T.Stops Blog

Extracting the best color from your RED camera. An informal experiment.

The ease of shooting digital is a double edged sword.  Seeing what you are getting, without delay, is    so easy and quick.  However, the simplicity of turning on the camera and getting to work, often leads to a false sense of security.  Seeing an image, off the bat, can make a cinematographer a bit lazy.   The image is there, and it looks good…. Good, isn’t enough for me personally.  I want the best I can possibly put forth to the digital negative.

In the film days, a DP had to know his stock inside out and backwards, being able to mentally calculate color correction on the printer lights for the dailies, how much to push, pull, and the various ways of processing.   Look to Darius Khonji ASC’s phenomenal work on Seven.   He pre flashed the 35mm negative, meaning he carefully exposed the entire stock to a very low light level, before shooting to get some base exposure in the “Blacks”.  He then exposed the film far darker than what would have been possible, but reatained incredible detail in the shadows and a smoky look, that was in the chemistry of the film itself.  Without his full and complete knowledge of the chemistry, exact light levels, and possibilities of prepping the emulsion both before rolling, and in the lab, he would have never been able to achieve this effect, repeatably, every take.

To be at the top of your game in the film days, this level of knowledge was paramount.  It still rings true today as an emerging cinematographer.  As I discover new techniques to add to my mental arsenal, I feel it necessary to share the information simply because it may help someone someday.

So….. How can a person know the digital sensor inside and out. Well, a good place to start is in the “lab”, with a good DIT who has video engineering experience   Tom Wong, IATSE Local 600, DIT an I took some time and began testing to solve a particular problem.   The rumor on the street is that the RED Epic is weak in the skin tones on subjects.  So we set out to look into this “problem” a bit.  It is true that out of camera the image falls a bit green.  This is the nature of bayer sensor cameras. My F3 was a bit green, and I also feel the F55 is as well, but is much better controlled.   Then there is the Alexa.   Its just perfect, out of the box.  How? Why? Whats different…..

Our findings were very interesting. The first thing we did, was examine the channels the camera captures, of a grey card left over from an old shoot. Since the image is RAW, age of the shot didn’t matter, and it was from a green screen shoot, so we know the shot was from a controlled environment.  Green Screen, under tungsten lighting, with big tungsten lights.  Big lights are important for this kind of test, and we used this clip for a reason.  Big tungsten filaments are very stable.  They have so much mass, that they burn very consistently.   This experiment was not 100% scientific, as it was more for experimentations sake, but its results were rather interesting.  (I was not even planning on writing about this, or I would have taken screen captures from our test.)

As we all know the digital sensor has a native kelvin rating.   For a RED camera it is 5000K.   So, in essence, this is the color temperature where you will achieve full balanced exposure on all three channels of color on the sensor.   This is why shooting in daylight, and under daylight balanced (5600K) sources have a tendency of producing better quality images on RED than under tungsten (3200K).  By better I mean slightly cleaner with a more natural color reproduction.   This would be all well and good in perfect world, but the truth is it seems the RED chip is more sensitive to certain colors.  Particularly red, and its least sensitive to blue, hence needing blue light (daylight) to render colors faithfully.   That said, this seems to causes another problem.  Too much blue.  Most daylight sources have color spikes.   HMIs and Kinos are not immune to color shifting.  (though plasma lights are seeming to be the answer…)  Colors thin out a bit and to get a golden beautiful skin tone, you need to set your white balance to a kelvin higher than 5600, to retain it.  This boosts the RED channel, and you get unpleasantness again.  Now you have to remember, im talking small tiny shifts of color.  But, small variations are perceptible by the eye.  The difference between a beautiful on camera appearance and an unimpressive one is just the way a camera renders skin color.  So small variations do make a difference.  So what to do?

Well, after some experimentation, we came up with the following.

Light with 3200K tungsten units, but use a 80C filter.  It is a 1 stop cut 3/4 power CTB glass filter on the lens.   Tom’s theory is this: The red(ish) light of the tungsten filament is being filtered out by the blue glass. However, because the source light is “red” it is still pushing through the blue glass.   So what appears to happen is, your color temperature on camera sits around 4900, I believe we were at 4917K when manually white balancing off a grey card on set.  Your blue channel is being fed properly, and you still get a healthy red channel. Green will naturally be taken care of.  Some magic happens, and everything kind of falls into a proper balance.

The results are this:  The latest Sesame Street piece I DP’ed.   Three Epics, three Fujinon Cabrios, 2x 19-90’s, and one brand new 85-300 Cabrio.  Tom Wong, DIT built a look based on the grey card from the scene, and set the WB in RCX and loaded the look into all three cameras.  Aside from a small boost in contrast, and a little saturation the look is mostly in camera. Juan Salvo, colorist did a great job with last touchups and matching between cameras.

Even on the highly compressed Youtube video the results are pretty darn good.  This is the only place these videos can be seen.  I wont have access to higher quality versions. Ever.

Pay attention to the park rangers skin tone. Golden and healthy, while remaining clean as a whistle with overall rich, accurate color reproduction.  The Sesame Street characters must be within a very critical spec for color.   Often correcting for fabrics can wreak havoc on skin tones. We were able, using this technique, achieve great color in both the skin and the fabric.

To all the RED shooters out there. I used a Formatt 80C filter.   I happen to like Formatt Filters.  I think they did the trick on this shoot.

Thank you for reading.

You can follow me on Twitter @Timurcivan for more updates from Tstops.

Coming up Next:

3D Sound capture… The Mitra Microphone.   Just so cool.  I have to write about it….

Leave a Reply

3 Comments on "Extracting the best color from your RED camera. An informal experiment."

Notify of
Sort by:   newest | oldest | most voted

What do you recommend if you are shooting outside in daylight?

The sun is naturally balanced.

When shooting under HMI or Daylight Kinos, it may help to use a 1/4 CTO filter on the lens.

But I haven’t tested this yet.

First of all, thank you for writing these posts! I discovered this site from your article in NFS I am a big fan of the Sony F3, and I was looking into this topic after reading that Jeff Cronenweth used an 80D filter on his work with Fincher. Do you think that all digital cinema cameras need this filter? Does the F3 need it? I could not find the kelvin rating for any camera other than REDs; but despite the difference in sensors, my understanding is that they all receive and process light in very similar ways. Jeff mentioned that… Read more »
wpDiscuz