One of the standard recommendations from the web are to create or buy various flash diffusers so as to make the light softer and less concentrated. I decided to test these out with various flash angles and so on. The three "treatments" are: no diffuser, film canister, and sto-fen flash diffuser. The film canister is a translucent film canister with a slit cut in it so that it fits over the onboard flash to form a DIY diffuser. The sto-fen is a standard store bought diffuser. While it is meant for a slave flash, it can be draped (inelegantly) over the onboard flash as well.
All shots are with a Rebel XT on a tripod with a Canon 50mm f/1.4 in P mode with manual focus. White balance is set to auto and the color parameters are at the default settings.
The subjects for the portrait are some of my son's stuffed animals. They're sitting about 6' away from the camera in a poorly lit bedroom. The flash is pretty much the only light source.
We start with the baseline--onboard flash with no diffuser.
|From flash test 4|
It's not a terrible shot, but it isn't good either. It's too contrasty and the colors are more garish than in reality.
Now we add the DIY film canister.
It doesn't make as much difference I was hoping. The light is a little more evenly spread. The contrast is less severe. Basically, the histogram is shifted to the left with less spiking at the top. The catchlights in the dinosaur's eye are a bit crisper. It's an improvement, but only a modest one.
Now we try the sto-fen diffuser. It is not meant for this purpose which means I'm shooting through the side of the device instead of out of the top as the thing was designed.
Of the three onboard flash shots, I like this image the best. It's again less contrasty than without the diffuser. The light is more even on the subjects. It hasn't shifted down the light as much as the film cannister. Overall, it's a slight improvement.
Now we return to the film canister, but add some paper in front of it pointing upward to try to imitate a bounce flash.
This helps. It's about the best of the choices.
Now we add the slave flash, which, in this case is a Canon Speedlite 430EX. It is aimed directly at our subjects.
This is about as awful as the onboard flash. While it's not as contrasty, the colors are garish and there's definitely a "deer in the headlights" quality to the light.
Now we add the Sto-Fen diffuser. (Note that the film canister DIY rig is too small for the Speedlite.
It does what it is supposed to do--cut the light and make it softer and more diffuse. It's a better image though it's still pretty bad.
Now we change aim angles. In this case, a 3/4 angle with the Speedlite to bounce some of the light off the ceiling (without and with diffuser, respectively)
This is a ton better than anything we saw previously. The light actually looks reasonably nice on the stuffed animals. With the diffuser, we end up with a classic bell curve histogram. Absent the diffuser, there is more light overall and especially in the background of the image, which isn't all that helpful. Either way, this is a vast improvement. (Of course, there's a problem if there's no ceiling off of which to bounce light. I'll have to try experiments in other rooms with vary high ceilings, but that's for another day.)
Since 3/4 high was so successful, why not go whole hog and rely completely on bounced light. Here, the Speedlite is aimed directly upward.
Apparently, this was too much of a good thing. Without the diffuser, the light is like that of the sun at high noon. There are lots of unflattering shadows under the eyes of the stuffies. The diffuser makes things a lot better. In fact, it's hard for me to choose between this shot and the 3/4 shot with the diffuser.
So what did I learn?
1. The diffuser makes a huge difference, especially when bouncing the flash. It makes a small improvement in the onboard flash, but it's still a lousy image.
2. Bouncing is key for halfway decent light. 3/4 is clearly the best without the diffuser. With it, either 3/4 or ceiling work. In situations where there are no ceilings or high ceilings off which to bounce, my guess is that 3/4 is the safe bet since you're still getting some of the light directly.
These experiments haven't revised my overall view of flash. I will still avoid it when possible, but at least I have a better idea how to shoot in flash situations to get somewhat decent photos.