Alexa jobs typically use tested types of light sources.

It is likely that most wish to solve the issues, where ever their origin may be, but before they are solved they should be properly determined and defined and one test of this sort is insufficient to provide all the required data and basis for proper conclusions.

 

[Eric Haase]

Two issues here - one is raw development which is always under constant R&D for improvements, 2nd is makeup. Best advice for makeup is to always ensure the light used to apply the makeup is the kind of light used to illuminate the talent. I did some testing on makeup and found that it would appear different to the human eye depending on lighting conditions, and that difference was also seen by all the cameras I have here. I also discussed makeup with some top makeup people, and they refused to blame any camera for makeup issues, even with lots of prodding from me.

Graeme

Graeme, I'm glad you are here reading. I would LOVE to help get to the bottom of this issue. From my experience with this issue, the makeup is always applied under same lighting conditions, which is usually tungsten or 2700K. When I look at it by eye under this lighting, it blends with the skin tone well and does not look green. When seen on camera with white balance set to 3200K the area where makeup applied is noticeably green or greenish/orange. I have had talent that usually appears on broadcast television shows (Access Hollywood for instance) come to set with their makeup done exactly as it is on their show. Mind you, we are shooting on THEIR set with THEIR makeup team. They step in front of the RED and the face looks green and the rest of the body does not. Same color temp lights illuminating them on their own set with their regular makeup. They step in front of their cameras and they look great- no green face. This is just one example of how I have seen the issue and known something is not right with the sensor or color science. I also pull out my 5D and check against that and have noticed that the 5D does not exhibit this problem. As for the Alexa, I've never run in to the issue with it and speaking to crew I work with they have never seen the issue on Alexa but HAVE seen the issue on RED on other DP's sets.

I would love to find some problem makeup and side by side test to show that this camera renders these tones in a strange way and hopefully find some kind of fix that does not require scrambling to redo the talent's makeup when the clock is ticking on set. It just isn't a practical solution. I've tried using a bluish filter, developing using RC2/RG2 and lastly, pulling secondaries with mattes in the final grade. None of these solutions are optimal nor do they produce a clean image comparable to what one would expect with proper tonal rendition.

I'm going to get started on a test and report back what remaining resources I need to make it happen. Maybe others here could pitch in.

 

[Scott Crawley]

Ivan,

I am asking serious questions and ignoring personal questions like "What is your deal?" because they are not relevant. If I am ever in the market for a personal growth coach I will look you up. Thanks for the tips.

Your demonstration of classic trolling technique; launch a personal attack with inflamatory statements then accuse the target of trolling, has been particularly helpful.

I believe I answered the question "Do you believe there is a problem?" several times.

As for the OP's example, I am happy with what others have said and have nothing better to add.

If you continue to make this personal, I will have to use the ignore function and encourage you to do the same.

 

[Elsie N]

Was this present when shooting the milk girls? Maybe it was their lighting but that was near perfection as I recall (nothing is ever perfection because someone will always see something in an image differently.)

OBTW, I'm not suggesting the ones seeing the issue are outliars.

 

[Ivan McCullough]

Was this present when shooting the milk girls? Maybe it was their lighting but that was near perfection as I recall (nothing is ever perfection because someone will always see something in an image differently.)

OBTW, I'm not suggesting the ones seeing the issue are outliars.

If you or anyone who looks at it cannot see it then it is not present.

When it is present it is obvious, if you had seen it you would know.

Generally under tungsten lighting. Milk girls was daylight afaik.

 

[Scott Crawley]

Was this present when shooting the milk girls?

As I recall, the first posts on this began to crop up some time after the MX release, but I could be wrong.

One of the first posts I remember seeing was in 2011, by a user named Axel Ebermann. I haven't seen him here in quite a while.

It is interesting to note that Mr. Ebermann was able to replicate his findings on his 5D.