If I wanted to interest children in the biological world by using macro photography I could think of nothing better to show them than the reproductive parts of flowers or close-up images of flies.
For me the fly’s eye is particularly interesting because with modern digital cameras and relatively inexpensive lenses it is not very difficult to see structures that are very different in shape and organisation from our own eyes. You might be aware that flies have 5 eyes: 2 compound eyes and 3 simple eyes known as ocelli that are present on top of the animals head. The ocelli form an obvious triangle known as the ocellar triangle.
The fly has five eyes; the 2 compound eyes marked with a C and the 3 ocelli that are circled
The compound eyes consist of individual elements called ommatidium (ommatidia is the plural). Each ommatidium acts as a sensor with its own cornea, lens, sensory cells and neural connections to the fly brain.
You do not have to be a university researcher to investigate for yourself the beauty of a fly’s eye anatomy. With simple macro photography equipment you can see the individual ommatidia yourself! By attaching microscope lenses to your camera via a bellows you can attempt to go one step further and perhaps resolve structure within the ommatidia of the compound eye.
105 mm Sigma Macro Lens used by Steve for the fly photo above
Automatic extension tubes were placed between the macro lens
and the camera to move the lens further away and increase the magnification
Seeing the Eye Surface in Detail by Conventional Macro Photography
The surface of the fly’s eye can be photographed with conventional macro lenses and macro-flash illumination. Where necessary automatic extension tubes can be added to provide increased magnification. The advantage of automatic tubes over manual tubes is that the electrical connection to the lens aperture is maintained so that small apertures with large depth of field can be selected. If required, supplementary magnifying lens such as those produced by Raynox will provide greater magnification and better resolution.
The compound eyes of flies have multiple sensory units called ommatidia, which are organised in a regular pattern.The image above was photographed with a 105 mm macro lens, automatic extension tubes and studio flash heads.
Using a Microscope Objective Lens
Macro Bellows fitted with an extension tube, a RMS threaded adaptor plate
and a Nikon x10 microscope objective.
Microscope Objective lenses, when used on a microscope or used on a bellows produce, a very shallow depth of focus. If you wish to carry out macro photography with a bellows and objective lens you need a bellows that will fit your particular camera, which in my case is a Canon. At the opposite end of the bellows you need an adapter ring, which the microscope lens can screw in to. Many microscope lenses come with a RMS thread, so for these lenses an RMS adapter ring is required to attach the objective lens to the bellows. In general you are best to try lenses in the range x4 to x20. Below that magnification range you are probably better using a macro lens. Above x20 it is extremely difficult to light the subject, as the working distances of most lenses are very short and so the lens tends to obstruct the illuminating flash. In addition, with higher magnification lenses, it becomes difficult to place the object of interest in the optical path as the field of view is very small.
A close up of the compound of a dead fly eye photographed with a low magnification microscope objective lens and bellows. The round corneal lenses on the ommatidal surfaces are packed in a dense regular array. The white objects are dust particles. Post-mortem discolouration of some ommatidia has occurred.
Seeing Beneath the Surface
In order to produce the image below, I photographed a fly’s eye using very bright studio flash heads and a 20x Nikon objective lens with a numerical aperture of 0.25. I focused on a plane just inside the ommatidial surface on those parts of the eye nearest to the lens i.e. near the middle of the picture shown below. Towards the bottom right-hand side of the image you will see sections through the outer ommatidial surface. Towards the centre of the picture you can see various planes within the ommatidia.
The initial image was photographed in RGB colour using a Canon DSLR.
The blue-green circles are thick ‘optical sections’ through the ommatidia at different levels.
A monochrome or greyscale image was produced directly from the RGB with the badly-named
Photoshop ‘Black and White’ adjustment dialogue box.
This type of images should not to be confused with a binary or true ‘black and white’ digital image. On the upper left you can see what look like the ommatidial corneas in semi-profile. Further down and to the right the image appears to look deeper into the surface of the eye.
Creating a Derived Image
The coloured (RGB) image was processed to form the derived images below. The new image was created using a masking tool within the Adobe Camera RAW Converter. This tool is designed to produce spatially restricted sharpening, so that parts of an image with little fine detail such as a sky in a landscape image, for example are not unnecessarily sharpened.
Adobe do not provide a direct route to produce the image using the thsi program so I have had to use the process of screen capture while working within the Camera Raw Converter. In practical terms what I did was as follows:
- The image was opened in ‘Camera RAW’. I then opened the ‘detail’ tab.
- I selected the ‘masking slider and moved it to the right with a finger pressed on the Alt key.
- I then simultaneously pressed the ‘screen capture’ button on the keyboard. Ideally one would have three hands to do this.
- I then ‘pasted’ the screen capture into a new blank image in Photoshop and then cropped it and flattened the layers.
This tool appears to analyse the initial RGB image and then create a new image in which brightenss value corresponds to regions of lower and higher local contrast or some derived arithmetic value of brightness such as the ‘local brightness gradient’. In general terms the algorithm appears to be using some measure of ‘brightness texture‘. At least, I guess that’s how the image processing operation works when creating a sharpening mask within Photoshop, as one would not want to sharpen ‘smooth’ areas of sky for example. Adobe do not of course reveal the inner workings of their spatial filter algorithms. They further compound this problem by using use silly terms for their tools, such as “puppet warp”, so users like me can only speculate about how they work based on the ouput.
2 different settings of the spatial masking filter were used to produce the images above
You can think of this image above as a map of local contrast differences within the original coloured picture. You can also think of it as a contour map, where the distribution of brightness of the new images is related to local variation in brightness within the original. Expert comments would be welcome. In order to make the image easier to view the images were then enlarged by 200% using the Adobe bi-cubic smoother interpolation function (image size menu). Two rounds of unsharp masking were used to make the enlarged image appear a little crisper. Within the initial image the hexagonal lattice network and internal features of the ommatidia were not at all obvious, however Adobe’s image segmentation function made the arrangement of the fly’s eye very obvious.
Alternatively you could start with the RGB image, apply the pseudo-coloured
Adobe Glowing Edges Filter within Photoshop and brighten it with a Levels adjustment layer
Understanding the Derived Image
In the case of the this image of the fly’s eye it seems reasonable to suppose that ‘smooth’ portions of the image with little brightness variation were converted to white. Other parts of the images above a certain threshold value of local texture or contrast gradient were converted to black or various shades of black. (For more information about segmentation based on ‘image texture’ or click here).
A hoverfly feeding on the pollen on the anthers of a Japanese Anemone
The same process was used as that applied to the image of the eye
You can understand what ‘image texture‘ or local contrast means if you imagine a three-dimensional (X,Y, Z) graph of brightness of an image, where brightness is the third dimension in the graph and the x-y values are coordinates within the image. The peaks and troughs of brightness on the ‘surface’ of the 3D graph represent the ‘texture’ of the image. A ‘rough image texture’ is one with lots of peaks and troughs of brightness
The actual image produced depends very much on the threshold values selected by the user (as shown above and below). In practical terms this depends on whether not a variety of pre-processing steps have been and used and what parameter (slider) values in the masking function are selected. In that sense the picture produced has an arbitrary appearance.
In order to demonstrate that the detail of the derived images produced is arbitrary
other values were used to create a related but substantially different picture
In this case a different starting point was also used.
The monochrome or greyscale image shown above was processed with the masking function.
An important question to ask yourself do the derived images have meaning in the sense that they corresponds to real (biological) structure. Even if the structures within the derived image are not ‘real’ the new images may still tell you something important about the underlying biology in a way that would be hard to discern by conventional photography with a camera. I would argue in this case that the image is meaningful although it needs careful interpretation.
Understanding the Biological Meaning
To understand the information content of the image above it is important to appreciate what ommatidia look like in cross-section. For an excellent and beautifully illustrated short video on insect ommatidia click here. For an excellent description of the ommatidium that will help explain the image above go the Pharyngula blog page on this subject by PZ Myers.
Ommatidium: A cornea, B crystalline cone, C & D pigment cells, E rhabdom,
F photoreceptor cells, G membrana fenestrata, H optic nerve
Taken from Wikipedia with thanks to the artist
A cross-sectional diagram of an ommatidium
Image copied from: Chance and regularity in the development of the fly eye
on the The PZ Meyers Blog, Pharyngula
My Biological Interpretation of a Derived Image
When faced with a completely new way of visualising a structure it can be difficult to be sure what one is ‘seeing’. My guess is that as one might expect the images to consist largely of light reflected from the most superficial part of the ommatidia, not deep within them. In that case so we might be ‘seeing’ structures near the top of the cross-sectional diagrams above. It would seem reasonable to suppose that the white hexagons, although not clearly corresponding to anything in the original coloured (RGB) image, result from the close packing arrangement of the ommatidia. Whether or not they ‘are’ the ommaditia is open to interpretation, for the black areas represent ‘peaks’ in local contrast and the white hexagaons, which the black areas define, are ‘valleys’ of local contrast (or vice versa). The inner black circles might be the pigmented cells surrounding the ‘sensor’ area and the central black spot might correspond to the inner sensory area.
On the other hand you should consider this image to be an artifact, in the true sense of the word (… something made or given shape by man). Expert opinion from a professional or amateur would be welcomed!
It strikes me that such quick photographic and image processing methods might have a minor role to play as a screening tool in research when studying the effect of genetic mutations on the structure of the eyes of the fruit fly for example. If such an approach were used in eye research a well defined texture-based segmentation method would have to be applied rather than an unexplained Adobe filter. In some cases studying the dysfunction of genes important in fruit fly eye development is also relevant to understanding gene involvement in Parkison’s disease in humans (more info>).
A Philosophical Point
We should be careful when interpreting images be they scientifically produced or by any other means. This is especially the case when those images do not form part of our every day lives. Even when images do form part of our everyday lives we should exercise a healthy degree of scepticism about what we believe they ‘mean’. The BBC documentary ‘Seeing is Believing‘ showed, by the use of powerful perceptual illusions, that we do not look at images in a completely value free way that is independent of our previous experience. The fact that there appear to be hexagons in the above image, although not in the original, should make you healthily skeptical. Skepticism, when not taken to extremes, is very useful.
I will leave you to ponder my favourite philosophical quotation on the need for skepticism:
“It seems to me what is called for is an exquisite balance between two conflicting needs: the most skeptical scrutiny of all hypotheses that are served up to us and at the same time a great openness to new ideas. If you are only skeptical, then no new ideas make it through to you. You never learn anything new. You become a crotchety old person convinced that nonsense is ruling the world. (There is, of course, much data to support you.)
On the other hand, if you are open to the point of gullibility and have not an ounce of skeptical sense in you, then you cannot distinguish useful ideas from worthless ones. If all ideas have equal validity then you are lost, because then it seems to me, no ideas have any validity at all. “
Carl Sagan, ‘The Burden of Skepticism’, Pasadena Lecture 1987, Quoted in ‘Why People Believe Weird things Pseudoscience and other confusions of our time’ by Michael Shermer
I would like to thank my friend Geoff Scordia, an amateur microscopist and avid microscope collector, for creating the macro bellows arrangement for me and very generously lending it to me along with a large series of microscope objectives and older macro lens designed for use with these bellows.