56K vs. 32K Color Rendering

W

William

Active member
Just wondering... If one lights a scene using 3.2K and set White Balance accordingly for 3.2K... Would/should the color rendition be the same using 5.6K lights with WB set for 5.6K? Would one set up be warmer or cooler than the other, or the same? Are today's digital cameras optimized for 5.6K or 3.2K?... Or no real difference with proper WB?
 
Last edited:
Depends on the camera but my understanding is that most image sensors are designed such that the blue and red channel exposure is best balanced around 5000K. As for color rendering, in reality there are a lot of differences in lighting technologies what make 3200K vs 5600K hard to match exactly. Basically 3200K is particularly useful because tungsten light fixtures have ideal and consistent color quality vs most methods of lighting for 5600K which are prone to color spikes (HMI/FLUO), color rendering errors and metamerisms (LED), and inconsistencies in control over color balance (sunlight/skylight/daylit bounce). On the other hand one often doesn't have a choice between if some existing daylight or practical lighting has to be matched and often style becomes more important than the intricacies of the difference between working white balances and color rendering, since if you are careful with your lighting units its very possible to overcome most color rendering limitations at any color temperature.

My rule of thumb if you happen to have a choice and don't have to match color temperature in particular is I go for daylight where possible while being careful about color rendering and green/magenta spikes since daylight is easier to gel warm than tungsten is to gel cool due to the comparative transmission of those respective gels. And most digital sensors do better with daylight color balance since the blue channel is the most likely to get noisy and underexposed. It's especially bad for cameras with digital rather than analog whitebalance like RED cameras where shooting in 3200K and white balancing for it may require underexposing your blue channel by as much as 2 stops relative to the other channels and applying digital gain to compensate.
 
Last edited:
nyvz said:
My rule of thumb if you happen to have a choice and don't have to match color temperature in particular is I go for daylight where possible while being careful about color rendering and green/magenta spikes since daylight is easier to gel warm than tungsten is to gel cool due to the comparative transmission of those respective gels. .
Click to expand...

I may be mis-reading this but it sounds like you are suggesting to balance for 5600 vs 3200. However you are also pointing out (correctly) that CTO has improved transmission over CTB so it is more efficient/less light loss gelling daylight units with CTO than tungsten units with CTB--which doesn't really help as an argument to shoot 5600.
 
I'm not an engineer, and arguably still a novice. However, I notice that when I shoot compressed into a codec at 3200k, my blue channel gets a bit noisey...specially if also amplifying the overall signal through gain/ISO.

I think most digital sensors are natively 5600k. If I understand correctly, cameras have to amplify the blue channel quite a bit to achieve a normalized white point under 3200k lighting...which can introduce some amount of noise. Also, blue and violet wavelengths are much faster and smaller than the red wavelengths. This makes it harder to penetrate the photosite bucket, thus needing more sensitivity (which I think may be a big reason for native 5600k sensor design).

So the combination of smaller wavelengths that have a harder time raising photosite charge levels off of the noise floor, the fact that tungsten lights have very little blue output, and that the camera has to then amplify the blue channel to compensate for it may be the reason why I see some noise in the blue channel when I shoot at 3200k white balance.

While I love tungsten, specially when shooting film T stock, I tend to shoot mostly at 5500k when I can, and using color temperature creatively on set. So when I want a more yellow look within a scene, I will generally use Tungsten with a 1/2 CTB so that I still feed the blue sensor but I still have enough color separation to push the yellow even warmer in post. That means my color temperature range is between 6000k to 4300k while camera is set to 5500k.

I do agree with the points made about metamerism with daylight balanced sources. But I don't find that it prevents me from getting a good image with HMIs, Plasmas and LEDs.

I do wish I could afford to work with huuuuuge tungsten lights with full CTO so that I get that even spectrum without metamerisms and still capture at 5500k, however I never have been able to afford the power and sheer number of units it would take under some of the scenarios where sunlight is present.

So yeah, that is why I often try to push for 5500k capture.
 
Frankly, outside of the older Red sensors in underexposed portions of the frame, I've never encountered any grievous issues with either colour balance/light source on digital.

Almost all of the studio work I've done has been purely tungsten, and the cameras I've used for those shoots (everything from Reds, to Sonys to Arris) have been fine and dandy with it. And I say that as the pickiest guy I know (so far as colour and spectral quality is concerned).
 
Grug said:
Frankly, outside of the older Red sensors in underexposed portions of the frame, I've never encountered any grievous issues with either colour balance/light source on digital.

Almost all of the studio work I've done has been purely tungsten, and the cameras I've used for those shoots (everything from Reds, to Sonys to Arris) have been fine and dandy with it. And I say that as the pickiest guy I know (so far as colour and spectral quality is concerned).
Click to expand...

In the end, that is all that matters. I know a lot of people who share your opinion. It is just the approach that has been working for me so far. And not sure why, but I do see some blue-channel noise.
 
CharlesPapert said:
I may be mis-reading this but it sounds like you are suggesting to balance for 5600 vs 3200. However you are also pointing out (correctly) that CTO has improved transmission over CTB so it is more efficient/less light loss gelling daylight units with CTO than tungsten units with CTB--which doesn't really help as an argument to shoot 5600.
Click to expand...

I didn't necessarily intend it to be a straight argument for 5600K, just a consideration of the relevant issues. But it could be said that an argument for carrying 5600K lights is also an argument for balancing your camera for 5600K in that you want to carry the lights that are the most flexible/useful/versatile but try to keep it simple where possible which may mean shooting unfiltered at their native color balance where possible. It sounds like you are saying setting your camera at 3200K could make more sense given my argument since it is easier to get a wide variety of lighting technologies to that white balance given the efficiency of CTO over CTB, but from my point of view, in practice CTO is more efficient but hardly ideal for gelling most daylight sources given that it exacerbates color (usually green) spikes in HMI/FLUO/LED sources.

As I mentioned, so often there are bigger factors that determine these for us, unless you are shooting on a soundstage where my understanding is tungsten is still somewhat popular for its low cost and high color quality and easy routing/dimming. What a waste of electricity though...
 
Last edited:
Thanks for your thoughts and insights guys! Yes... a few variables to contend with for sure. However all the lights I have are 56K LED's rated at 85+. Yeah I purchased them a couple of years back before the current crop of 90 - 95+ CRI's were available (at reasonable prices). So I do use a 1/4 CTO and up to 1/2 CTO on occasion. That gives me supposedly 4.6K and 3.9K. With the rather short throw of my particular LED's combined with the CTO lumen falloff... not ideal... but I'm ok with the performance. A Fresnel type LED is on my list. Using the 1/4 CTO gel on 5.6K's does separate (subject from environment) and take the edge off for me nicely when needed during most interior shoots where incandescent lights or low color temp bulbs are prevalent. The aesthetics of lighting is always a factor. I'll be watching out for those pesky metamerisms :)

Noah, Grug and Rafael, I checked out your Demo Reels... All very different, all Fantastic!
 
You must log in or register to reply here.
Back
Top