Filters For Simulating Candlelight

Hello, maybe you can recommend me some lighting filters that work well on simulating candlelight using tungsten lamps.

I have been doing some tests with "Straw" filters from several manufacturers, but I don’t like the results very much.

I am currently working on a project that will be shot on HD.

Leandro Fessia
Gaffer
Cordoba, Argentina

>>maybe you can recommend me some lighting filters that work well on >>simulating candlelight using tungsten lamps.

I like to bounce tungsten light off of a board mounted with Roscoflex gold for firelight / candlelight / torchlight. It's not exactly a filter- and it takes a bit more work to control and shape than, say, a gel on the doors of your lamp. But I think that the effect is really nice. If it has to be a gel combination, you might try Chocolate (#99) mixed with a shade of yellow or straw, or even CTO... Just remember that the Chocolate gel really cuts down the output of the lamp.

Toby Birney
DP,

Based in Lithuania

Leandro Fessia wrote :

>>...recommend me some lighting filters that work well on simulating >>candlelight using tungsten lamps.

In this case, by seeking a filter solution, you express the interest in matching the colour appearance of candlelight, as opposed to its flickering characteristics. This is a situation where using mired values can help. Without getting into much of the background detail, mireds (micro reciprocal degrees) are calculated by dividing the colour temperature for each light source, in degrees Kelvin, into the number one, and multiplying the result by one million.

So the mired value for tungsten, at 3200 degrees Kelvin, is about 131. The mired value for candlelight, at about 1800 degrees Kelvin, is 555. The difference between them is 244 mireds, which is known as the mired shift.

Since the transformation from tungsten to candlelight is one of increasing "warmth" you will want to use the CTO range of lighting filters. The useful thing about mireds is that they are additive, so you can use any combination of filters whose individual mired shifts add up together to the total desired.

In this case, a 244 mired shift can be created fairly closely by combining a Full CTO (about 131 mireds) with a 3/4 CTO (about 112 mireds) which total about 243 mireds. I keep repeating the word "about" here since the actual mired shift for a CTO will vary somewhat depending on the manufacturer, but this assessment should get you very close.

Remember that for any such filtering effects to work, in either film or digital formats, you have to be recording using a base setting for 3200 Kelvin.

Ira Tiffen
Basking Ridge, NJ

Ira Tiffen wrote :

>>So the mired value for tungsten, at 3200 degrees Kelvin, is about 131.

Please change "131" to the correct value of "312" and the rest of the math works fine.

Ira Tiffen
Basking Ridge, NJ

Ira,

I think you have a typo

>So the mired value for tungsten, at 3200 degrees Kelvin, is about 312. >The mired value for candlelight, at about 1800 degrees Kelvin, is 555. >The difference between them is 244 mireds, which is known as the >mired shift.

This got me to wondering if in video you did a white balance with a CTB in the same mired amont,-244 would it give you the same results once you removed the filter?

Perhaps you know the answer to 2 other mired questions I have :

1) I thought mired reflexed colours on the red/blue line as Kelvin does. So why does -Green and +Green filters both have a + mired value?

2) Why do ND gels have a +mired value? Is it just because it's not a perfect world?

Thanks,

Steve Golden, DP
Chicago

Ira Tiffen wrote :

>>increasing "warmth" you will want to use the CTO range of lighting >>filters

I have done some tests, and there are several filters combinations that work well. But using only CTO filters doesn't get me the look I was looking for. The resulting light turns out very red, and I am looking for a more yellow/orange light.

Apart from the candlelight thing, searching for filters I realize that there are conversion filters that don’t go under the CTB or the CTO series, they were dyed with whole different colours than the usual CTB/CTO.

Someone can explain me how filters with different colours can provide the same mired shift? How the mired shift relates with the "dominant colour" of the filter?

Leandro Fessia
Gaffer
Cordoba, Argentina

Steve Golden wrote :

>>if in video you did a white balance with a CTB in the same mired >>amount, -244 would it give you the same results once you removed the >>filter? So why does -Green and +Green filters both have a + mired >>value? 2)Why do ND gels have a +mired value?"

White balance systems typically don't go beyond the range of about 3000 to 10,000 degrees Kelvin. That's a range of about 233 mired. Candlelight is lower and bright blue sky is higher.

Note that a filter producing a given mired shift alters colour temperature more at higher colour temperatures than at lower ones. A 21-mired filter produces a 200 degree change from 3000 to 3200 Kelvin, but also a 586 degree change from 5000 to 5586. It would only take an 8 mired shift to go from 5000 to
5200.

So it depends somewhat on where in the scale you are whether you can accomplish similar effects electronically as optically. The other issue is whether you are going beyond the limits of the WB circuitry. The closer to the center of the colour temperature range, and the smaller the change induced, the more closely will the WB simulate optical filtering.

As for plus/minus green filters having mired shifts, this would reflect only a part of what the filter is doing. A mired shift value here would be misleading as it would not tell you anything about the other changes to the spectral bias the filter creates.

And you're right about ND's not having mired shifts in a perfect world. Which it isn't, but some filter manufacturers get closer than others...

Ira Tiffen
Basking Ridge, NJ

Leandro Fessia wrote :

>>But using only CTO filters doesn't get me the look I was looking for. >>Someone can explain me how filters with different colours can >>provide the same mired shift?"

If the world were as definable as to allow us to say that there is only one way to create the look of candlelight, and that it will work perfectly for everyone in all situations, then it would be a very different place. And very boring. Colour temperature matching is only one of many considerations. Ultimately your artistic sense is the final judge.

Mired shifts related to anything other than filters specifically designed to produce ONLY mired shifts tell only part of the story, and should not be considered comparable. Its as if to say that if there are four wheels on a car, the ride will be the same for all cars. Of course that's not true. A filter that
produces a similar mired shift as another, but that's otherwise doing different things to colour, will produce obviously different results. You need to know more about those other things its doing.

Note that the standard conversion and correction filters, like the Wratten 81 and 85 (reddish-warming) series and 80 and 82 (bluish-cooling) series were originally designed to accomplish what film needed to be properly corrected in the specific situation of going from one particular starting colour temperature to another. Some customisation was done to accomplish this. Filters designed more generically for specific mired shifts can be configured more readily for a wider range (almost infinite) of CT changes. The coloration of reddish and bluish filters is similar, but does differ among the two types accordingly.

It could theoretically be said that for every film or imaging chip of different colour sensitivity, you should create a different conversion filter, but in reality, that hasn't been necessary.

Ira Tiffen
Basking Ridge, NJ

>>In this case, a 244 mired shift can be created fairly closely by >>combining a Full CTO (about 131 mireds) with a 3/4 CTO (about 112 >>mireds)

Although this theoretically will give you the exact Colour Temperature the candle light gives, aesthetically it will be too much red. Make tests with stills, see how much CTO suits your needs and your taste. I usually use 1/2 CTO on tungsten lamps plus a dimmer for the flicker effect (the dimmer will add red as it goes down). There are plenty of different effects to simulate candle flickering and they have been posted in this list before.

I just like to keep it simple.

Cheers!

Agustin Barrutia,
Gaffer/Electrician,
Buenos Aires, Argentina.

Leandro, I missed your last post where you say that you’ve already done tests with CTO.

Another option if you are looking for a yellow/orange light could be using CTS (Colour Temperature Straw) filters. Or CTO plus an amber filter of your choice. There are plenty of options on different types of yellow filters (with brown, orange, pink tints). Chocolate, Straw, Amber, Coral, Sepia, Apricot. I’ve read about this last one in an AC article some years ago.

Take a look at this list :

http://www.rosco.com/us/filters/roscolux.asp.

Let us know what you decide and which flicker effect you’ll use.

I’m not related to Rosco filters.

Cheers

Agustin Barrutia,
Gaffer/Electrician,
Buenos Aires, Argentina.

To me there is nothing like a real good old type A household low wattage lamp on a dimmer. Gives about as much of a candle glow as I've ever seen any gel do.

--
Disclaimer :

My opinions, thoughts, and beliefs are my own and may not reflect yours. The use of the pronouns "you, "some", and "many" to name a few are generalizations and without a proper name attached to them are not references to anyone reading my posts.

Walter Graff
Director
BlueSky Media, Inc.

I prefer CTS to CTO, but as Augustin says you don't have to stop there...

This is simply a question of what says candlelight to you I think; even if CTO & mired values might literally get you there, our eyes/brains are not so "lin" ---

Did Rembrandt use a Minolta Colour temperature meter ? Hell I think he WAS a CT meter - with some AI built in (sorry Chris I'm stealing your line now and in the future too.

Sam Wells
film/.../nj

Creating Candle Light – CML Posting of 2/7/2006

If you want convincing candle light, you need more than just filters. Perhaps that's why you've not been completely satisfied with the result. For starters, the flame isn't a uniform colour. The colours range from blue to red-orange depending on which portion of the flame you're looking at. These colours vary slightly with the guttering of the lit wick.

Even the shadows cast from a candle "dance" as the candle responds to movement of the air. To be more convincing, you might explore how the candle light changes if a person out of frame crosses the room. The shadows lengthen or shorten in response to the growing or shrinking of the flame. Perhaps part of "selling" the effect is recreating these subtle changes caused by the flicker of the light.

The artificial source will be more believable if it contains some of the variations that exist in real candle light. That means you need not just random flickering, but motivated variation. Not just a single piece of gel on all the created light sources, but broken up bits of other shades. Not just a stable artificial source, but dynamic within an expected range of movement to mimic the flickering of a real flame. These elements are more important to creating a convincing candle-lit effect than a correcting gel alone.

Bruce Aleksander
ABC / Disney
Houston