Gestalt - The Limitations of G.I.S.S. (General Impressions of Size and Shape)

At the heart of any discussion around gestalt we have G.I.S.S., meaning the general impressions of size and shape, apparently derived from a WWII Royal Air Force term.  Curiously, there is also a valid association with a similarly sounding word JIZZ, apparently of Irish origin (HERE) and pre-dating WWII by some decades.  It is unclear what is implied by the Irish use of JIZZ in the context it was first used and whether it related to size, shape or the general 'gist' or 'energy' of a living species.  In any case, whether referring to the German term gestalt, the British term G.I.S.S. or the Irish term JIZZ, we are essentially dealing with the same concept.  We are referring to that almost indescribable uniqueness of a species in terms of its size, shape and how it carries itself, including it's behavior.  This is a much more complex concept than one derived solely from size and shape.

When the British Armed Forces came up with the concept of G.I.S.S. they were referring to the identification of fixed-wing aircraft.  Due to the limits of design, war planes of the 1940's would have all moved about in very much the same way.  So, realistically G.I.S.S. would have been mainly about the relative proportions of different aeroplanes.  It probably wasn't envisaged that such a concept would be adopted and applied to the general impression given by living creatures.  After all, there is far more to this subject then simply the size and general proportions of a bird.

Photographs represent an excellent opportunity to study a subject in fine detail and it is highly tempting to believe that we can take accurate measurements from photographs and use those to determine the accurate shape or proportions of a bird.  If we were going to start anywhere we might start with solid structures like the bareparts - consider perhaps the relative proportions of the eyes, bill and legs.  We might then include length of the primaries beyond the tertials, and the tip of the tail beyond the primaries at rest.  All these structures, on the face of it at least, seem like very valid and predictable measurements from photographs, and as birders we have become accustomed to studying these proportions in the field.  But, as I have demonstrated in recent postings (HERE, HERE and HERE) these are not reliable measurements from photographs.

At the heart of this challenge we have to accept that a photograph is merely a two-dimensional projection of a three-dimensional world.  We can only reasonably accurately measure objects which lie perfectly parallel to the plane of the lens and sensor, as discussed HERE.  When we try to measure something which is not parallel to that plane we run into the problem of perspective foreshortening.

And yet, even when we take away all field marks and leave only the simple outline of a bird the brain is quite good at interrogating general proportions along with our library of knowledge and beginning the process of identification.

If we are to approach gestalt in bird images from an objective, reasonably scientific perspective we may have to rethink some of our long held beliefs and discard some of the tools we have been using in the field and in our study of photographs.  We may have to come up with some new tools to work through this challenge of gestalt from bird images.

Link to quiz solution.

Gestalt - The Limitations of Leg Proportion Analysis

It is within the subject of gestalt that I tackle the question of bird size and shape in this blog.

Primary Projection and the Eye/Bill Ratio analysis so far have proved to be fairly unreliable analytical tools owing to the problems of perspective foreshortening and anatomy.  What about the analysis of the proportions of the avian leg?   

Anatomy of the leg

As we have found with the analysis of primary projection, no discussion about the proportions of the leg can be made without some understanding of the anatomical functioning of avian leg bones and joints.  Below I have illustrated the leg bone anatomy of a Bertholot's Pipit, Anthus berthelotti.  Clearly, like the upper limb or wing bones, a considerable amount of the leg anatomy of birds is hidden inside the body.    It is easy to forget while watching birds that all we are seeing are the equivalent of the shin and foot of the mammalian leg.  In effect birds walk around on the tips of their toes.  Everything above the shin is tucked away inside the body cavity.  For passerines and many other birds the 'thighs' as they are called in birding (in reality the shins) are muscular and feathered.  Wading birds tend to have more of the tibia exposed and it tends to be less muscular and unfeathered.

Movement of the leg

When a bird crouches the motion is akin to a person, standing on the tips of ones toes, simultaneously bending ones knees and ankles.  The 'thighs' lie close to the body, hidden inside the flank and belly feathers. As a bird raises itself tall leg joint articulation is equivalent to a human standing tall.  The knee and ankle joints open and in doing so the 'thighs' are exposed through the flank and belly feathers.  Clearly posture therefore has a bearing on the proportion of a measurement like tibia/tarsus ratio.  As illustrated below tibia/tarsus ratio can be highly variable.  There is more than one factor at play in this example however as discussed below. 

Angle of view and foreshortening

When we look at a perched bird it should be obvious that it's legs are not positioned vertically below the body but tend to be splayed to assist with balance.  While the tibia and tarsus may be aligned in the same plane unless our angle of view is perpendicular to this plane we inevitably have perspective foreshortening.  This makes it impossible to take an accurate comparative measurement between these two structures.

We do have a slightly better chance of comparing toe and claw length as these usually align favourably for the camera.  One does need to be careful however with the articulation of the bones of the toe due to the terrain and the potential for the claw to get snagged and skew the toe.  In short, measurements of structures and objects that can move relative to one another is highly challenging from photographs.

Judging points correctly

We have seen how variable the tibia length can appear simply due to posture and feather position.  Typically legs being relatively small features tend to be obscured more often than not.  Legs and joints are also often uniform in colour and it can be hard at times to judge the position of joints and where we should start to take our measurements from. 

In Summary

At risk of beating the same old drum I find myself drawn to the same conclusions I had reached earlier on Primary Projection and the Eye/Bill Ratio analysis.

No.1  Leg proportions are not strictly measurements because more often than not we have to accept some perspective foreshortening.

No.2  Leg proportions only make sense when we can view our components in profile.  Once again perspective foreshortening is going to mess with our measurements.  It wouldn't for instance be advisable to try and compare the accurate lengths of all four toes based on a photograph.  It simply can't be done correctly.

No.3 Leg proportions are unreliable and at best only an approximation.  I would couch that remark with the point that provided two structures are properly aligned and totally in profile it is possible to make a good comparison.  But for the most part we don't tend to have that luxury so most leg proportions from photographs tend to be estimated at best. 

Gestalt - The Limitations of Bill to Eye Ratio

It is within the subject of gestalt that I tackle the questions of bird size and shape in this blog.

Having drawn some rather pessimistic conclusions about the usefulness of Primary Projection as a measurement directly from photographs I find myself picking on another favourite form of comparative measurement, the Bill to Eye Ratio.  This is the ratio of the width of the eye (front to back) and the length of the bill.  As with primary projection typically this measurement is described in terms of a perfect side profile view.  But, as with primary projection we are not dealing with a straightforward linear measurement and it can be difficult also to accurately find points to measure from because the bill base and occasionally eye can be partially obscured by feathering.  

Typically birds have a conical skull shape, spherical at the rear, tapering towards the bill.  The eyes sit in large spherical sockets and the bill protrudes from the tapered end of the skull.  Vision in birds is very interesting in that more than 50% of birds have two fovea in the eye and so they have a choice of two centres of focus from which to view the world.  Nonetheless most birds tend to look back at the observer or camera with one eye and, presumably mainly sharply focused with the fovea that is located directly opposite the pupil.  To the camera the pupil looks perfectly spherical when birds are looking at the camera in this way, just as human eyes do when they are trained on the camera.  So we might expect then that a perfectly round pupil indicates the bird's head is in perfect profile.  

However when we compare images of the eye and bill at different angles to the camera we can see a problem with the bill to eye ratio.  

When the bill is in profile the eye is not, and visa versa.  As with the primary projection question we are trying to take measurements from a two dimensional projection of objects which are not directly comparable as they are not positioned along the same plane.  

The same caveats apply to the bill to eye ratio as to primary projection:-  

No.1 Bill to eye ratio is not a measurement as such.

No.2 Bill to eye ratio only makes sense when the bird is viewed in side profile.  Determining what the side profile of the head is may be a matter for debate.  The eye and bill are not necessarily in profile together.

No.3 Bill to eye ratio is unreliable and at best only an estimate.

Gestalt - The Limitations of Primary Projection

It is within the subject of gestalt that I tackle questions of bird size and shape in this blog.

Primary projection or primary extension is a form of analysis commonly used in bird identification.  Despite its widespread use it is not a particularly well defined or well understood analytical tool.  Put simply, primary projection is a comparison between the length of the exposed tertials (A-B) on a closed wing relative to the length between the tip of the tertials and tip of the primaries (B-C).  Before I go into detail here I am going to make a few general, rather pessimistic statements:-

No.1  Primary Projection is not a measurement.  Far from it in fact.  It is as close to a comparison of apples and oranges and one can find in birding.

No.2  Primary Projection really only makes sense when we view a bird in perfect side profile.

No.3  Primary Projection is unreliable and at best only an approximation.

Why so negative?  Well when we explore what primary projection actually involves unfortunately what we end up with are a whole load of questions without any answers!

Anatomy of the wing

Despite the fact that we witness the articulation of the joints of the wing in flight all the time we rarely stop to consider how the bones sit and joints articulate when a bird is at rest.  Below I have approximated the position of the wing bones at rest in this juvenile Common Rosefinch, Carpodacus erythrinus.  I have chosen Common Rosefinch intentionally as primary projection would be one of the key distinctions between this old world species and  House Finch Haemorhous (Carpodacus) mexicanus from the New World.

Straight away we can see the problem.  The tertials and primaries are not connected to the same bones and therefore can move quite independently of one another.  If the humerus bone moves on the closed wing it moves the rest of the wing with it.  If the elbow extends the radius and ulna bones then the bones of 'the hand' are also affected.  If the bones of 'the hand' moves only the primaries and/or other feathers of the outer wing are affected.  So the primaries, connected to the bones of 'the hand' can move independently in various planes relative to the tertials.  Clearly, if the tertials and primaries align differently there can be no such thing as a standard primary projection value for an individual bird, let alone a species.  In other words, primary projection is ever changing.

Movement of the forearm
Note how the position of the secondaries in the juvenile Red-backed Shrike (Lanius collurio) below differs in these images from closed in one to fanned in the other.  Meanwhile the tertials and primaries appear to remain much the same in both images.  This suggests an articulation of the forearm, i.e. a movement between the humerus and radius/ulna.  And yet if the 'elbow/lower arm' moves position so must the 'wrist/hand'.  A lowering of the secondaries equates to an extension of primary projection and visa versa.  This point can easily go unnoticed.  Worth pointing out also that the feathers which we refer to as tertials may actually represent different feather tracts in different species.  In many passerines the tertials join the ulna bone together with the secondaries.  In effect, in these species the tertials are actually innermost secondaries.  On many other species the tertials are connected to the humerus bone and are therefore distinctly separate from the secondaries.  This also means that the tertials articulate differently relative to the other flight feathers on different species depending on which bone they connect to.

In Europe two extremely similar Hippolais warbler species Icterine (H. icterina) and Melodious Warblers (H. polyglotta) differ most notably in terms of their primary projections.  Icterine has a long primary projection while Melodious is typically shorter.  Coincidentally, Icterine has a brighter wing panel but Melodious can approach Icterine in appearance, particularly when the secondaries are bunched, as shown by the Red-backed Shrike example above.  Luckily, by bunching the secondaries primary projection is further reduced.  So, overall the ID in this case is made simpler, provided that is that primary projection can be properly assessed.

Unlike the clearly visible movement of a mammalian elbow and forearm, in birds the elbow is hidden beneath feathers and tissue.  It is very difficult to envisage a way to accurately account for a movement of the hidden forearm and thus it's impact on primary projection.  We may be able to qualify a primary projection analysis with the observation that the secondaries look very bunched, thus primary projection is probably slightly shortened from the norm.  That is about as good as it gets.

Movement of 'the hand'
There are multiple joints in 'the hand' of a bird so primaries can be held in various different ways.  Typically, however unless a bird is preening, cleaning or displaying primaries are fairly tightly bunched.  Despite this wings can still lie in a variety of positions from hanging low, almost parallel to the tertials to high, often crossed together over the rump.  Unlike the forearm we may feel we can visualise the arc that the primaries might take from fully closed to partially open, right?  But the problem this time is we lack a frame of reference for the 'normal position' of the primaries at rest.  What we end up with is a range between typically held low to typically held high.

Note the angle of the wing in this case is approx. 45 degrees off of side profile.  This is more in line with the side profile of the wing as opposed to the side profile of the whole bird and I think this is probably the best angle to actually try and measure primary projection properly, if a case could be made for such a measurement.  But, as explained above the frustration in trying to accurately compare tertial length and primary length is in the fact that they rarely if ever line up properly to be measured.  And if this is the case an accurate measurement doesn't serve a real world purpose.  We are left with the realisation that...

No.1  Primary Projection is not a measurement.  Far from it in fact.  It is as close to a comparison of apples and oranges and one can find in birding.

and...

No.2  Primary Projection really only makes sense when we view a bird in perfect side profile.

Angle of view and foreshortening
The angle of the bird and wing relative to the camera is an equally important consideration.  If the wing is in profile we are only considering the X and Y axis, aligned to the plane of the camera sensor.  However if the bird is not in profile then the Z axis (perpendicular to the lens) comes into play.  We also have the problem of perspective foreshortening.  Put simply, we cannot account for the Z axis properly.  The images below are a good case in point.  The left wing on the right hand bird is quite well in profile and the primary projection measurement matches the left hand bird well.  But in the right hand image if we try and measure primary projection from the right wing, which is not in profile the result is very different.  Note the right hand wing neatly illustrates the problem of wing joint articulation.  When we see the wings crossed like this it should be a strong warning that the primaries are at a very different angle to the tertials and that a primary projection measurement will fall at the short end of the range.  However, as we are typically measuring primary projection from a bird in perfect side profile we cannot see what the angle of the primaries are relative to the tertials.  We are limited by our angle of sight.

In other words...

No.3  Primary Projection is unreliable and at best only an approximation.

Judging points correctly
When trying to measure primary projection in photographs observers typically point out that the location of point A at the base of the tertials is a bit arbitrary.  If the bird is in profile typically the greater coverts will obscure the base of the inner tertial.  However, if viewed from above the base of the tertial is visible inside the greater coverts so the tertial length measures longer.  so typically A is defined by the point where the tertial becomes visible past the inner most greater covert.

Solutions
Having thought long and hard about this it pains me to report that I don't think there are any solutions here.   Primary projection is what it is.  It is an interesting comparison that can be made between the visible length of the tertials and visible length of the primaries past the tertials on a two dimensional image of a bird viewed in side profile.  It is not accurate and any attempts to turn it into a scientific measurement are probably pointless.  Above all, it is important not to take primary projection too seriously and avoid the temptation to dive in and measure primary projection from every single image.  Best to take note of the hazards identified above...and probably others I haven't even explored.  Primary projection should be used sparingly and in combination with other identification pointers.

Gestalt - Measurements From Photographs

Measurement by definition requires a numerical scale and a comparable frame of reference.  The physical distance between two points is measured along a straight line using a measurement instrument such as a straight ruler or callipers, or perhaps remotely using a calibrated laser.  What counts in all cases is the accurate scale and an accurate frame of reference.  Photographs present numerous challenges and moreover some of our typical frames of reference in birding are a little questionable at times.

Frame of Reference

Let's take a typical birding scene, a small group of shorebirds, Dunlin Calidris alpina and Common Ringed Plover Charadrius hiaticula.  Let's say we are interested in taking accurate measurements of the aberrant Dunlin and comparing it with other Dunlin in the same image.   

Straight away we have a problem.  In order to take an accurate measurement we need a frame of reference.  Current camera technology can't help us out but let us pretend that we have access to futuristic technology which places a ruler inside an image for us.  The concept is simple enough to achieve, the camera needs to note the distance to our chosen subject and then fancy software will overlay the ruler at that point, while avoiding obscuring anything in the foreground.  The weatherman uses this kind of technology every day so why not us?

This is an appropriate frame of reference for our photograph because our photograph is a flat, two-dimensional representation (X and Y axes) of the three-dimensional world which we are observing.  And, our ruler is located exactly perpendicular to the camera at the location of our key subject of interest.  There is no doubt that this type of technology would be very useful for taking some measurements in the field.  But within this scene, right away we can see we are missing something.  We have a lot of things going on within that third (Z) axis which we are going to have trouble measuring accurately.  Things which we need measuring in front of and behind our frame of reference cannot be measured accurately using that scale.  Only those points which fall exactly along the line of our ruler can be measured accurately against it.  We find a fix to this in version 2.0 of our imaginary measurement software.  Using bird recognition software our camera is smart enough to distinguish a number of subjects and kindly overlays a different coloured grid over each.  Not only that, as the camera has measured the distance to each subject each grid is calibrated, providing four unique frames of reference, one for each subject.  Super!  Right?

Once again, we have improved our range of useful points which we can measure.  But looking closely at each individual bird and grid we find that the grids do not align perfectly along the same plane on each subject.  So why are we still having difficulties?  Well the bird's aren't cooperating.  Our main subject is nicely in profile while the other Dunlin have their back's slightly to the camera and the right hand bird isn't looking at the camera at all.  Therein lies the problem of accurate measurement from a single digital image.

3D Imaging

A quick internet search will reveal various methods to create 3D models from 2D images.  Whether or not this is a practical application for birding remains to be seen and consumer-level technology may be some time away yet.  The basic principal is that multiple images are taken of a subject than stitched together to create a three-dimensional representation.  Such a technique, matched with the kind of sophisticated measurement techniques outlined above would ultimately give us accurate measurement in all three dimensions.  

However if our goal is to be able to measure accurately based on a single moment and we don't want to be reliant on the subject giving us multiple angles, more than likely we are talking about a 3D camera or some other bespoke technology.  A 3D camera is essentially two or more separate cameras, separated from one another by a short distance, triangulating images of roughly the same scene.  While the measurements from this type of setup are possibly not 100% accurate they are far better than the guesswork involving in measuring from a single 2D image.  

Another possible solution would be a 3D laser scanner which would image the terrain in front of our camera and create an accurate 3D model.  The elusive Z axis coordinates are normally recorded by measuring laser "time-of-flight" from leaving the scanner to reflecting off the subject and returning back to the scanner.  One problem with that technique would be the amount of time and file storage required to generate a detailed scan.  Not very practical when trying to photograph a swarm of waders, all constantly moving about, though this technology is improving all the time (eg. see HERE).

Stepping forward again into an imaginary future, I have put my faith instead in a 3D camera setup involving three separate cameras placed approximately 30cm apart.  They stitch together three images into one and I interrogate them using sophisticated software for an accurate bill length comparison.  Some day...

For more on the challenges of measurement from photographs see HERE.

Human Bias - Tonal Gradient Illusions

In recent postings on this blog I have been interested in the causes and control of tonal variation.  What really started this thread was a look at Grey Scales and Gulls.  Aside entirely from the complexity of trying to accurately capture and reproduce exact tones, we also have various different elements impacting on the uniformity with which tones are distributed across an image.  

The properties of light

We know that the angle of a surface relative to the light source impacts on surface illuminance as explained by Lambert's Cosine Law.  But the lighting in a scene comes from multiple directions.  On a bright sunny day the lighting in an image is often overwhelmed by uni-directional sunlight.  In the shadows we may detect a bluish tone and this is due to blue light from the sky dome.  On an overcast day the sun's influence is diminished.  Sunlight is scattered by cloud and arrives at the ground from all directions as diffuse light.

The properties of surfaces and perspective

Surfaces are rarely uniform and may consist of a combination of specular highlights and more diffusely-lit areas.  Surfaces may also vary in terms of reflectance and absorption of light.  But even if objects are considered virtually identical and are carefully aligned in front of the camera, they rarely if ever look identical in the final image. Unless a surface is perfectly Lambertian light will not be reflected evenly in all directions, and so the angle of view of the camera relative to each subject comes into play. Tones vary slightly even between identical subjects, simply because of perspective.

The properties of the lens

To further compound the problem we have artefacts introduced by the camera including the 'Cosine Fourth' Law of Illumination Falloff and other forms of vignetting.

The properties of the human visual system

Last but not least we turn to the human visual system.  Consider the following tonal optical illusion.  

If we are interested in studying and comparing tones across an image this type of illusion is worrying.  While many optical illusions might not appear to have practical consequences I think this illusion is particularly relevant to birders.  The three elements that make up this illusion are the presence of a tonal gradient, a regular pattern of objects and a contrasting white or bright background.  As birders we encounter these three elements together in almost every one of our field experiences.  Lets replace trees with feathers and repeat the experiment.

It is sometimes stated that the Checker Shadow Illusion and similar computer generated graphics eg. used by the Lottolab Studio could not be replicated in nature because lighting and surfaces are more variable in nature than in the optically perfect, Lambertian, theoretical computer-generated world.  I am not so sure personally.  I think these illusions probably happen all the time but are simply harder to detect in nature than in these computer-generated examples.

For more see HERE.

Birds and Light - Lighting and Perspective (Part 2)

A rainbow is a reminder to us of the majesty of light.  As kids we try to find the rainbow's end until we come to realise that the rainbow always stays ahead if us, always positioned opposite the sun.  We then begin to realise that our individual perspective on the world matters.  This particularly luminous rainbow was photographed against a dark cloud, which also provided a convenient neutral grey for white balance correction.  I gave the image added saturation to punch out the colours.  Alexander's band to the left is a darker, unlit area of the sky between the primary and secondary bows.  The portion of sky inside the bow has added illumination, hence the contrast between inside and outside the primary bow.  At the outside red light at a wavelength of approximately 700nm fades off to invisible infrared, while inside violet light of wavelength roughly 400nm fades to invisible ultraviolet.  A camera's sensor can detect both IR and UV but these are filtered in most cameras before they reach the sensor (for more on IR and UV see HERE).

As an interesting aside there appears to be one or two very faint rainbows offset from the primary bow on the inside of the arc in the image above.  These might be an example of an uncommon phenomenon called supernumary rainbow, caused by wave interference, or possibly even triple-split rainbow which is rarely seen or photographed and only explained by science in the last couple of years as being due to the presence of non-spherical rain drops.  These faint bows were not obvious to the naked eye and may have been expressed as a result of increasing image saturation.  It is sobering to read that there are still new discoveries being made about rainbows - one of the most recognizable natural lighting phenomena.  Not surprising perhaps therefore there are lots people still don't know or take for granted when it comes to the subject of birds and light.

  
Earlier, on Lighting and Perspective
In Lighting and Perspective Part 1 I mainly focused on human binocular vision and it's implications for how we perceive the world versus how cameras record it.  I also carried out a specific lighting experiment with multiple identical multi-faceted targets to illustrate a point that subtle variations in the angle of the camera versus the subjects and light source affect the lighting and therefore the appearance of each subject.  I thought it might be useful to carry out a few more similar experiments, this time using more complex, oval and spherical shapes and to consider the possible impacts of different variables.

Right away I need to confess that I don't think I will ever totally understand and nail natural lighting in all it's complexity.  With each little experiment comes a new insight.  That is part of the challenge but also part of the real enjoyment of trying to tease out this complex subject.

Lens Comparisons

I have taken two lenses which I think most birders who use a DSLR are likely to have and use in the field.  The first is a typical camera kit lens, which often tends to be a multi-purpose zoom lens.  The kit lens of the Canon D70 is a wonderfully flexible 18-135mm, ultra-quiet zoom.  At full zoom 135mm the image is reasonably rectilinear but vignetting starts to creep in very slightly at the corners.  At 18mm the lens has a strong barrel distortion but vignetting seems to be under control (possibly due to curvilinear distortion).  The other lens I use is a 300mm fixed lens which is quite rectilinear and has no obvious vignetting.

In the first experiment I took five identical plastic eggs (raiding the kids' toy chest once again).  I arranged these symmetrically and glued them to a board.  I also placed a grey card in the scene and then took multiple images in low, diffuse evening light, spot-metering to the grey card to try and obtain consistent exposures with different lenses.  Obviously I had to adjust the distance to the subjects as I changed the lens or alternated the zoom setting.  The purpose of the experiment was to see if the targets remained similarly lit in each image and if there were any obvious differences based on lens type or focal length.

Once again I have used the versatile freeware Color Quantizer to postarize then artificially re-colour individual tones in order to map tones.  Note my choice of colours for each tone is purely arbitrary.  I explained how I did this in an earlier posting (HERE).

The most obvious thing is the major difference in perspective foreshortening between 18mm and the longer focal lengths.  The barrel distortion created by the 18mm lens is also quite striking.  The next point to make is that I didn't obtain a comparable exposure with the 18mm lens as evidenced by the paler-looking grey card.  I suspect this is due to the angle at which the grey card was lying.  So it wouldn't be fair to compare the 18mm image with the other two directly.  Obviously there would have been a delay of a minute or two between images as I needed to change lenses and realign the camera and scene between shots.  So lighting may have changed slightly between images.  Though I did pick a dull, low light and used the grey card for spot-metered exposure to try and iron out these differences.

On first impressions all targets within the same image are strikingly similar and this seems to contradict the findings of my earlier experiment HERE.  However a closer analysis of the images shows that indeed all subjects in all of the images have subtly different tonal maps.

On balance the type of lens doesn't appear to greatly influence the degree of difference between how subjects are lit and appear in the image, notwithstanding barrel distortion, vignetting etc.

Subject Size

In an experiment like this I considered that the actual size of the targets might influence the results because larger targets means working with greater distances from the camera and larger surface areas.  That said, relative angle of view doesn't change from before.  For my next experiment I switched to larger, and also more uniformly spherical targets.  This time I dispensed with the 18mm lens image and just compared 135mm and 300mm focal lengths.  The results show an even closer match between the 135mm and 300mm images than in the earlier experiment, though this may just be coincidental.  Once again, though superficially very similar looking, on closer examination all targets have a unique tonal map.

  

Bright Light Versus Dull Light

With the next experiment I was keen to see if there would be a difference between how the experiment would perform in low or diffuse light versus high contrast sunlight.  I was fortunate to have fast moving clouds and sunny spells so I was able to make this comparison happen in a matter of seconds.

In bright light the camera's dynamic range is challenged, contrast is increased and tonal range is reduced.  There was a small amount of clipping at the white end of the tonal range but other than that I was able to obtain a reasonable comparison.  It is interesting as an aside to see how an out of shot object managed to catch one of the targets with a long shadow in the sunlight image but the subject was entirely unaffected by this object in dull light.  One further consequence of photographing in bright sunlight I guess.  In this limited experiment I found that the more brightly lit scene produced more consistent-looking subjects.  Whether this is simply due to the reduced tonal range or a real effect of strong lighting I can't say.  In the lower light image there is considerably more variation.  

Point Source Versus Diffuse Lighting

What has come as a real surprise is that the brightest point (roughly at the centre of our concentric rings of colour) on our spherical targets above appears to change from a position of roughly 45 degrees on the sunlit image to a more vertical centre on the cloudy image.  The only possible explanation for this that I can account for is that the overall lighting on the subject may be shifting from one predominantly point source lit on the sunlit image to a scene predominantly ambient lit (i.e. lit by the sky dome) on the cloudy image.  

Further evidence for this comes in the image below where I photographed a number of targets from above.  In the diffusely lit image the centre-point in each target appears to shift with the lighting of the left hand targets pointing to the left and the right hand targets pointing to the right.  In stark contrast, with the point light source the lighting is unidirectional.  This finding has implications where we are trying to gauge lighting direction from images (for more see HERE).

Note how the shadows in the upper image appear to fall towards the centre of the image.  This is Lighting and Perspective in action. I think I am finally starting to grasp some of this subject.  Please expect a Lighting and Perspective Part 3!