Dreaming of color balance

[Raul Quintanilla]

First of all excuse me for my silly question, but I guess this is the best place to ask about the future of cinema.

Could it be possible that one day color balance in a camera won't be neccesary. What I mean is, could we have in the same frame a room lighted with tungsten and another room with fluorescent, and that the two look balanced (and not by balancing to the middle), but to really match what the eye sees. Maybe not done in-camera but in post (I guess you can do that manually, but with a moving frame it seems difficult). Some kind of software that can detect the different light sources and combine them naturally.

Could this dream come true or is this physically impossible?

 

[Alexander Ibrahim]

When I look at a space with mixed color temperature lighting, I see different color temperatures.

In other words the different color temperature lighting results in different color as seen by the eye.

The major complaint is that photographic media (both film and digital) exaggerate the different tones as my eye sees them.

So, could sensors be made that are less sensitive to color temperature? Sure- but you risk accurate color rendition.

Could sensors or camera systems be made whose handling of such color differences is more subtle? Possibly... but only when cameras are smarter and have better aesthetic sense than DP's, gaffers and the entire G&E department.

Could camera systems be made that attempt to switch where they perceive "white" to be on the color temp scale as our eyes do? Easily... its called autowhite balance, and most cameramen hate it.

Can you color correct an image in post to balance lighting color temperatures? Absolutely yes... its a feature that has been available to colorists for more than a decade.

For myself the current state of affairs is pretty good. There could be some improvements, but basically I am pleased.

In fact I often use different color temperature light when working. Technically I suppose gels and such always alter color temp, but that isn't what I mean.

I often mix daylight and tungsten in a scene- because that is how real spaces are. You see people whose face is filled with tungsten... but the hair and rim light comes from the sun via a huge window.

That's the reality of the spaces we inhabit.

So, maybe I am completely missing the point of your question, but I don't see the problem.

 

[Charles Angus]

Nope.

 

[Vico Martin]

the day we will not need to do a White balance on-set it´s the day we use a RAW camera. ;-)

 

[Raul Quintanilla]

I know the benefits of RAW and I know most things can be done in post. Let me explain with an example. I white balance a scene to tungsten lights, but in the background there is an open door that leads to a room with fluorescent lights. I'm seeing the fluorescent lights much more blue than with the naked eye.

In post maybe I could divide the frame and white balance the door opening to fluorescent light. But then a person passes from the fluorescent room to the tungsten room. If he is standing straight in front of the doorway, he is going to be balanced to fluorescent light, even if he already entered the tungsten room. Did I explain well? Of course this would imply a much tedious work of balancing frame by frame so the transition from one room to the other looks real in the person's skin. That's what I wish could change.

That in post, since working in RAW, a software could understand the different light sources and could do seamless transitions, just like the eye would. That I wouldn't need to divide frames, just point at the different light sources to be identified and let the computer do the rest (or even let the software identify the sources). And if there is a space where the two light sources hit, let the software figure the exact balance.

Yeah, I know automatic white balance is not used by professionals, but that is just because the technology isn't there yet. I read somewhere that this problem is soon to be tackled by the big camera companies.

Maybe the answer to my question is simply "nope", like Charles just posted...

 

[michael webb]

color balance

color balance

"The major complaint is that photographic media (both film and digital) exaggerate the different tones as my eye sees them. "

The real problem isn't that the camera exaggerates but that they don't automatically compensate like the human eye and brain do. the system humans have is really pretty nifty in that the auto white balance and color replace in their brain, something cameras and computers don't do as well yet.

 

[Scott Crawley]

the day we will not need to do a White balance on-set it´s the day we use a RAW camera. ;-)

He's talking about mixed light within a frame. At the risk of flogging the pony... RAW itself is not a solution to that dilemma.

What I don't understand is why folks are so interested in a system that would auto correct en-masse in the first place. It's analagous to a painter complaining that they have too many colors to paint with, or choosing to work in grey simply because he can't be bothered to master color theory.

Learn to control the tools you have. ;-)

 

[Raul Quintanilla]

Hmm I am a one-man crew, and not only for events but for short films (even a feature film). And there is certainly a lot of work to do specially with sound and color, but something I really haven't learn to use is After Effects, or green screen, or complex special effects (which I never really need). But I recently had this problem I just described, to fix it I would have needed to work with something like After Effects or pay someone (I don't have the budget). So I finally left it as it is, it's nothing too serious.

I don't want a camera or editing software to do the creative work for me, I just want them to replicate what I see, and then if I want to change something, I'll work from there.

Your metaphor actually made me think of those first filmmakers who painted over the film, since the lack of color; if they dared to complain and ask for color film? or like Antonioni having to paint the trees, streets and fruit to balance the colors in the technicolor film (and still look a bit fake), that made the production costs much higher and troublesome (he was actually very happy when in Mystery of Oberwald he could finally manipulate colors electronically, he thought that was the future and it is!).

Yeah, I'll try to learn the tools I have, but I also want them to evolve and make things easier and better (should I say more realistic?), I thought that's what RED is doing. I'm sure if RED comes with a solution for this in the future, everybody will use it and think of it as a great revolution in color science.

 

[Scott Crawley]

Your metaphor actually made me think of those first filmmakers who painted over the film, since the lack of color; if they dared to complain and ask for color film?
Yeah, I'll try to learn the tools I have, but I also want them to evolve and make things easier and better (should I say more realistic?)

A pioneering photographer wishing for color film as opposed to B&W was looking to represent more reality than he was getting. Our cameras are delivering more reality than we perceive. So no I wouldn't say more real. Your example is all mixed up.

 

[Todd Norris]

At first glance, the idea of a sensor that can "intelligently" handle mixed color temperatures smacks of someone who doesn't want to learn the craft of lighting. However, upon thinking about it, the idea (in the abstract) seems cool. Imagine a sensor that somehow interpreted light in the same way the human eye/brain does. I can imagine a hypothetical technology (a sci-fi solution to this whole idea) that could analyse the photons that struck the sensor, determine it's source (daylight, fluorescent, tungsten, etc) and attenuate the final result in-camera to replicate the colorimetry of the human eye. This doesn't seem like an outlandish request for an artist. If a different look than "human eye reality" was desired, one could simply turn the function off and revert to a traditional sensor.
I don't know how this technology would work. Again, it's science-fiction at this point. But eliminating the need for CTB on tungsten lights, or the need to gel windows with CTO in mixed lighting scenarios seems cool in my book. Not to mention green/magenta issues in fluorescent lights that seem to pose no problems to the human eye. The trick, of course, is in the cases where you're using gels to intentionally tint the light...how would the "intelligent" camera know the difference?

 

[David Mullen ASC]

In the abstract, it seems like a bad idea actually -- you want a camera to distinguish between different subtle shades with minimal crosstalk, and honestly, how is a camera to know whether a subject is a certain color or it is merely being lit with a certain colored light that is different from the rest of the light in the room? It all comes through the lens and hits the sensor as color information. So do you really want a camera eliminating shades of blue, green, or orange merely because it thinks it's due to mixed lighting?

Better to capture all that color information and deal with it in post than to toss it away in-camera ("What happened to my orange bedroom?" "Oh, the camera thought it was a white bedroom shot under uncorrected tungsten light..."

Besides, color temperature is a HIGHLY creative visual tool and I find it more useful than problematic that different lights have different colors. I find it only annoying when I'm dealing with a green cast on blonde hair and pale faces, then it's a bit tricky to deal with in post. But contrasting the warmth and coolness of different sources is a great way to create interesting looks.

 

[Raul Quintanilla]

Thanks David, it made it more clear to me as to why its impossible. I guess what I need in my line of work is for cheaper and brighter LED lights with an adjustable color temperature.

 

[Graeme Nattress]

David's spot on - any algorithm that tried to do this would in effect undo the good work the DP has done setting up the colours of the lighting just-so for the scene. When our eyes white balance they adapt like a camera would, and also adapt in the brain too where we "know" certain objects and things are "white", and no longer see their actual colour, adapted or otherwise.

Graeme

 

[Alexander Ibrahim]

Thanks David, it made it more clear to me as to why its impossible. I guess what I need in my line of work is for cheaper and brighter LED lights with an adjustable color temperature.

It sounds like, for your work, you are absolutely right- variable color temperature light fixtures are ideal.

If that is your main line of work, I question if a RED is the right camera.

I think I'd prefer a Panasonic HPX 2000 or Sony XDCAM EX F350- a more traditional ENG type shoulder mounted camera. (My pick would be the Sony with a NanoFlash actually)

I'd probably stay away from RED, for your work, until the Scarlet 2/3" is available, then follow as people like Steve Gibby (do I have the name right?) build rigs out of it for ENG/EFP work. Steve does run his RED for this sort of work currently, but I'm impressed at how he manages focus. Its not as easy as he makes it look. I wouldn't plan things that way, but he manages it.

 

[Raul Quintanilla]

Actually I want a fixed lens Scarlet, but right now I work with a Panasonic DVX100a and a Canon 7d that for most cases is just fine (specially since here in Mexico the HD transition is far from ready). Thanks for all the responses!

 

[Scott Crawley]

It was interesting to think about Raul. Here is a little more food for thought: what would such a system do with this?

 

[Todd Norris]

I still stand behind my assertion that it's an interesting idea IN THE ABSTRACT.
In my first post I qualified that it would need to be something that was an option that could be deactivated. I also mentioned that it was simply a "what-if" science-fiction speculation. And at the end of of my post I brought up the paradox of how the camera would know what the DP wanted when it came to the color of the light. I agree that a sensor of this type is not likely to happen, nor is it necessary. I just liked to speculate rather than immediately call it ridiculous. I guess it's the sci-fi nerd in me that likes to at least consider things.
I suppose what I'm getting at (and perhaps what I the original author of this thread was getting at), was the discrepancy between the human eye and film/video when it comes to color perception. If the aim is realism (and I'm not saying it is), wouldn't something that replicates what the human eye sees be valuable? Wouldn't color fidelity be as valuable as dynamic range?
Again, I'm just providing food for thought. I'm not trying to minimize the job of the DP or limit creativity.
As an example, when 3-strip technicolor superseded 2-strip technicolor, wouldn't that have been a step in the evolution towards a more realistic color replication?
I'm rambling and going down a road that has no real answers, so I'll stop. I just wanted to defend my desire to at least consider the value of a sensor that saw the world the way the human eye does.

 

[David Mullen ASC]

Not sure that the human eye doesn't perceive color temp differences -- my eye does. Our brain just tends to white balance very quickly, just as our brain auto-exposes and auto-focuses very quickly, neither of which we tend to want a camera to do either. In other words, if you had a decent auto white balance on a video camera, it's going to do a lot of similar things that the human eye does in averaging color temp. But we all know the problem with auto WB just as we do with auto focus and auto exposure.

The only area where the human eye is a bit insensitive to differences is in partial spectrum sources which have a strong color spike, like in green wavelengths. That's one of those cases where the camera is overly sensitive compared to the eye-brain. Over the years, in film stock technology, there was some attempt to address this with the Fuji Reala stock, which essentially seems to partially counteract green spikes with an increase in magenta. Unfortunately, since digital sensors often get most of their luminance and resolution information from the green photosites, I don't see a way of decreasing green sensitivity without affecting overall speed and sharpness.

Ultimately "realism" is just a subjective quality -- a stylistic decision -- as anything else. After all, we don't see reality at 24 fps with a 180 degree shutter either. And we can focus quickly from near to far (except when you get older) and we can adjust quickly from a bright background to a dark background. All of which give the illusion that our eyes have deeper focus and see more dynamic range than they actually do. Same goes for color temp -- it's only an illusion that our eyes can't see the differences between colored sources. Certainly a trained set of eyes sees them. It's just that our brain is constantly averaging out color temp like a very fast auto white balance.

But in a movie, we generally don't want to see the color temp constantly being corrected, or the focus shifting rapidly from near to far, or exposures constantly being averaged out as the camera pans around a room.

 

[Scott Crawley]

The difference between reality and perception may seem a technicality, but it is at the heart of the matter.

Perception is a cognitive process. Our eyes accurately convert light energy into image data that we can use. Our brains then process that data. This is perception. Barring defect, the ability of one person to perceive diferences in the color temperature of light where others can not is due to careful observation and training.

CMOS sensors provide the same function as our eyes, but in our camera systems and do so with relative accuracy. Accuracy is really the most that we can expect from a sensor.

As Photographers, our jobs are part technician and part artist. Accurately capturing and recording image data and conveying it in a way that it can be peceived the same as it would be in the first instance (Without our media as middle man) is the technical part of the job. Documentary and news type of shooting done well is almost purely technical. When shooting those kinds of material, the idea proposed by the OP makes sense.

The art side of our jobs involves the manipulation of perception, both in the rendering of the image and in the message that is ultimately conveyed. With that in mind, it could be said that the art of narrative cinematography is primarily concerned with image processing. (Incidentally, that is why I am as much in love with DI as I once was with the darkroom.) Artistic shooting requires control of this processing, and those shooters aren't so much interested in auto-balancing. We want to make those creative decisions ourselves and not leave them to the camera's brain.

It is apparent to me that if the kind of processing proposed by the OP were to evolve into an actual technology, it would need to be an optional process that could be turned off.

 

[Todd Norris]

Scott,

Well said.

I think the first time I really noticed that my eyes were "auto white-balancing" was in my teens when I looked out the window at dusk. The sky and all the houses on the block looked blue while the lamps in my house looked white. I went outside and after only a minute everything outside in the fading twilight looked "normal" while all the lamps inside the houses on the block looked orange-yellow. It was my first inkling that color perception wasn't fixed and constant.

BTW, there are scenes in "Tucker" shot by Vittorio Storaro that have some of the best "evening blue" scenes. His lighting not only looked realistic, but the colors he rendered on film held an emotion that was pitch-perfect. Genius is in the details, I guess.