Color calibration for DT under halogen quartz lamp 3200K light

Nowadays: what else is available?
You either find LEDs or halogens, and
of those, halogens seem to have much
better color rendering properties…

Exactly. As far as I know halogens are still better than the best flash light. I’m not aware of LEDs that are in pair of halogens as well. May be I’m wrong but I think halogens still rock.

I have heard about better tungsten bulbs for video with 5500K temp. But they have some drawbacks for home use

My mistake. Not 5500K but 6000K. Movie makers used to use HMI bulbs in the past. Not sure about them now. But HMI should be even better. Its spectrogram is identical to sun’s. But HMI is inconvenient at home.

and you did measure your targets ( to build a camera profile ) with spectrophotometer with POL filter on it ?!

No I didn’t. I don’t have a spectrophotometer. You’re right. It would be an interesting experiment.

The polarized light and camera filter are responsible for making dark patches of the color target darker because of lack of light reflections on them. And what I had to do in DT to achieve better results in color calibration module was to decrease the level of black before doing any color optimizations. Look at these 2 pics. The second one gives me much better delta E in calibration module

Sure if I’d have access to a spectrophotometer and managed to put the exact same b+w polarizing filter on it I would’ve had better results. But are there any way to do this in this setup? Are there any specrophotometers that may allow you to measure through a third party filter using a third party light?

it is not an experiment - it is what you do IF you attempt at reproduction quality … and IF you do indeed you also try to make a target with similar oils and measure that target and incorporate it in your repro camera profile.

i1Pro3 Plus ( Basic ) model / pol filter included / prescribed (3rd gen does not have tungsten vapor induced drift as the light source is halogen no more)

yes, I understand the idea and considered this device. However I’m a bit skeptical about polarized filter inside the device. They are all different. The filter I put on my camera may have different color reproduction. As well as I put filters on my tungsten bulbs and that gives me another color shift. If only could get colors by measuring through the same filters…

But I also really doubt about oil paints. They look different under different light sources. Also one artist may use different combinations of paints to get the same color. Blending different pigments of different manufacturers will give you different colors. And adding some thinner or lacquer will change everything again.

Still worth experimenting? I doubt. Anyway the same oil picture looks a bit differently under miscellaneous light sources in gallery.

I already have great results with WB, Color checkers and DT. I almost do not see any difference between artwork and its photo.

well , if you aim is just to satisfy “your eyes” ( almost ) then case closed … better be happy than to sweat about dE-whatever - it hurts your wallet and can descend into an obscession

Except introducing one another level of complexity I do not see how spectrometer like i1Pro3 Plus may help with the setup like that simply because it isn’t designed for measuring custom light setups.

@SpaceDreamer do yourself a favor and get some CTB (convert to blue) filters for those Halogen lights. 3200K Halogen has a nice spectrum but your camera sensor is not being used to their full extent.

For example: 201 Full C.T. Blue - LEE Filters

Thank you. That is an interesting idea. I did some research but I didn’t find any proof that using filters like 80A put either on camera or on light might give any improvements for a digital sensor.

I understand the physics of the process. The 80A filter cuts some red and adds some blue. Sounds good but will it help a modern camera to improve the color reproduction? Will digital sensors capture more information with the 80A filter than they can capture without it in a 3200K setup?

Can you give me a link with some more information about it?


The second part of these experiments might interest you: Here is why WB is not enough

Also, Rosco and Lee filters have published quite a lot of data on the Web
regarding the nobel art of changing light temperatures.

Have fun!
Claes in Lund, Sweden

I don’t have a link ready, but a look at the raw histogram will tell the story.

Halogen lights have a very continuous spectrum, which makes conversions easy and without much of a quality loss. You lose some exposure but since your models are static that should not be a problem.

From experience I can say that filtering light sources gives better quality than filtering the lens.

1 Like

As far as I understand white balancing and then applying optimization in DT using color checkers should solve any problems with color shift if you have all colors/data in your RAW file.

If you don’t have you won’t be able to recover your colors. IMHO the main question is ‘does a modern digital sensor see all colors that it’s cable to see when the scene is lit with the spectrum of sun and with the spectrum of a 3200K tungsten bulb’. If the sensor’s sensitivity allows it to capture the same amount of data in both cases we do not need to bother playing with any filters at all. We just need to adjust every channel to get the exact same result. Am I right?

i believe we’re talking about colors fallen outside the gamut of the camera sensor (or it’s hardware sensitivity to the whole spectrum).

Let me rephrase the main question. ‘Do our sensors lose some colors when we illuminate the same scene with either sun light or tungsten light assuming they both have uniform spectrograms with red color prevailing in tungsten setup

The simple answer is yes.

In an ideal world you want neutral white light - where all pixels can receive the full information and then you can use it. Now consider a light that has a strong bias into a single channel, for example tungsten into red. To use the dynamic range on the green and blue channel you would have to blow the red channel completely. If you expose correctly for the red channel the green and blue will only use a fraction of their possible dynamic range and therefore your image will suffer in color fidelity.

before and after white balance:

R ===========================|
G ===========================|
B ===========================|
before whitebalance:

R ===========================|
G ==============|
B ==============|

after whitebalance:

R ===========================|
G = = = = = = = = = = = = = =|
B = = = = = = = = = = = = = =|
1 Like

it can measure spectrum of your illuminant of course … it has a diffuser for ambient light measurement.

provided that you don’t melt the filter in the process if your light source is a powerful halogen light

it helps -for example - with S/N if you do just one shot of a target and have dark blue patches under tungsten illuminant and can’t give them good exposure to preserve some other patches from clipping in raw … plus it is better to have a single illuminant profile to be done under light somewhat closer to D50

Well, self-responsiblity can be called on a lot of things.

“The problem with common sense: it’s not so common.”

Yes, you’re right. It was my mistake. It is a good idea to use a spectrophotometer with a diffuser feature to find out the truth.

I don’t have any problems with my polarizing film filters put in front of the two 1kW halogen bulbs since I have glass between polarizing film and halogen. Also I installed small but effective 12V turbine fans from a server to cool down the halogens. To make a new photo I switch the bulbs for not more than just 5-7 seconds. So now I can put any gel filters in front of the halogens

Here within this DT forum we’re talking about software and post processing but my main concern is rather hardware. Even with the best algorithms we wouldn’t have our image improved if we lost some information on the hardware level. And we do loose some CRI when putting CTB filters (up to 10%).