Part 1: take a photograph in which both the interior lit by tungsten lamps and the exterior at dusk are both visible. Wait until the light levels inside and outside are approximately equal, and take three photographs, as follows: with the white balance set to Auto, with the balance set to daylight, and with the balance set to tungsten. Compare the results and note the differences.
Part 2: find two different interiors lit by fluorescent lamps. Take two or three photographs, identically composed, in each location. The first image should be with the white balance set to Auto, the second to fluorescent, and if there is a choice of different fluorescent settings, the third to the alternative fluorescent.
Part 1: tungsten and daylight
As recommended I waited until the outside daylight had weakened enough around dusk. Taking sample meterings from around the room there was a clear variation in shutter speeds needed to capture an image at f/2.8 and base ISO (200). Evidently the eyes do compensate for different levels and qualities of light as this variation wasn’t clear to the naked eye.
The results were startling. The scene seemed to be well-balanced to my own eyes, with no strong colour tint to the light either inside or outside the room. However, the Auto WB version (3850K) tries to cope with the different types of light and ends up with a compromise that simultaneously makes the outside light slightly blue and the inside light slightly orange. The second shot, using Daylight WB (4900K) renders the view outside in natural-looking colours but increases the orange tone to the interior. Conversely, the Tungsten WB version (2850K) makes the room look quite natural yet throws the outside scene into an extremely strong blue tint. None of these looked true to life.
Part 2: fluorescent
First, a scene in the main thoroughfare of a shopping mall. My camera has three Fluorescent WB settings so I tried them all.
- For the Auto WB version my camera selected 4700K and this looked pretty close to how my eyes saw the scene
- Fluorescent 1 gave colour temperature of 6550K and looked too warm, too orangey – overcompensated
- Fluorescent 2 gave colour temperature of 5150K and still looked too warmed up
- Fluorescent 3 gave colour temperature of 4450K and looked a little too blue
In this set, the Auto WB actually looks most natural.
Secondly, a scene inside a supermarket.
- Auto WB gave colour temperature of 4000K and this looked close to how my eyes saw the scene
- Fluorescent 1 gave 6550K and looked too orangey
- Fluorescent 2 gave 5150K and still looked too warm
- Fluorescent 3 gave 4450K and this looked better, but still a little ‘off’ – slight red tinge
So again in this set, the Auto WB looks most natural, albeit at a lower colour temperature than the previous image. I conclude that my camera is pretty good at selecting the right WB for the situation, but two of its three Fluorescent presets were significantly off, for both of these particular locations.
What I’ve learned:
The exercise has really brought home something that I’ve understood in theory for a while, namely that the human eye/brain can quickly adjust to different lighting conditions and will normally refuse to acknowledge the wide variations in light colour and strength that the camera sensor will clearly reveal.
The tungsten/daylight shots at different colour temperatures were so significantly different that it took me by surprise. I now have a much better appreciation of the differences in light that my eyes can’t detect but the camera can.
The wide variation in fluorescent lighting, and the fundamental problem of their limited colour spectrum, is something that again I was aware of in theory but has really come to life here. I now understand why my camera has three Fluorescent settings! I’m glad that I always shoot RAW+JPG these days, so that if needed I can reprocess the RAW with an appropriate WB adjustment. I will be particularly wary of shooting in fluorescent lighting situations in future.