Concave Flash Diffuser Test

[SteB]

These are the results from some preliminary tests of the Concave Macro Flash Diffuser I proposed on this thread http://www.juzaforum.com/forum-en/viewtopic.php?f=2&t=12151

My aim with this diffuser was to be able to create more directional and slightly more contrast than the end of lens plastic cup diffuser gives, but without the central hotspot problem you tend to get with conventional flat or convex macro flash diffusers. Whilst I may not have explained this as my aim, if you look at the illustrations of these macro flash principles you can see this is what I was showing i.e. catchlights less prone to blown highlights. Basically I believe that flat or convex diffusers work in spite of their design and not because of their design. In other words the larger diffuser area softens the shadows and light, but there is still a central hotspot that has a tendency to blow highlights.

These prototype diffusers were quite crudely put together in a way to make it easy to quickly modify them. I wanted to test how well they worked in the field and any problems before finalizing the design and showing them. However, they have worked beyond expectation and so I am sharing the results for others to test the designs. My altruism is just because I believe more heads are better than one, and I hope to learn as much from other people's experimentation. I am already learning from other people's experience with the end of lens cup diffuser, both from those that have acknowledged it and others who are trying it (the catchlights give it away). All these diffusers are made from is translucent deodorant plastic tops (Dove - Invisible Dry). I was looking for slight bigger diffusers, but had to make do with these. However, considering their small size they have worked very well. They are just contained in a tube of aluminium sheet cut from drinks cans, and the reflective surface is an important part of their design.

One additional and very welcome discovery is that they have hugely improved the E-TTL performance of the MT24EX on the Canon 40D. No longer does it go mad and grossly over-expose if there is any part of the background it can't light, and it know behaves as I would expect a pattern metering TTL system to behave. Occasionally it does over-expose a bit, and does sometimes over-expose a bit at high magnifications. However, generally the metering is slightly conservative, tending towards slight under-exposure and amazingly consistent compared to how the MT24EX usually behaves on the 40D.

What the diffusers look like.



First some comparison shots showing the catchlights on an old Dragonfly fly exuva. I have used screen grabs of the comparison feature on Faststone Image Viewer. Pay special attention to the histograms. I shot at both life size and twice life size, because with the MP-E 65mm this entails the flash heads being at different angles. Past 2x life size the flash angle remains similar. The background is just my cup diffuser. Note how the Concave Diffuser produces consistently different histograms - without the mid-tone dip of the others. These images are just default Lightroom 2 conversions to TIFFs. All comparisons are Concave Diffuser, Bare Flash Heads, StoFen Diffusers - left to right. There are the whole images and then crops for catchlight comparisons.

Life-size:


Crops:


2x life size:


Crops:


In my next post I will post the results of the field test, including similarly composed shots of a Miner Bee - one taken with the end of lens cup diffuser and the other with the concave diffusers. The crops from these images show the diffuser behaves as predicted.

[Toby]

about time Canon came up with their own diffusers - you wonder if they ever field tested the MT-24EX.

[TheBat]

Hi Stephen,
A great report on a very well conducted test. I know from personal experience how hard it is to put these things together and how hard it can be to get the results that you are looking for. I am really happy that your results turned out just as you predicted.
I am very surprised at the lack of any decent diffusion from the StoFens? The results from the histograms indicate very strongly that they do nothing, or very little, when compared to the histogram of the bare flash head. . .
I look forward to your subject images. Congratulations.
Bruce

[SteB]

Here is the results of a field test comparison. By good fortune I noticed that a Miner Bee on a Coltsfoot flower shot I had taken on 28 March with the end of lens plastic cup diffuser, was very similar in composition and angle to one I had taken on 3 April with the experimental concave diffusers. So I have used this shot as I believe that is a very valuable illustration of how these diffusers work. Just click on the photos and they will lead you to a 1400px higher resolution image. The 1400px images have been fully processed for posting, sharpening etc, tailored to each image. So you won't see much difference. The only exception to my normal post-processing regime is that I didn't use any local adjustments to even out the lighting that I normally apply. However, the further comparisons and crops displayed using the Faststone Image Viewer comparison feature - use the fairly flat images from the RAW converter that I normally prepare for post-processing - show far more difference. It is important to understand that my aim with the cup diffuser was to produce flat low contrast images, not because I want the end result to look like that, but because it allows me to easily apply far more aggressive sharpening. If you apply too much sharpening to an image with already hight contrast and lots of small bright highlights, it looks horrible.

Please note that the finished concave flash diffuser shot looks slightly over-sharpened in comparison to the end of lens cup diffuser shots because I have not perfected my sharpening for these images yet - as I only first got some yesterday.

The 2 finished images - click on the image for the higher resolution version:

28 March - end of lens cup diffuser:


3 April - experimental concave diffuser:


The comparisons with the flatter pre-post processing images. These are just from Lightroom 2, but they have been passed through Neat Image to remove noise and add slight sharpening as none of these are applied in Lightroom 2, as Neat Image does a better job.. At default Neat Image removes a slight bit of detail from the compound eyes. I don't normally bother with this for web size posting as even at 2000px none of this detail is visible, but if I was preparing a file for a large print etc I would use masking and fine tuning of the Neat Image settings to preserve this detail. Please remember these images are deliberately flat and not too punchy for post processing. I think it is fairly clear that the Concave diffuser provides a slightly contrastier image with more catchlights, but they are subtle and the highlights aren't blown. This was my aim with this diffuser so I am very pleased that the diffuser acted as I predicted it would act, which shows that my theory about this aspect of diffuser design is correct. Without blowing my own trumpet too loudly I think it also demonstrates that I have good insight into how macro flash light behaves in that I can theoretically design a diffuser, and know how the light from it will behave, before I've actually used it in the field.

Also note that there has been no local adjustments of the catchlights on the eye, they are as shot.

The whole images:


2 crops to show the catchlights in detail. Click on the images and they will take you to 1400px higher resolution versions:




Finally a couple of habitat shots showing where the photographs were taken, and a Miner Bee feeding on a Coltsfoot Flower. They show how low these flowers are and how dark these Bees are. Please note they were just taken single handedly with my Panasonic TZ3, whilst I held my Canon 40D and MT24EX in my other hand, so they are not meant to be great photos. If you look closely at the central flower in the shot the small black speck is the feeding bee in the following photo. You'll also notice how it still looks fairly bleak and winter like with very few flowers. I had to work hard to find these bees and any flowers. The Coltsfoot flowers are only about 2cm or so in diameter, so it gives some idea of scale. These dark little bees are a serious test of any diffuser, as they are both very dark, and have very shiny eyes, which cause blown highlights with a lot of flash diffusers.

[TheBat]

Without blowing my own trumpet too loudly I think it also demonstrates that I have good insight into how macro flash light behaves in that I can theoretically design a diffuser, and know how the light from it will behave, before I've actually used it in the field.

Hi Stephen.
Are you planning on going into production or design consultancy? This question seems reasonable following the statement above. . . .
I believe that your results show exceptional promise; there is clearly more detail in the mid-tones on the new diffuser histograms.
As I said above, well done and congratulations. With results like these I don't think that there is any reason why you shouldn't have a little blow on your trumpet. . .
Bruce

[liewwk]

wow .. Stephen .. u manage to test with bee ..

again thanks .. some how I start use to the cup diffuser even it lack of highlight .. but I think I prefer the soft light instead

and I think I make similar to Concave diffuser before just little bigger compare what u have ... and I still like your's cup

[SteB]

about time Canon came up with their own diffusers - you wonder if they ever field tested the MT-24EX.

Yes it is a bit puzzling why Canon can't produce their own diffusers, and even other accessories for the MT24EX, like the Kaiser Bounce Flash Shoes I suggested using a couple of years back. I could quite easily do some quick sketches of some diffusers that would slide on and off the MT24EX and work a lot better than any other commercial product. As all these things would just be made from moulded plastic it should be a breeze for a big corporation like Canon to produce. If I can come up with solutions to basic problems MT24EX users face, it makes you wonder why a big organization like Canon with their presumably huge R&D budget couldn't come up with a few bits of moulded plastic to greatly enhance their products such as the MT24EX and MP-E 65mm.

Hi Stephen.
Are you planning on going into production or design consultancy? This question seems reasonable following the statement above. . . .
I believe that your results show exceptional promise; there is clearly more detail in the mid-tones on the new diffuser histograms.
As I said above, well done and congratulations. With results like these I don't think that there is any reason why you shouldn't have a little blow on your trumpet. . .
Bruce

Thanks Bruce. As I said in response to Toby's response I could quite easily sketch out some basic diffuser designs and head adapters like the Kaiser Bounce Flash Shoe that would greatly enhance the MT24EX or other flash gun. These are basically just bits of moulded plastic and it would be a doddle for anyone with the facilities to make this sort of thing. Ideally the plastic dome would be slightly thicker at the top than the sides. Also it could be made with a sort of telescopic tube (one tube fitting inside the other) so you could adjust the distance from the flash head. This would also allow you to have a few different sizes and designs that could be changed for different subjects. The actual shapes aren't too critical and just need to be adapted to the particular flashgun, such as the MT24EX here. Unfortunately I don't have the resources or connections to really do this type of thing myself.

wow .. Stephen .. u manage to test with bee ..

again thanks .. some how I start use to the cup diffuser even it lack of highlight .. but I think I prefer the soft light instead

and I think I make similar to Concave diffuser before just little bigger compare what u have ... and I still like your's cup

Thanks again. This is not really meant as a replacement for the cup diffuser, but an addition to it for a slightly different type of light. In fact this type of diffuser could be made to work with a cup diffuser as it is very effecient. This particular design is in fact more or less as efficient at the bare flash head because the focusing of the light seems to cancel out the light loss due to the diffusion. I found that with my dragonfly exuva test at 2x that I could get a correct exposure at 1/64 using the power ratio control in in manual instead of 1/8-1/16 with the cup diffuser. I think each of these types of diffuser has their advantages and there is nothing to stop them being used together. Sometimes you might want softer light with more light being thrown onto the background such as with the cup diffuser, and sometimes you might want contrastier light.

[liewwk]

thanks Stephen ...

[orionmystery]

Excellent post Stephen and another great DIY diffuser! It's also amazing that you have two similar shots of miners bees. What's the odd of that?:D

What do you think of using a pair of GF Puffers as the front diffusers?

[SteB]

What do you think of using a pair of GF Puffers as the front diffusers?

Hi Kurt

I did think a while back about using some GF Puffers in reverse, but building the softbox was too complicated. One of the reasons I am using a concave diffuser is because the shape of the diffuser acts a bit like a lens and has an effect on the way the light is distributed. The light from a convex diffuser like the GF Puffers (in the convex position) is distributed differently, in that there is a brighter central part and the light is scattered sideways. Whereas the light from a concave diffuser is more focused, but more evenly distributed and without such a central hotspot. Try it out yourself with this simple experiment. Get a sheet of thin aluminium sheet such as piece from a drinks can or aluminium foil, so you can form it into a tube. Get a white domed deodorant roll on top. Form a tube around it. Then with the lights dimmed or off, shine a small powerful torch through it, both in the concave and convex position onto a light coloured painted wall or door. Notice how different the light spread is, when the dome is reversed. One (concave) gives a more focused but even circle of light, and in the convex position, light is scattered more to the sides, but you will notice the light is brighter in the central part. The brighter central part is the problem, so putting a convex diffuser on top of a concave one would be self-defeating.

Light from a macro flash is very different than a flash at normal distances. With conventional flash in a room, it is good that light is thrown to the sides, because it bounces off walls and ceilings, filling in shadows. However, due to the inverse square law - in a macro situation - light bounced off something more than a few cm from the subject is totally wasted, because it is too weak to have any effect. This is because of the relative, not actual distances. If you were 3m from a person in a room, then the light would have to travel 6m to double the distance. Whereas if your macro flash is 3cm from the subject, something only 6cm away would be the same relative distance as regards the inverse square law light drop off, of around 2 stops. So with macro flash it is no advantage to scatter light to the sides. The brighter central part of the light from a flat or convex diffuser enhances the specular reflections, making them brighter.

It would be possible with a careful blocker design to evenly distribute the light, but if it is only slightly duller at the centre, you will get the effects you could see with your diffusers in the jumping spider eyes. The concave diffuser gives a far more even light spread, without the need for such a careful design, and the yet the brighter central part is nowhere near as hot. Yes shiny surfaces will show the brighter centre. Although this could be fixed if the top of the dome was slight thicker and the sides thinner. It is also easy to just push a bit of thin white packing foam into the central part of the inverted dome and this dulls the centre even more with the concave diffuser - although you risk a darker centre type reflection if this is not carefully balanced.

Also the idea with my cup diffuser design is that it is like a mini room, a mini studio - it doesn't just diffuse the light. The light is reflected around inside it, so all that reflected light doesn't just bounce outwards. I am going on a bit, but there is actually a lot of careful thinking behind these apparently simple diffuser designs. Unfortunately it takes a lot of words to explain all this. I am not saying that these are the only diffuser designs that will work, but within the confined space you have with the MT24EX bracket I doubt there are many other methods that would work better. The bigger bounce card designs like some are using successfully with bigger flash units, work a bit like the cup diffuser. It is easy to see when they are being used, especially with the jumping spider shots.

[Alex]

Hi Stephen!

You're putting quite some work into this - hats off!

Just out of curiosity, you don't happen to have one of these fellas around, do you:

If you do, I'd really love to see how your setup does when shooting any line of text on the bunny, since this has been my diffusion-proving ground in years past.

Regards,
Alex

[orionmystery]

Thanks for the explanation, Stephen. You must be a scientist .

Yes, I do mean with a reversed GF Puffers. You read my mind!

[SteB]

Hi Alex, thanks

Unfortunately I don't have one of these around and of course you come from the area where they make lots of these fine chocolates. What''s the size of this chocolate bunny so I don't have to chase around the shops looking for one? It looks a good standard target though as it has shiny surfaces. At the moment I only have these concave diffusers for my MT24EX and for high mag stuff on the MP-E 65mm. However, I am going to make some for larger flash units for lower magnifications than 1:1. They will basically just be softbox type designs, but with a concave instead of flat or concex front. They will be quite like Gary Fong Lightspheres in shape, and I might actually use one of these. The main difference is that I will line the insides with reflective material as I only want light to go through the inverted dome part, and not through the sides. The main idea of them is to reduce the central hotspot intensity that you still get even with softbox designs. I have found that the concave shape gives more even light coverage, and it is also more effecient with less light loss.

[mackay123]

Hi stephen
Do you have, or are you setting up your own website?, if you aren't then I really think you should be!, another in depth and very useful article for many.

[SteB]

Rob Ault who runs a very nice Blog on macro photography and HDR, did a very interesting and well written comparison between my cup diffuser, his own very effective diffuser and other designs on the MT24EX. His comparison photos are particularly good. I think it is well worth a look and he has some excellent images and very well written pieces.
http://www.robault.co.uk/blog/