Exercise: Tungsten and fluorescent lighting.

This exercise looks further at white balance and how artificial light needs extra attention. 

The first part of the exercise shows how our eyes adapt to the different colours of light, but that in fact, as we saw with daylight/shade/low sun, the difference between the different types of light can be quite significant. In the first part of the exercise I stood in my kitchen (which is lit with tungsten halogen lamps) at dusk and allowed my eyes to adjust to the light outside. Once they had done so I turned to look into the kitchen and noticed that the colour of the light was quite yellow/orange. Once my eyes adjusted to this light I turned to look back outside and noticed that the light there was very blue.

Measuring the light levels in the kitchen, with an aperture of 3.5 (which was the widest on the lens I was using), at ISO 100 I needed 1.3secs in the darker areas of the room and 1/13th sec close to the lamps themselves. This I believe is around 4 stops difference. Clearly hand-holding the camera wouldn't have been ideal in these shooting conditions.


After measuring the light levels, I had to set up a scene where I could shoot part indoor (lit by tungsten lamps) and part outdoor, at the same time - but with the light levels being approximately the same. I used a hand-held light meter to check the measurements. Outdoors was actually slightly lighter than indoors, although this isn't apparent in the photographs. The first shot was to be taken with the white balance set at auto, the second at Daylight and the third at Tungsten.

Auto white-balance setting. The camera has chosen a white balance setting which it thinks is correct for the scene. The indoors looks quite orange and the outdoors quite blue.

Daylight white-balance setting.With this setting, the indoors has been adjusted to be slightly more orange and the outdoors slightly less blue.

Tungsten white-balance setting. In this setting, the indoors is slightly less orange but the outdoors is far too blue.

What this exercise shows is how the camera can detect the overall colour cast and adjust the white balance accordingly; but clearly it can only do so much with a scene that has two separate and distinct light sources. The daylight setting corrects the blueness of the outdoor scene but makes the indoors seem too orange, whereas the Tungsten white balance corrects the orange tinge of the indoors but increases the blue tinge outdoors. On balance the auto setting is the probably the most accurate option.


In this final part of the exercise we look at how fluorescent light affects the colours in an image and how the in-camera colour correction deals with it.

I have taken three images below, the first with auto white balance, and the second and third with a fluorescent white balance. The first fluorescent is 'cool-white' and the second is 'warm-white'.

Room lit by CFLs. White Balance: Auto

White Balance: Cool White Fluorescent

White Balance: Warm White Fluorescent

It is clear, looking at the three images above, that there are problems with at least two of them. In particular it is clear that the Warm White Fluorescent in-camera setting is not right for the CFLs being used, as there is a very clear green tinge. The second image is also incorrect as there the image is now slightly too warm (red). the Auto setting is again probably the most accurate. 

This exercise was repeated in a different location - thistime lit by a 2D fluorescent ceiling light.

2D Fluorescent Lamp. White Balance: Auto

White Balance: Cool White Fluorescent

White Balance: Warm White Fluorescent

The colour tints are clearer in this image. The Warm White Fluorescent setting hasn't corrected enough for the lack of blue. The Cool White Fluorescent setting has overdone it with the red (trying to compensate for the expected 'coolness') and like the Warm White setting is still short on blue. Again, the auto does a good job of fairly faithfully reading the situation and producing a pretty accurate colour balance - it does this by adding sufficient blue to the mix. Interestingly the green component remains almost the same in each image.