Flash Color Variations
When working with flash, one has to be aware that the color of the burst might vary slightly as the power is changed. Common wisdom has it that the more you decrease the power of the flash, the more it warms up (with higher end units typically maintaining more consistency than cheaper ones). I had never really bothered to check how obvious the variation was, so I did a little comparison.
Naturally, this is not an issue when there is a single light (or when all lights are identical in every way), since white balance will fix the cast. It only might become a problem when using multiple lights at various intensities simultaneously.
Note: This is not the strictest test in that the aperture varied (obviously), I used a 4-stop ND filter (which tends to add some red) to be able to shoot at maximum power, the light might have bounced off stuff in the location I did the test, etc., but the same camera/lens/filter/location was used in every shot, and all images use the same base white balance setting. What's important is that the relative variations between the shots are fairly accurate.
For the first series, I used a 1200Ws studio strobe (a Lightrein LR12DR), starting at full power and decreasing by a stop until minimum power (a 7-stop variation between both extremes).
A 1200Ws studio strobe fired at various powers. RGB color variations are relative to the top left shot.
Everything else being normalized (I used a "Blur => Average" filter to determine the variations), we note that there is indeed a slight variation. The outlier in this test seems to be the 1/1 shot, with all the other ones being pretty close indeed. Moreover, keep in mind that it is unlikely that far extremes would be used in the same shot—when do you have a 7-stop variation between two lights (a 1:128 ratio) in a single shot?
Although there is a difference, it won't usually be something to worry about, in my opinion.
For the second series, I used a flash gun (a Canon Speedlite 580EX II).
A flash gun fired at various powers. RGB color variations are relative to the top left shot.
There are a number of lessons to be learned from this second series. First, the color variations between the powers are still not outrageous, albeit ever so slightly worse than with the studio strobe used in the first series. Second, in this case, the lower the power, the cooler the color—the opposite of what one might expect. Third (and more importantly), the studio strobe was significantly warmer than the Speedlite, regardless of the slight variations caused by different powers.
At the end of this test, I would conclude that if I only use studio strobes, or only flash guns, in a single image, I shouldn't worry too much about color variations stemming from power variations. However, if I intend to mix studio strobes with flash guns in the same image (which I sometimes do), I might want to add some gels to the lights to bring them closer. I believe that the difference is enough to be noticeable in many situations.
Unfortunately, the discussion cannot stop there, because the power of the flash is not the only factor that can alter the color of the light. Indeed, most of the times, lights will be modulated by some kind of modifier, such as a softbox or umbrella.
In the third series, I fired the same 1200Ws studio strobe used in the first series, at full power, only this time I used a collection of light modifiers. Some of my modifiers are newer than others and are of a number of different brands—a fairly typical situation. The top left shot is the same as the top left shot from the first series, for reference.
A 1200Ws studio strobe fired using a number of light modifiers. RGB color variations are relative to the top left shot.
From this last test, we notice very distinctly that the color is thrown in every direction from one modifier to the other—even two similar 45" umbrellas are dramatically different, adding a grid to a reflector added some cyan, adding a baffle to a beauty dish added some blue, an older softbox is significantly warmer, whereas a newer one from a different brand is much cooler.
Frankly, the color variations introduced by the use of various light modifiers is such that it renders the problem of color variations from different powers moot—don't be impressed by that detail when deciding which brand of studio strobe to buy. In my experience, the difference in color between studio strobes and flash guns is quite considerable and should be dealt with, but other differences have gone largely unnoticed!
Unless I'm doing color-critical work (reproduction, products with exact color specifications, etc.), I wouldn't worry unless I mix and match big lights with small lights.