Diffusers shootout, vol. 1
Nikolai
Registered Users Posts: 19,035 Major grins
Preamble
Originally me and DavidTO were planning to meet and try our flashes with all available gear. Unfortunately, our lives caught us with several different "no-go" reasons. And while our plans to meet-and-shoot are still valid, I decided to do something on my own, missing David and his LightSphere.
Since I didn't have a partner and a model (my family hides under the table when they see me with the camera) I have no choice but do something different than previously planned "model shooting". I decided to test relative differences and, most of all, light fall-out.
Setup
For the test I used my Canon 20D with EF 28-135, wide open at 28mm. Originally I thought to use my 50mm f/1.8 prime, but the size of my dining room would not let me test "full height" shot, so I opted for a wider angle.
Since I didn't care about each shot actual quality (I was interested ONLY in RELATIVE difference) I have chosen my manual settings pretty arbitrarily: ISO100, 1/125s, f/3.5. Camera was set on a tripod just a couple inches above 5 ft (~158cm) and triggered with a wired remote. All shots taken inside the dining room with high ceiling (about 13 ft, i.e. ~4m), at night (about 10 pm) with no other lights.
My target was Botero 8'x16' white backdrop hanging from Amvona Dynatran rack. I also set a 10' light stand just a few inches in front of the fabric - it gave me a focusing target and at least some source of shadow.
Shots taken in "RAW+Large fine jpeg" mode, WB set to "flash".
I used jpegs straight from the camera with absolutely no postprocessing to upload for the light fall-out test. I used RAWs in ACR full manual (all settings "as shot") to create central cutouts for better shadows display.
Test Matrix
I used the following flashes and diffusers:
All in all, I got 3*9=27 test cases.
Legend
for the file names
BI: built-in flash
PP: built-in flash with ping-pong ball
EX: "bare" 580EX
OB: with Stofen Ombibounce
SF: with LumiQuest mini-Softbox
_1: +1EV in flash
_45: at 45 degrees
_451: +1EV and at 45 degrees
Pictures
Here's how my test target looked like:
Note: this picture has been taken from 15 ft position. As you can see, the test area covers more than an average person height (I'm 6'2"), even though for the test purposes I pointed the camera parallel to the ground, which is slightly higher compared to what would have been used in the actual "group/full height" shot.
First test matrix (light fall-out):
Second test matrix (central shadows):
The actual shots are available here: http://nik.smugmug.com/gallery/1162303
Results
General
While the shots do not pretend on any actual quality, they seem to reflect the natural differences in flash unites/diffusers used, and as such this test is fully a representative one.
However, the high ceiling severely impacted the quality of shots taken at 45 degrees (with Omnibouncer), since it only led to less light and additional vertical lightness gradient.
Light fall-out
All cases shows similar and easily explainable fall-out pattern.
At 4ft (~120 cm, "head" shot) the central area is lighted much higher than the edges with a rather steep radial gradient. Difference between the lightest and darkest point on the canvas ranges anywhere between 80% to 35% in the most uneven case and 70% to 50% in the most even one (data from Luminosity channel in LAB mode).
At 9 ft (~2m75cm, "above waist" shot) the gradient is still visible, but has different, cylindrical shape: lightness decreases slowly towards the horizontal edges (55% to 66%), and then falls down quickly on the edges down to 75%-80%.
At 15ft (~4m75cm, "full height/group" shot) the whole area is lit rather evenly, although understandably dimmer.
All the diffusers dimmed the original light, and, sadly, slightly affected its color/temperature, especially the ping-pong ball. The tints, while relatively easily fixable in RAW (or LAB), may create a problem for a jpeg-only shooters or people unfamiliar with advanced post-processing techniques.
Apart from the facts mentioned above, to my surprise (and may be due to the lack of praised LS:-), all the diffusers ended up pretty close to each other.
At 4 ft, where the gradient is the steepest, the slight leader (the most even shot) was Omnibouncer (I expected mini soft-box to be on top here with its larger surface, which, in theory, should have been most effective at the closest range). The worst case (most uneven) turned out to be "bare" 580EX in its full auto ETTL mode, and all the shots at 45 (once again, with the ceiling that high it's not wise to use this technique, which works fine with the conventional ceilings under 10 ft).
I shall repeat: best and worst are only relative figures, the absolute difference is very minor.
At 9 ft, and especially at 15 ft, the difference between different diffusers or lack of thereof is practically indistinguishable. Again, totally understandable: at these distances the difference in size of reflectors is negligible, so it's better to remove them and save some lumen (and juice:-).
Shadows
Once again, the biggest difference is observed at 4ft, while 9ft and 15ft already shows practically no difference what diffuser (or none at all) to use in this respect.
At 4ft all the diffusers seem to be "better than nothing" since all of them created less sharp shadows than "bare" flashes. Interesting enough, even with the celing thus high, at this distance aiming at 45 changes shadows position quite noticeably, which may be desired (or undesired) effect.
While all results being very close, it's possible to select the winner - mini-softbox (phew, at least here my initial thoughts were correct:-). Worst is bare built-in flash, followed closely by bare 580 - again not surprising at all.
Tint
Everything you put on the way of light affects it. Ping-pong ball is the worst. However, even commercial products change the spectrum slightly, with Omnibouncer being a bit closer to the original flash than miniSoft-box, which made the light a bit warmer (again, it may be a desired effect). Both of them are really easy to fix, though.
Conclusions
Unless you shoot with flash from a very close distances (4 ft and less - but think first what your subjects will experience if you discharge 580EX at full power in their faces from this distance:-) it does not seem to matter what diffuser you use. Better use none at all, you'll draw less power for the same light coverage. Expect some vignetting at 9ft and closer.
At 4ft mini Softbox produces the softest shadows, but has a tad more tint and a bit more gradient than Omnibouncer.
What's next
Once we get together with David, expect LS to enter the battlefield. Also, we will try shooting some faces :-)
HTH
Test design, execution, presentation and pictures: copyright © 2006 Nikolai Sklobovsky.
Any non-personal use is prohibited without proper authorization.
Originally me and DavidTO were planning to meet and try our flashes with all available gear. Unfortunately, our lives caught us with several different "no-go" reasons. And while our plans to meet-and-shoot are still valid, I decided to do something on my own, missing David and his LightSphere.
Since I didn't have a partner and a model (my family hides under the table when they see me with the camera) I have no choice but do something different than previously planned "model shooting". I decided to test relative differences and, most of all, light fall-out.
Setup
For the test I used my Canon 20D with EF 28-135, wide open at 28mm. Originally I thought to use my 50mm f/1.8 prime, but the size of my dining room would not let me test "full height" shot, so I opted for a wider angle.
Since I didn't care about each shot actual quality (I was interested ONLY in RELATIVE difference) I have chosen my manual settings pretty arbitrarily: ISO100, 1/125s, f/3.5. Camera was set on a tripod just a couple inches above 5 ft (~158cm) and triggered with a wired remote. All shots taken inside the dining room with high ceiling (about 13 ft, i.e. ~4m), at night (about 10 pm) with no other lights.
My target was Botero 8'x16' white backdrop hanging from Amvona Dynatran rack. I also set a 10' light stand just a few inches in front of the fabric - it gave me a focusing target and at least some source of shadow.
Shots taken in "RAW+Large fine jpeg" mode, WB set to "flash".
I used jpegs straight from the camera with absolutely no postprocessing to upload for the light fall-out test. I used RAWs in ACR full manual (all settings "as shot") to create central cutouts for better shadows display.
Test Matrix
I used the following flashes and diffusers:
- built-in flash
- built-in flash with self-made white ping-pong ball diffuser
- "bare" Canon Speedlite 580EX in auto ETTL mode
- 580EX with Stofen Omnibouncer
- 580EX with Stofen Omnibouncer at 45 degrees
- 580EX with Stofen Omnibouncer +1EV in-flash
- 580EX with Stofen Omnibouncer at 45 degrees +1EV in-flash
- 580EX with LumiQuest mini Softbox
- 580EX with LumiQuest mini Softbox +1EV in-flash
All in all, I got 3*9=27 test cases.
Legend
for the file names
BI: built-in flash
PP: built-in flash with ping-pong ball
EX: "bare" 580EX
OB: with Stofen Ombibounce
SF: with LumiQuest mini-Softbox
_1: +1EV in flash
_45: at 45 degrees
_451: +1EV and at 45 degrees
Pictures
Here's how my test target looked like:
Note: this picture has been taken from 15 ft position. As you can see, the test area covers more than an average person height (I'm 6'2"), even though for the test purposes I pointed the camera parallel to the ground, which is slightly higher compared to what would have been used in the actual "group/full height" shot.
First test matrix (light fall-out):
Second test matrix (central shadows):
The actual shots are available here: http://nik.smugmug.com/gallery/1162303
Results
General
While the shots do not pretend on any actual quality, they seem to reflect the natural differences in flash unites/diffusers used, and as such this test is fully a representative one.
However, the high ceiling severely impacted the quality of shots taken at 45 degrees (with Omnibouncer), since it only led to less light and additional vertical lightness gradient.
Light fall-out
All cases shows similar and easily explainable fall-out pattern.
At 4ft (~120 cm, "head" shot) the central area is lighted much higher than the edges with a rather steep radial gradient. Difference between the lightest and darkest point on the canvas ranges anywhere between 80% to 35% in the most uneven case and 70% to 50% in the most even one (data from Luminosity channel in LAB mode).
At 9 ft (~2m75cm, "above waist" shot) the gradient is still visible, but has different, cylindrical shape: lightness decreases slowly towards the horizontal edges (55% to 66%), and then falls down quickly on the edges down to 75%-80%.
At 15ft (~4m75cm, "full height/group" shot) the whole area is lit rather evenly, although understandably dimmer.
All the diffusers dimmed the original light, and, sadly, slightly affected its color/temperature, especially the ping-pong ball. The tints, while relatively easily fixable in RAW (or LAB), may create a problem for a jpeg-only shooters or people unfamiliar with advanced post-processing techniques.
Apart from the facts mentioned above, to my surprise (and may be due to the lack of praised LS:-), all the diffusers ended up pretty close to each other.
At 4 ft, where the gradient is the steepest, the slight leader (the most even shot) was Omnibouncer (I expected mini soft-box to be on top here with its larger surface, which, in theory, should have been most effective at the closest range). The worst case (most uneven) turned out to be "bare" 580EX in its full auto ETTL mode, and all the shots at 45 (once again, with the ceiling that high it's not wise to use this technique, which works fine with the conventional ceilings under 10 ft).
I shall repeat: best and worst are only relative figures, the absolute difference is very minor.
At 9 ft, and especially at 15 ft, the difference between different diffusers or lack of thereof is practically indistinguishable. Again, totally understandable: at these distances the difference in size of reflectors is negligible, so it's better to remove them and save some lumen (and juice:-).
Shadows
Once again, the biggest difference is observed at 4ft, while 9ft and 15ft already shows practically no difference what diffuser (or none at all) to use in this respect.
At 4ft all the diffusers seem to be "better than nothing" since all of them created less sharp shadows than "bare" flashes. Interesting enough, even with the celing thus high, at this distance aiming at 45 changes shadows position quite noticeably, which may be desired (or undesired) effect.
While all results being very close, it's possible to select the winner - mini-softbox (phew, at least here my initial thoughts were correct:-). Worst is bare built-in flash, followed closely by bare 580 - again not surprising at all.
Tint
Everything you put on the way of light affects it. Ping-pong ball is the worst. However, even commercial products change the spectrum slightly, with Omnibouncer being a bit closer to the original flash than miniSoft-box, which made the light a bit warmer (again, it may be a desired effect). Both of them are really easy to fix, though.
Conclusions
Unless you shoot with flash from a very close distances (4 ft and less - but think first what your subjects will experience if you discharge 580EX at full power in their faces from this distance:-) it does not seem to matter what diffuser you use. Better use none at all, you'll draw less power for the same light coverage. Expect some vignetting at 9ft and closer.
At 4ft mini Softbox produces the softest shadows, but has a tad more tint and a bit more gradient than Omnibouncer.
What's next
Once we get together with David, expect LS to enter the battlefield. Also, we will try shooting some faces :-)
HTH
Test design, execution, presentation and pictures: copyright © 2006 Nikolai Sklobovsky.
Any non-personal use is prohibited without proper authorization.
"May the f/stop be with you!"
0
Comments
I can send you a LumiQuest Pocket Bouncer for a few weeks if you want to add it to the mix. Probably not too much different, but ??
http://www.lumiquest.com/lq871.htm
I knew, of course, that trees and plants had roots, stems, bark, branches and foliage that reached up toward the light. But I was coming to realize that the real magician was light itself.
Edward Steichen
Thank you for your comment and for the offer!
I used to have a similar system (Lumiquest ProMax System, http://www.lumiquest.com/lq931.htm) with my (now sold) Sony 828 and its HVL-F32X. I found it draining out a lot more power from the flash compared to, say, mini soft box, mostly due to the fact it's working only when the flash is in upright position. Since it's reflective size is not *much* different from the softbox I currently have, I think we're fine for now.
However, if we decide to add "power effectiveness" to our test case, I will take you upon your word :-)
Cheers!
Portfolio • Workshops • Facebook • Twitter
To present the light fall-out. I think was too tired last night (I mean, this morning:-) since I only thought of it when I already finished the report, so I didn't want to start it all over:-). But I surely use it next time.
On this picture left part is split into 5% regions, thus providing very clear indicator how much light gets where:
In case anybody wonders how this is made, here is my workflow, which can be easily converted into an action since all the shots are equally framed.
- Select left-hand part with the marquee tool, copy to a new layer (Ctrl+J for Windows, something similar for Mac)
- Open the Surface Blur tool (CS2 and better, Gaussian blur otherwise - and in the latter case make sure you haven't selected the central pole), use it with 100 pixel radius, 20 threshold. This gonna be you Source layer.
- Using the eyedropper tool (and Lab/Luminosity channel if needed, although RGB works fine, too) locate the brightest part of the source layer. You don't have to be very precise, just make sure you're close. Once you found it, help yourself with two guides to mark it in a cross hair way.
- Select magic wand tool, make sure "Sample all layers" check box is OFF.
- Select 5% tolerance
- Click on the brightest point
- Copy selection into a new layer (same Ctrl+J), and make sure the new layer is an active one (CS and better makes it automatically, I'm not sure about PS v7 and before)
- Select new layer (copy+click on the layer icon)
- Press Shift+Backspace to fill the new selection with some contrasting color of your choice, confirm your choice.
- Switch layer blending mode to color
- Hide the newly colored layer, return back to Source
- Repeat steps 6..11, each time increasing tolerance 5% and selecting another contrasting color, until you have almost full coverage.
- Hide the Source layer, show all the colored ones.
- Using Text tool, type in the values and then arrange them appropriately.
- That's all folks!
HTHThis step is necessary to avoid canvas fold-marks affecting the general picture. If you have an impeccably uniform target this step may be unnecessary, but it should not hurt anyway.