Lighting with fire & IR

[Michael Collier]

So I am starting this thread to share an effect I came across on a film I shot recently. We were doing a campfire scene, and as we got more and more into the closeups, I started to light with actual fire more and more.

 

The obvious bit I found-lighting with fire can be a hell of a lot of fun! Either having someone squeeze lighter fluid on the fire, or placing a coke can with the top cut off, half full of fluid really gives you the feeling of fire. Of course-it is fire.

 

However there was one thing I could have probably remembered from science class, I didn't discover until the grade: most of fires heat is given off in the form of IR radiation. This causes the reds to saturate, and shift the overall hue of the light.

 

In wides, I was lighting with 6- 212 & 211's aranged behind the fire, in a triangle made of 20" C-arms. In closeups (especially with our bespecaled charecter) we started going more and more to fire. In the color grade, that led to a VERY saturated red taking over skin tones. I corrected that with a 3deg hue shift, which to me made sense. Afterall, the extra IR would be shifting the hue outside of the visible range into the red, so a hue shift would bring it back. I tried doing it on the wheels, but the hue shift felt better to me.

 

I thought I would pass along that discovery in the hopes that it someday helps someone out. Obviously if you are doing a shot with propane gas and concrete logs, it won't be as big of an issue, since it seems most of the heat of a fire (and thus IR) comes from the coals. But if you are doing a real wood fire, maybe consider adding a hot mirror to the camera package. I would have never thought to include that in the package, but I think from now on, if I see fire as a major light source, I will be using a hot mirror.

 

I can't post pictures until the film comes out, unfortunately, but maybe once it does I can share some before and after grabs.

[David Mullen ASC]

There could be more IR contamination but that would not necessarily make red faces go redder, the lower Kelvin of the real flames are more likely the culprit. Also, exposure is an issue too, a very reddish fire would look less saturated if exposed brighter.

 

You didn't say what gels you used on the 212's and 211's to match the color of the flames or if you took a color temp meter reading of the fire itself to check what it was naturally, but I think a lot of people are surprised at just how low in Kelvin a real flame can be.

[Michael Collier]

It certainly could have been that. The photofloods were ungelled, but dimmed down to almost nothing. The match between fire and artificial seemed to balance to eye and on monitor, and between MED and CU was constant. As the night went on, there was a marked change in the fill light when we went into the CUs, even though both were lit by only fire. The red shift wasn't so dramatic that I saw it during the night, but it was fairly apparent in the grade. It was much easier to see in the pictures I can't post.

 

Our light (on mids and CUs at least) remained constant, how we fluffed the fire remained constant, it was true dark for all of it. Exposure remained constant, the only thing I can point to that gave such a dramatic difference was the coals. They don't give off any appriciable light given our light level and relatively low ISO, so IR is all I can pinpoint. After all the IR polution tests I have done, I wouldn't expect IR to shift skin red either. But if that isn't the case, I am at a loss to explain the change.

 

My sense during the grade was there was something odd going on. When there was variation between kelvin in visible light (ratio of fire to photoflood), that was simple enough to grade away using the gain wheel. But with the later CUs it took a radical amount of correction on the wheels. The direction I was moving the wheel in while watching the skin tone gratical, was markedly more in the yellow direction than I had ever done before. Usually the wheel naturally goes more towards orange. Something felt very, very weird it almost felt like the light was non-kelvinic. which makes even less sense because both the fire and the photofloods, even if they dont match are definitely on the kelvin scale.

 

If I applied the medium shot grade to the CUs, and applied a very small hue shift, everything fell into line. It was very odd, since I haven't ever seen that happen, and have never had a normal use for hue.

 

 

 

(Sidenote-there was a long period of time between Med and CU due to an unfortunate moose visitor, allowing lots of extra coals to build up)

Edited October 5, 2015 by Michael Collier