If shooting flash indoors with mixed ambient lighting, you have to pick one of the two choices about white balance... We cannot mix two colors of lights (flash with incandescent or fluorescent), because any compromise white balance attempt would simply be wrong for both lights. So do you want the room lighting to contribute into your picture, or do you want to minimize the room lighting white balance problem?
If you don't want the continuous room light messing with the white balance in your flash picture (easy way - the subject of a previous page), then simply use low ISO and the maximum shutter sync speed to keep it out, so Flash White Balance can be used (high ISO indoors boosts the orange incandescent ambient, but reduces the necessary flash power, emphasizing the orange). Faster shutter speed does not affect or reduce the flash, but it does affect and reduce the continuous ambient. 1/200 second or so will really knock down the ambient in most cases indoors. A lower ISO, or a smaller diameter aperture (f/5.6 or f/8 instead of f/2.8) will increase this effect (this ambient reduction at 1/200 second), but that will require more flash power, which must be available. Studio lighting will always use low ISO AND maximum shutter sync speed to keep the continuous ambient out of the controlled environment. Even the 150 watt modeling lights in the close flash units cannot survive f/8 at 1/200 second at ISO 100 (we should verify that the same picture exposure without the flashes does come out totally black).
If you do want the room light to contribute in your picture (assuming it is bright enough, or ISO is high enough), then incandescent lights are orange, and fluorescent lights may be green, while the flash is daylight (white). Sometimes we like a mild incandescent orange warming, but more can be ghastly too. You cannot process with two White Balance values for mixed lighting, you have to choose one. Meaning, we have to make the light colors match. We can use colored filters on the flash to convert the flash to be orange or green too, and then set camera white balance for incandescent or fluorescent, to match the ambient lighting. If you are using high ISO (or Auto ISO) to balance indoor lighting with flash, then keep reading below, you definitely need a filter on your flash, to be able to balance color too.
Said again - to combine orange incandescent room light with daylight flash: We can put a CTO filter (Color Temperature Orange) on the flash head, to make its light orange too, to match the incandescent room light. Then we can properly set Incandescent White Balance in the camera, or later in Raw processing.
Below are a few Rosco color filters from the (near free) Roscolux Swatch Book - and there are other choices too. The significance is 1) these are sized to fit a speedlight as is (attach with masking tape), 2) they are near free, and 3) a vast variety of choices are provided. Descriptions and spectrums of these swatch books are at Roscolux and Cinegel. These Rosco Swatch books may be available fully free from your local camera store if you ask about them... they are intended as free samples, but shipping charges complicate it. These swatch book samples include almost every filter (the book is about two inches thick), and are suitable for taping on a speedlight — just not directly touching the flash lens (see below). The Rosco Cinegel Swatch book is "only" 1 3/8 inch thick, but it contains dozens of these same standard filters too. Mine are 3.25x1.5 inches size, suitable for direct use on a speedlight. The color filters are polycarbonate or polyester plastic today, but they are still commonly called gel filters, or gels, from when gelatin was used.
Note: the text printing shown below is on a small white paper separator that comes between each clear filter in the swatch book.
From the Rosco spec page (search for numbers like #3304 as R3304),
CTO, 5500K to 2900K, 47% transmission (~1 EV light loss)
3/4 CTO, 5500K to 3200K, 58% transmission (~0.8 EV loss)
1/2 CTO, 5500K to 3800K, 73% transmission (~0.45 EV loss)
1/4 CTO, 5500K to 4700K, 81% transmission (~0.3 EV loss)
1/8 CTO, 5500K to 4900K, 92% transmission (~0.12 EV loss)
TTL will meter the filtered flash properly, but the flash will need this additional power.
Nikon furnishes filter sets, see Nikon SJ filter sets. The SJ-3 looks like it should work with most speedlights (held in the pullout card slot). Mounted similarly, the red filter on bottom flash here is from SJ-1. My best guess is that the Nikon filter numbers approximately match Rosco this way:
This match is close, but we don't know the color of our lamps anyway, lights vary too.
Rosco also packages their popular filters (55 filters in 20 colors) into a Rosco Strobist Collection It includes multiple copies of several of the useful colors. Size is 5.5 x 1.5 inch (38.1 x 139.7 mm). This longer strip can be held in place with a simple rubber band around the speedlight flash head. You probably want to add a bit of tape marked with the filter color too. Looks like a bargain. That Rosco data sheet is here.
These filters also come in larger sheet sizes for professional use (stage lighting, Hollywood, etc), like 20x24 inches for around $7 each, but the Rosco Swatch Book offers many choices to try them and see, also suitable for speedlights.
The CTO filter from the Rosco Swatch book is shown here on top flash on a SB-800 speedlight, simply using masking tape to attach the filter to the flash. But with a slight bow, so it is not touching the lens, so it does not melt at full power. Shown on bottom flash is a Nikon Red filter, which has been a little too close to a full power flash at time or two. Doesn't affect its light, but just a slight air spacing is surely better (bow it slightly, as shown on CTO Rosco at top). After you figure out what filters you may need, then you can do a more elegant mounting if desired — or not, masking tape works fine too.
Said again, the speedlight lens gets hot at high power levels. The flash power is intense for an instant. A speedlight flash fired when face down on the couch WILL SCORCH the fabric seriously (so be careful with it). Planning a mm or two of air space between flash head and filter is a very good thing. If the filter is attached bent around each side of the flash head, its natural bend curve probably suffices to hold it slightly distant.
You can use the filters to change your flash to match the ambient light, to set white balance for the ambient. Or you can use the filter to change the color of your portrait background (medium gray backgrounds work best for this). I have used the filters on the flash to add color into water drop splashes. You could for example use a 1/4 CTO filter on the flash with regular Flash White Balance just for a slightly warmer tone in the picture, but the same thing can also be done with just a slightly warmer White Balance setting in the later processing software (it's good to see things first, before committing).
Incandescent: Any one specific room lamp can be one of several shades of color (Soft White, Warm White, Daylight, etc), so these conversion filters come in several shades too. Selection may require some testing to match your specific lights (but close is better than none). These orange filters are named CTO (Color Temperature Orange) to match flash to various incandescent lamps. Incandescent lamp types vary, but start with 3/4 CTO to convert flash or daylight to match the lamps (and then use Incandescent White Balance). This CTO filter on the flash makes it be orange to match the incandescent lights, and then Incandescent White Balance in the camera works (any daylight coming in from windows will then appear blue). The CTS set (Color Temperature Straw) are sometimes used instead of CTO, being slightly less red, considered possibly better for human skin.
These filter densities can be stacked, for example, stacking 1/2 CTO + 1/4 CTO makes 3/4 CTO, which is why they are named this way. These filters reduce the effective flash power somewhat, perhaps 0.3 EV to 1.0 EV (Transmission 50% is a one stop loss).
The opposite of CTO filters is CTB (Color Temperature Blue). CTB filters can go on the lights themselves to convert tungsten lights to match daylight white balance. We used to have blue flash bulbs to use with Daylight color film (but the photo lab also did a pretty good job of fixing it if we didn't). The terms tungsten and incandescent mean the same thing regarding white balance, but there are differing colors of these lamps.
Fluorescent lamps have a non-continuous spectrum, hard to match completely. The actual internal gas spectrum has ultraviolet and blue lines of mercury similar to mercury vapor lights, but then the phosphorus coating glows to re-emit better controlled color. Fluorescent typically has a green component, so PlusGreen filters (above) are used on the flash to match flash to various common fluorescent lamps (so fluorescent white balance can be used). Colors of different types are:
Shutter speed issues with flicker: Fluorescent lights are continuously on, except they are powered with AC voltage. When using the old magnetic ballast, they flicker with the AC line frequency. Flicker is at twice the line frequency. Twice per cycle of 60 Hz in North America (120 times a second), 50 Hz in many other places. The 60 Hz AC voltage swings plus and minus, passing through zero 120 times a second. This is a big problem for shutter speed duration, which can randomly sample partial cycles and see varying color shot to shot (often this problem randomly sees a strong brown cast). I am speaking 60 Hz, but the solution is in setting the shutter speed to match the AC line frequency, either to 1/60 second, or double at 1/120 second (in North America, or 1/50 or double at 1/100 second in 50 Hz locations) helps to see only complete cycles for more stable color. Those are not the actual exact precise shutter speed values, but it seems close enough, and is best try. And slower multiples of 1/60 second (specifically like 1/30, 1/20, 1/15, 1/10 second) are even better to capture multiples of complete cycles of flicker, but faster shutters, like 1/80 or 1/100 or 1/200 second would be a bad match, capturing incomplete flicker cycles, affecting color and brightness.
However, modern electronic ballast (since 1990's, BUT magnetic ballast is still available too) has changed this situation greatly today, which converts fluorescent operation to about 20,000 Hz — for no flicker (in newer light systems). CFL bulbs, and desk lamps too, use the electronic ballast. And many new ceiling light installations too, but the old magnetic ballast is still available. I recently replaced an ordinary two 40-watt tube fixture, and the 25 year old ballast was magnetic, and the new ballast was electronic (and much smaller).
You can determine if your fluorescents do flicker, this way: For this, do Not use Auto WB. Shutter speeds of 1/60 or 1/120 second should match the cycles, but you are seeking flicker, so any intentional faster shutter would intentionally capture incomplete cycles of the flicker. 1/200 second shutter at 60 Hz captures 120/200 = 48% of one incomplete cycle for old magnetic ballast, incomplete color, which varies randomly from shot to shot. Or 1/200 second shutter for 50 Hz lights captures 50% of a cycle. But for newer electronic ballast, 1/200 second captures very many complete whole cycles, more complete color. You can aim the camera directly at the lamp for this test, for brighter light to achieve the faster shutter.
Take several, 5 or 6 all the same picture, without Auto WB to mess it up. If results vary dramatically (a few are randomly brown), that's flicker. If all of several results are the same color, then no flicker. If all the same, but color is off, that's just White Balance, not flicker.
If using a focal plane shutter, you might see bands of brighter and dimmer light, which are flicker variations during the shutter duration.
But regarding White Balance... Even if the lamp is called "Daylight", no fluorescent has a continuous color spectrum, which makes matching White Balance be a special problem, an exact match is never possible. Fluorescent lamps have a CRI rating (Color Rendering Index). See the spectrum test on this link, shown by single slit diffraction — discrete lines vs. continuous spectrum. When choosing new bulbs, a high rating (CRI 80+) has better color — not perfect, but better, often nearly acceptable. But a lower CRI is poor for color photography (and poor in the closet where your wife selects her clothing — she wants high CRI too, CRI 80+). Perhaps many colors look OK, but the colors not in its spectrum won't look right. In contrast, incandescent lamps do have a continuous spectrum and they are the definition of theoretical maximum CRI 100. Incandescent may be orange, but it can easily be matched, all of the spectrum is present. Direct sunshine is continuous too, but fluorescent and LED lights simply are not continuous spectrum.
Sodium vapor and Mercury vapor lamps have similar flicker problems, and much lower CRI ratings than fluorescent bulbs. Saying, you may see this flicker effect shooting under football stadium lights too if you can manage a faster shutter speed.
The whole idea of using the filter on the flash is to change the color of the flash to match the room lights, and then to set the camera white balance to match the room lights.
Alternately, if instead setting white balance to match the flash (no filter used), flash tubes are generally the same as daylight color, and hopefully overpowers the dim room light. But speedlight color varies significantly with the power level used, more red at high power (far or bounce), and more blue at low power (close or direct). In the Nikon and Canon systems, the speedlight knows the power it actually used, and if on the hot shoe, and if Auto White Balance is selected (allowing WB value to be changed), it can report an better estimated color temperature to the camera (called "Flash Color Temperature Information"). This is a very special deal (hot shoe flash can totally change the meaning of Auto White Balance — but only works in TTL BL mode), and this information can be used for a more accurate White Balance temperature. Whereas, if you instead set Flash White Balance, that standard guess is what you get, but it may need slight correction for any picture. Still close maybe, but the flash color does vary with power level.
However (just my opinion), otherwise, other than with the system's hot shoe speedlight flash, Auto White Balance is a mixed bag... it tries to analyze the picture results, and shifts the three RGB channels to have equal brightness levels. That may or may not be correct, it depends on chance (scene content, light color, etc). So then (without any Flash Color Temperature Information reported), Flash White Balance seems better, at least it knows the light should be generic flash color (which is about the same as daylight).
The background shown here is Savage #27 Thunder Gray seamless roll paper. You can change its brightness with the background light power level or distance (anywhere from white to black), or radically change its color with a colored filter on the background light. A red filter shown used here.
You can create colored backgrounds with a colored filter on the flash that is illuminating your studio background. The filtered light will change the color of the background, perhaps to better compliment your subjects clothing color. The background flash power level adjusts the brightness of the new color.
Keep the background back far enough back so there is no risk of colored light reflection back onto your subject. And your main light may wash it out some unless the background is farther back. At least six feet behind subject, maybe eight or ten feet is better.
A medium gray background is very practical, and is normally a great choice itself. Lighting intensity changes its shade of light/dark, and it is also easily changed to other colors with filters. The filter colors can come out a strong color on a darker background, arguably a bit much color for portraits. And arguably rather weak on a white background. Middle gray is hard to beat. Both the filter selection and the background light power level can affect color results. Some experimentation will be helpful.
Copyright © 2009-2024 by Wayne Fulton - All rights are reserved.
