We spend all our waking hours in our own company. Not surprisingly we come to see ourselves as unique and special. Self-serving bias is a form of self-preservation and is a very significant bias in our decision-making processes particularly when it leads to irrational and entrenched modes of thinking. It is the tendency to see oneself as responsible for desirable outcomes but not responsible for undesirable ones. Self-relevance effect is the unsurprising tendency to recall facts referring to oneself better than facts referring to others. Placement bias is a tendency people have to rate themselves as better at the skills they believe they excel at then they actually are and to rate themselves poorer at the skills they believe they suck at then they actually do. Risk compensation is a tendency to take greater risks when perceived safety increases. Thestatus quo bias and system justification bias are tendencies people and groups have to want things to remain the same.
Delusional behaviours
Subjective validation is the perception that something is true if a subject's belief demands it to be true. This can lead to perceived connections between coincidences. The overconfidence effect is excessive confidence to one's own answers to questions. Pessimism bias is somewhat the opposite. The Ostrich effect is the tendency to ignore an obvious, negative effect. Bias blind spot is the tendency to see oneself as less biased than other people or seeing more biases in others than in oneself. Sometimes we imagine there is greater support for our position than there actually is (the false-consensus effect). The forer or barnum effect explains the belief that people have in astrological predictions as applying to themselves when they are written to apply to almost anyone. The hot-hand fallacy applies to people who believe they are on a winning streak when there is no statistical significance in their gambling. The illusion of control is the belief we can control external factors we have no control over.
Entrenchment
Choice supportive bias is the tendency to see one's choices as better than they actually were. Bayesian conservatism is the tendency to insufficiently revise one's beliefs when presented with new information. Backfire effect is when people react to disconfirming information by strengthening their beliefs. Irrational escalation is the phenomenon where people justify increased investment in a decision based on an accumulation of prior investment, even if new evidence suggests the decision was probably wrong. Loss aversion is a tendency to prefer avoiding losses to acquiring gains. Post-purchase rationalisation is the tendency to rationalise a purchase as good value. Reactance is the urge to rebel against a perceived constraint of one's freedom to choose. Reactive devaluation is the devaluing of proposals from someone we don't like.
Group Think
Call it group think, bandwagon effect, or herd mentality we all have a tendency to follow the crowd at times.
Of all the different types of bias, we may be more aware of how fickle and biased our memory can be than any other conscious form of brain functioning. A memory bias is a cognitive bias that either enhances or impairs the recall of memory. Memory is highly selective. Memory Formation
There are different ways of encoding information to memory and how a particular memory is encoded will determine how accurately it is remembered. This is the levels-of-processing effect. False information or misinformation can affect people's reports of their own memory (misinformation effect). Missatribution and source confusion refer to situations in which memory detail is retained but the source of the memory is not or cannot be later recalled. That snippet of information may become associated with a wrong event in time. Suggestibility is a form of misattribution where ideas suggested by a questioner are mistaken for real memories. A false memory is misattribution where an imagined event is mistaken for a memory.
The picture superiority effect says that concepts are more likely to be remembered experimentally if they are presented in picture form. This might be one of the reasons I like to present my thoughts in that way, for example on this blog. There is a link between emotion and memory. The bizarreness effect is one where bizarre material is better remembered than common material. This is similar to the humour effect, where perhaps the distinctiveness and the emotional stimulus associated with humour make it more memorable. The Von Restorff effect states that an item that sticks out is more likely to be remembered than other items.
Emotions and perhaps memories associated with unpleasant events tend to fade faster than those associated with happier events. This is the fading affect bias. This might also explain the positivity effect and rosy retrospection, where older adults favour positive over negative information in their memories. One's current emotional state can also affect one's ability to recall information (mood-congruent memory bias). Information which is heard frequently is likely to be remembered as fact even if it is fiction. This is the illusion of truth effect, Self-generated information is more easily remembered than information which has come from another source. This is referred to as the generation effect. Illusory correlation is inaccurately remembering a relationship between two events. Interestingly it has been found that tasks which have been completed are remembered less clearly than those which are uncompleted or interrupted (Zeigarnik effect).
Memory Modification and Recall
Levelling and sharpening are memory distortions that may coexist. Sharpening is the selective recollection of certain aspects which then become exaggerated or take on a greater relevance than aspects which have been levelled and lost. The repeated retelling of the memory may further exaggerate the effects. Seems akin to repeatedly re-saving a lossy JPEG image file. Verbatim effect is the tendency not to recall an exact wording but the general gist or context of the memory. Tip of the tongue phenomenon, where we know a memory is right there but we cant quite grab it, is believed to be due to multiple related memories blocking and preventing the memory we want from being retrieved.
The peak-end rule is another interesting aspect of memory recall. We tend not to recall the sum of an experience but the average of how it was at it's peak and also how it ended. We have a greater tendency to recall items near the end of a list (primacy effect/recency effect/serial position effect) which is why it is always handy to be at the end of a job interview process. Conservative or regressive bias affects memory by remembering higher values and higher probabilities as being lower than they actually were. Consistency bias is incorrectly remembering one's past attitudes and behaviours as resembling present attitude and behaviour. Egocentric bias is a recall of the past in a self-serving manner. The hindsight bias is the inclination to see past events as being predictable. Memory is often associated with context and is often retrieved by triggering contextual memory pathways. This is the context effect. Telescoping effect is the tendency to displace recent events backwards in time and remote events forward in time.
All these biases might paint the picture that memory and recall is totally unreliable. There are tricks to improving and perfecting memory, including the testing effect - frequent testing of information that has been committed to memory improves memory recall. The spacing effect refers to one such mode in which information is better recalled if exposure to it is repeated over a longer span of time. The method in which information is received can affect how it is retained eg. whether heard, viewed or read (modality effect) which is why writing down information can aid in consigning it to memory ... possibly worth combining with a sketch and a bit of humour!
When it comes to identification there may be a whole range of questions that need answering and these answers are rarely simple. Anyone who studies biology will know that there can be great variation for any given trait. Thus bird identification is complex and very often open to debate. When it comes to evaluation of the evidence we clearly have a lot to consider. It doesn't help that we can be distracted from our goal by inherent biases in our patterns of thought.
Some memory and availability biases in evaluating evidence
Some biases and heuristics are caused by recent memory. Availability heuristic/recency bias are tendencies to overestimate the likelihood of events due to their greater availability in memory. Availability cascade is a reinforcing process where an idea gains credence simply by being repeated - 'if we say it enough we will eventually believe it'. This can be an example of group think - eg. on a social forum. Our memory can be subject to confirmation bias, where we tend to remember information in a way that confirms our preconceptions, possibly even ignoring or suppressing information that may be at odds with our conclusions. Sometimes if we have recently seen a word or object, all of a sudden it appears to be everywhere - we might imagine it's frequency has increased but really we are just more aware of it than before (frequency illusion/observation selection bias). Hindsight bias is the tendency to see past events as being predictable. Distinction bias is a tendency to view two options as more dissimilar when viewed side by side than when evaluated separately.
Some belief biases in evaluating evidence
Regardless of the logic of an argument or the statistical probability behind it, we can be swayed by whether or not we actually believe it to be true or not (belief bias). Selective perception is the tendency for expectation to affect perception. Sometimes we may be biased by how hard or easy an identification might appear (hard-easy effect). Restraint bias is the tendency to overestimate one's ability to show restraint in the face of temptation. Semmelweis reflex is the tendency to reject new evidence that contradicts a paradigm. The zero-risk bias is a preference for reducing a small risk to zero over a greater reduction of a larger risk. In terms of an identification based on a collection of variables I guess this might equate to spending too long on a fairly irrelevant part of the puzzle because it might be considered low hanging fruit (like unit bias - the tendency to want to complete a unit task). Thereby we miss the bigger picture.
Some biases in evaluating statistical probability
When faced with an image of a bird it is human nature to ask, is this species likely in this context. Neglect of probability is a tendency to completely disregard probability when making a decision under uncertainty. On the other hand, disregarding a possible answer due to low probability can also be problematic. Some of us suffer from exaggerated expectation while others suffer conservatism or regressive bias - the tendency to underestimate high values and likelihoods while overestimating low ones. Some people exhibit base rate fallacy, or the tendency to ignore the first choice. People often see patterns in occurrences and draw conclusions though there may be no statistical significance (clustering illusion). A gambler's fallacy is the tendency to think that future probabilities are altered by past events when in reality they are unchanged. The illusion of validity is the belief that further acquired information generated additional relevant data for predictions, even when it evidently does not. The subadditivity effect is the tendency to estimate that the liklihood of a remembered event is less than the sum of its (more than two) mutually exclusive components.
There are certain biases that distract us, whether we are considering evidence from our own original field observations, a written description, or images taken in the field. These include tendencies towards anchoring or focalism on certain aspects of an identification at the expense of others and to have our attention drawn to dominant stimuli. We are liable to be influenced by contrasting evidence, be it in life, or based on documented evidence. A decoycan throw the identification and that decoy might be an artefact posing as a field mark for example. The curse of knowledge, our tendency towards subjective analysis based upon our existing knowledge, may work similarly for or against a correct identification, whether it is based on a field observation, or based on analysis of a description or images. Bare in mind though that while we may be very familiar with a species we see every day, we may never have written or read a description of it, or seen a photograph of it in flight. There may be anillusory correlation made between objects or subjects just because they exist in the same space. The next-in-line effect may be a form of memory distraction where a person in a group has diminished recall for the information they heard directly before and after they themselves spoke. Another is the part-list cueing effect, where being shown part of a list of items makes it hard to retrieve the other items from memory. Pareidolia is a human tendency to create patterns where they don't exist. Once again a well placed image artefact may be mistaken for a field mark if we try hard enough!
Some of you may have seen this video before. Please try to pay attention...
Okay so our brains are not wired to be able to pick up on everything all at once. This makes the challenge of identification in the field all the more difficult. Surely though, when faced with a set of digital images we wont be so easily fooled. The next time you are presented with a set of images for analysis I encourage you to write down your initial impressions and what prompted them. Then write down your analysis in the order in which you analysed the images, the tools and techniques you used and what conclusions you reached during the process. This may provide you with an insight into how your own cognition works and if you are led to progress identifications in a certain way.
This is the first of a number of postings looking at different types of cognitive biases. As I approach an end to the series I hope to compile a simple tool that we can use to circumvent cognitive bias to keep us on a path to the correct identification of birds from digital images.
Consider a pathway following an image formed by light leaving our subject. It passes through a camera lens and is transformed into a digital camera image. A human, viewing the image analyses it's content with care and uses an array of tools to validate a firm identification based on the available evidence. This is the very pathway we are exploring throughout this blog.
All along this pathway there are parameters that can affect a favourable outcome - the correct identification of the bird in the image. We have the composition of the image, consisting of the posture, angle of view and the terrain in which we find our subject. We have environmental considerations such as lighting, temperature, moisture, dust and pollution in the air. We have the qualities of the lens and camera sensor and in-camera settings. We have image processing parameters and the image viewing platform, be it the screen on the back of the camera, a computer or phone screen or a print. The image has traveled a long way along our path...but we are only now beginning to ask what it is we are looking at. The remainder of the route to the identification of the subject from the photograph is strewn with an even greater array of complex parameters which we will collectively call human bias.
Up until now the blog has been devoted to reviewing the various parameters that go into creating a digital image. I have also started looked at tools and methods for forensically analysing digital images, so that we can overcome some technical limitations and peer into the finer details of an image. Along the way I have touched on certain aspects of the human condition, including aspects of the design of the human visual system. I am now going to broaden this out to cover a much wider array of parameters relating to human vision and cognition.
The prevalence of human cognitive bias
Cognitive Science is a broad discipline looking at a whole range of fields relating to the human mind. We don't often stop to consider the functioning of our own minds, how we receive and record sensory information, analyse and transform it into logical thought. It probably goes without saying that this is of far more importance than any other when it comes to the actual identification of a bird. Though the first bird identification software has started to emerge, when it comes to difficult bird identification puzzles it is probably safe to say we will always be totally reliant on human cognitive skills. How much of our thought process is deliberate, objective and logical and how much is based on automation and subjective reasoning. Well there seems to be growing evidence that a lot of our decision making is not logical at all, is based on hard-wiring, emotion and memory.
Take the following hypothetical scenario. Three small birds fly past an experienced observer along a street in a busy city. The observer raises a camera and takes one photograph of the three birds as they pass. Before the observer has had a chance to check the camera, we ask the observer what the birds were and the answer is a confident "they were House Sparrows" (Passer domesticus).
We ask the observer to write a description detailing their observation. The description is very simple and honest. "Three birds flew past which were easily identified as House Sparrows. There are always House Sparrows on this street. The identification is based mainly on gestalt and on some detail observed on one of the males as seen through my camera lens as the lens focused on the subjects". The observer says they are 100% confident of the identification (though quietly might like to check the photograph for total confirmation). Writing the description, the observer remains confident though is somewhat more aware of the lack of strong evidence as they write. There are a great many other species that could be involved. The observer is mainly 'playing the odds'.
Though this is just a hypothetical scenario, we have probably all experienced this type of situation. Cognitive bias sets us up for these fails all the time and it can take a lot of discipline to really learn from these simple experiences. If the observer were being critical of their observation the correct answer would have been "I am certain one bird was a male House Sparrow as I observed it clearly enough to ID it properly. The other two birds were not seen well enough to identify them correctly so I will leave them unidentified". How many birders do we know who actually talk like that? The fact is, life's experiences encourage us to learn and use short cuts or heuristicsas we grow and develop...close enough is usually good enough.
There are many different biases at play in this example. We have the observers prior experiences (curse of knowledge bias). The observer has evaluated that, based on past experience, 'there are always House Sparrows and only House Sparrows on this street' (attentional bias/availability heuristic/availability cascade/belief bias/congruence bias/illusory correlation/levelling and sharpening effects). In reality, the observation was too brief to make a firm judgement but this fact has been ignored. The observer identified one House Sparrow so this appears to confirm the initial opinion (framing effect/anchoring bias/confirmation bias/decoy effect/illusory correlation/survivorship bias/insensitivity to sample size). The observer possibly wasn't heavily invested in the observation so didn't take the time to look carefully enough or verify their initial assessment. However, on being asked what the birds were the observer is now forced to become more invested in the event. The opportunity for objective reason has been missed - instead the oberver digs their heels in (irrational escalation/overconfidence effect/optimism bias). On being asked to prepare and consider a description, the observer might have felt some doubt creeping in, as there really wasn't a lot to go on to make a firm ID (choice-supportive bias). There were a lot of inferences that needed to be made from the available evidence and the observer has either chosen to ignore that fact or filled in some of the gaps with false memory (including missatribution).
A photograph that might help clinch the ID is now presented. The moment of truth. The photograph confirms that there were in fact two male House Sparrows but the third bird was actually a male White-throated Sparrow Zonotrichia albicollis. The observer remarks - 'A simple error, no biggy'. 'We all make mistakes right'? Already the observer is moving on from the event, choosing to maintain the status quo in their cognitive pathways rather than challenge them (cognitive dissonance).
The example only serves to illustrate some of the underlying processes that might be going on during our thought processing. These cognitive biases and heuristics should not be seen as an entirely negative thing. After all, they underpin our daily lives and exemplify our incredible abilities to navigate life as independent, successful sentient beings. But, if our goal is to analyse images in a entirely objective and unbiased fashion then we have to consider how stimuli can trigger certain automated responses and thought patterns, and it might help to find ways to circumvent them.
This blog was started in February 2014 to scratch a persistent itch of mine. I had always felt that birders needed a resource like this and I felt it was a good time to have a go at creating it myself, while the kids are still small and I am at home much more than I am out birding. For me this has been one of the most rewarding of my birding years. Stopping to take stock of 30 years of field experiences, and revisiting my collection of images and videos with a renewed vision and purpose has provided the greatest thrill.
The top three highlights of the year in the blog were...
In 3rd Place
I felt I needed to start the blog with something tangible and useful. The Image Quality Tool set a good foundation for much of the next few months and it provided the impetus for me to keep up the effort and interest. It helps objectively score the quality of an image based on five key quality parameters, namely resolution, focus, exposure, colour and artefacts.
In 2nd Place
While studying the broad and challenging subject of colour in birds I started reading about ultraviolet. On a hunch I purchased a Baader-U filter and put it together with an old Sony digitial-8 camcorder I had in storage. The result is a relatively inexpensive and effective UV reflectance imaging system which might bring this cutting edge science more into mainstream birding. Some discoveries may be mundane (like the fact the Common Moorhen's bill tip is more reflective in VIS than in UV), but no doubt there are plenty of more useful discoveries to be made. While UV patterns in birds may be subtle, particularly in this part of the world, I have found this effort worthwhile, not least for the insights into floral nectar guides and butterfly UV patterns.
In 1st Place
I was quite nervous and reluctant to take on the challenge of colour in birds. While there was a great deal of useful research material and resources to tackle colour theory and colour management, there were a few notable gaps that needed attention. For a start, I could find no useful technique to sample colour patches from images. This was relatively easily overcome using a postarizing method. The biggest surprise was the total lack of a colour nomenclature for the digital sphere (i.e. the sRGB colour space). After much effort it was a big disappointment not to be able to revive Ridgway's Color Standards and Color Nomenclature in sRGB - Ridgway's iconic standard exceeds the gamut of sRGB and therefore can never be fully illustrated online. But this effort all helped crystallise the subject matter and I did eventually produce what I think is a good practical Birder's Colour Pallet - illustrated below.
White Balance is another aspect of colour which I have always enjoyed exploring. The X-rite colorchecker passport was a very worthwhile purchase. Having calibrated my camera gear with it by creating a DNG profile, I then used it to create a special rig to put lighting under the microscope. The standard use of the colour checker is for grey card exposure and white balance correction.
I may have shown my geeky side a bit through the use of various experimental rigs. A lot of this work could probably be better explained through the medium of mathematics, but maths isn't my speciality. I also like the simple effectiveness of visual experimentation and illustration. Anyone can replicate and build on the ideas in this blog without requiring a degree in maths or science. I encourage anyone with an interest to challenge and develop upon my findings. This is a journey of discovery for me too and I love to get your constructive feedback.
The Quick Reference Guide is a good summary of the progress so far with the blog. Some might find this pdf format easier to navigate than the blog. The content can also be effectively navigated using the page links on the top right of the blog.
2015 Exposed
I have a lot of ground covered but I estimate I am, at best half way towards my objective of researching and presenting the main aspects of identification of birds from digital images. So what are the main objectives for 2015?
Birds and Light
In a nutshell, this series of postings has been all about the lighting conditions under which we observe and photograph birds. Like most aspects of the blog, this section is coming together organically. Having already featured lighting at sea, on snow and ice and in arid and semi-arid areas, I hope to include some more 'special environments' where lighting plays a key role in observation and photography. There are also other various loose ends to tie up, but this series is nearing completion I think. I hope to produce a handy summary guide when I'm done. Even if, as a birder, you are not into photography, this content should certainly resonate with anyone who spends time watching birds in the field.
Forensics
I am going to keep working towards a Forensics Manual throughout the year. I will be trying to put some order on things in the near future and may start to pull together some sample analysis in a standard format so that it is possible to see where the various forensic tools start to fall into place. I have no doubt this aspect of the blog may be boring for many of you but for me this is the heart of this whole exercise. Hopefully the value of it will be clearer in time. I must stress that this is very much an exercise in trial and error. I have no training in digital forensics, nor access to a wealth of research material. This is purely a case of getting to know the existing software tools, their uses and limitations, identifying challenges and trying to find novel ways to overcome them.
Most birders, myself included, are probably content to analyse most bird images using a web browser or a simple photo viewer with only a zoom tool for closer analysis. How many of us actually take the time to download an image and open it up in Photoshop or a similar package, to analyse it further? Most of the time a web browser or image viewer may be sufficient for ID purposes. But when the ID is a challenge it certainly helps to apply some form of forensic analysis. Learning what tools to use, when to apply them and by how much is what forensic image analysis is all about.
What I am working on ultimately with this series of postings is a guide to help with and encourage a more critical analysis when one is required. Wouldn't it be great if a forum existed where contributors could put up bird ID images for analysis and others could take the images,work on them and resubmit together with their analysis. I think that could be very instructive for birders learning how to ID birds from photos. I am happy to facilitate that through this blog if there is an appetite for it. For now I am just going to keep working on useful tools and gaining insights ... like the example below, illustrating the intrinsic mechanism behind blown highlights in digital images.
NEW* - Human Bias
Identification of birds from images as we know doesn't just rely on an understanding of image quality parameters or the ability to use forensic tools to interpret images. We all bring our individual experiences to the table and we must use our own judgement and skill in the end to make that ID call. This is all part of the appeal of mystery photo challenges.
Sometimes the less knowledge we have of the circumstances of a photo the better placed we are to make a good call. The urge to latch on to certain clues at the expense of others can sometimes be overwhelming. Context can throw an identification way off course. Non-birders often try to describe a puzzling bird in terms of common birds they know. The term "like a Sparrow" can mean a small bird or a large bird with sparrow-like plumage. A Wood Sandpiper Tringa glareola standing at point blank range out in the open in a farmyard can throw even the most experienced birder, more accustomed to seeing this species in a wetland at far greater distance. Occasionally those studying photos for the first time can even find themselves at an advantage over those who have spent time actually watching a bird in life! I want to explore a series of tools and techniques to counteract human bias. I hope to get started on this in the early new year.
NEW* - Field Marks Under the Spotlight
I will be putting together a new series of postings in 2015 looking specifically at field marks and how they are presented in digital images. There are phenomena that introduce false field marks and those that remove real ones from images! Take for instance, the subtly darker lateral crown area in Booted Warbler Iduna caligata. What are the conditions that obliterate this kind of subtle field mark and when is it enhanced in a digital image?
More importantly, when can false field marks like false feather fringes, false scalloping, false streaking or more complex patterns, and false colours all begin to manifest in our images and throw an identification?
Your Input
I estimate that about half of this years postings came about due to interesting questions and puzzles raised from emails and from forum discussions. I have no doubt this will also be the case in 2015. If anyone has a challenge they would like to pose feel free to drop me an email.
It is hard to imagine we are about to reach the shortest day of the year here in the Northern Hemisphere. This winter so far has been another exceptionally mild one here in Ireland. Garden flowers which should have been burnt by frost by now have continued to bloom here on the south coast. Some confused birds here have even been breaking into song throughout the winter. At the same latitude as Newfoundland, Ireland is pretty far north but is warmed by the gulf stream from the Caribbean. With an oceanic climate, weather influences mainly from the Atlantic and a prevailing southwesterly wind, our winters tend to be wet and mild.
Ireland's Sunshine
At this time of year light plays as big a role as weather in the quality of bird images. Mid-winter sunlight has been playing its usual tricks and websites and forums have a share of poorly white balanced images as a result. I never tire of revisiting the images below which illustrate just how poor the light quality is at this time of year. Light temperature is in a state of constant flux here in late December, whereas in late June it is stable for most of the day. White balance correction is a must at this time of year. More on white balance HERE.
Winter Ultraviolet Light
Unfortunately the uvawareness.com website seems to be having trouble at the moment, but having been watching the UV index over the winter, the index throughout these latitudes remains permanently low in winter. I recently took my UV reflectance camcorder outside and was surprised to find that it could still generate usable images, even in this low ambient UV. More on UV imaging HERE.
When you click on the link you will be directed to the screen below. Make sure to click on the download icon to download and interact with the pdf file.
Once downloaded you will be able to navigate through the manual as illustrated below.
In An Introduction to Gaussian Analysis (HERE) I outlined how most image quality parameters exhibit a Gaussian distribution around an optimum quality standard. This presents an opportunity to look for 'Gaussian Signatures' left behind in digital images. Here I am analysing the Gaussian signature for the fifth of our five Image Quality Parameters, namely Artefacts.
First off, I would recommend having a read of the page HERE to become reacquainted with the various common artefacts, including my individual take on what is and what isn't an artefact.
The Gaussian Signature for Artefacts
A number of artefacts are directly associated with our other image quality parameters. Image Quality itself becomes the signature for these artefacts as a whole and the Image Quality Tool can be used to provide a rough measure for these artefacts. I.e., the poorer the quality, the greater the artefacts. There probably isn't a whole lot more to be said here, other than to look at those artefacts which we can associate with our image quality parameters, as compared with those which are unrelated.
As we can see, the frequency or extent of certain image artefacts is linked to image quality. On one side of the graph we have Moiré, Defocus (inside), Camera Shake, Underexposure, Noise (frequency of). On the other, we have Defocus (outside), Camera Shake, Overexposure and Blooming. All of these have one interesting thing in common - they are all, to an extent, within the control of the photographer. By concentrating on obtaining a detailed, sharp and optimally-exposed image, quality-related image artefacts should take care of themselves. On the other hand, where image quality is compromised, this issue will be further compounded by the introduction of artefacts.
When one lists and considers all of the remaining common artefacts we can see that most of those are uncontrolled by the photographer and indeed they are also difficult to rectify at the photo-finishing stage. There isn't a whole lot we can do about these other artefacts, and we don't have any useful way to measure or estimate their likely impact on image quality.
When we are out birding, most of the time we are not trying to make subtle judgements about plumage colouration or tone. But occasionally we are posed with a question like, "which species has a darker mantle in this flock of terns?" Just how well equipped are we to make an accurate assessment?
Predatory animals and for example primates have forward facing eyes which provide a three-dimensional, Binocular view of the world. Benefits include the ability to judge distances extremely accurately. It takes a lot of skill and coordination for instance to hit a bulls eye on a dart board or score a 3 pointer in basketball. From an evolutionary perspective this form of vision obviously helps with hunting, navigation, foraging etc. But there are also some limitations introduced by this method of vision.
If you have looked at a 3D television without 3D glasses you will have noticed an annoyingly, blurred double image. This is like the image that our eyes actually produce and can also be seen by intentionally crossing ones eyes. Our left eye sees the world from a slightly different perspective to that of our right eye. This is obvious by alternately opening one eye and closing the other. Our brain takes these two perspectives and recombines them to create one sharp, 'properly aligned' image. In doing so, the brain must actually distort the lines of the image to match up these two different perspectives. This is a distortion of reality and therefore should be seen as a limitation, not an advantage in terms of forensic analysis.
By comparison, a camera has only one perspective. The lines captured by the camera are therefore more accurate than those perceived by human eyes. Normally we are totally unaware of this difference because straight lines are not very common in nature. But when we photograph in the built environment for example we are immediately drawn to this perceptual differences between the human eye and the camera image. Take this image below of Adare Manor in Ireland.
The lower left image accurately depicts the perspective and depth of the scene, given the position of both the observer and camera relative to this tall, impressive building. Note how the left edge of the building is not vertical - this is correct in terms of normal perspective. Human vision looks more like the image on the lower right. Our vision tends to distort the environment around us, making lines straighter and more vertical in appearance. This provides an advantage in terms of spatial awareness and navigation, but it is a distortion of reality. Note a digital image may also suffer from lens distortion. It is important to check for this anomaly rather than assume that the camera is always correctly recording angles and proportions (see HERE).
This series of postings, Birds and Light, is all about light. So what is the relationship between perspective and lighting one might ask? Well, because our eyes distort perspective, we don't often appreciate the difference a subtle angle change might make to lighting in a scene. And, because our eyes are effectively seeing two different perspectives, our eyes are seeing two subtly different lighting regimes. We may sometimes be surprised or even confused by the lighting in a scene. Consider the experiment below.
I have printed out a regular grid of rectangles. Each column is identical. Each row consists of rectangles offset by 15 degrees from the previous row.
Having printed this grid I now look down the line of the page from right to left. This is what my eye's see.
At the moment I am looking at a flat grid. I need something three-dimensional in order to appreciate the effects of ambient lighting on what I am seeing. On top of each of these rectangles I have glued a target. Each target is identical, consisting of a disk with a hexagonal shape protruding from it. Each target is centred and angled exactly in line with the grid box upon which it sits so that all the targets are angled to match the grid.
I have taken a series of photographs of this target grid. Viewed from above, these targets have been laid out in the regular pattern shown above, and everything looks very uniform. From the perspective of the camera looking down the line of targets from a slight elevation, these targets actually all have subtly different angles and and analysis of the tonal distribution on each of the targets shows they all have subtly different lighting signatures as a result. No two targets look exactly the same.
To the eye there may be just a subtle tonal gradient apparent between the front and back rows. For a clearer picture of what is actually going on I have used a very handy software program called Color Quantizer to postarise and then recolour individual tonal levels. This allows me to map the distribution of individual tonal levels across the whole image. For simplicity I have greyed out the tones representing the background.
From the perspective of the camera, we can now verify that each of these targets is subtly different in appearance. To the brain, which is combining two different perspectives on the same scene, this is an even more complex puzzle to deal with.
Mach Bands, Cornsweet, Checker Shadow and Chubb Illusions
I won't elaborate on these illusions here, but if you click on each of the links above they will take you to various web pages describing these individual optical illusions. While each illusion may be explained by subtly different mechanisms (some possibly yet to be fully proven) they all impact on the same general aspect of vision, namely our interpretation of ambient lighting. Are these phenomena all a direct consequence of the complex nature of light and perspective. Are they compounded by our binocular vision? Have some or all of these illusions arose for evolutionary benefit? Is there any real benefit to these illusions at all? I don't have these answers unfortunately. All I can say is, the more I read about all of this the more uneasy I have become as regards the accuracy and reliability of human vision when it comes to interpretation of subtle lighting and colour.
Take the example below. In the field we might often be presented with a scene where, while scoping a large flock of gulls or terns we are presented with a bird in the background which appears interesting. We decide to compare it's mantle tone with a bird in the foreground. They seem very similar. Are they the same or is one darker than the other?
In the field, this is not an easy distinction to make. However armed with a digital camera we can make a more helpful comparison.
So, as it turns out, the targets in the background are all a couple of shades darker than the ones in the front row. When objects we wish to compare are separated our eyes cannot make an accurate distinction (global analysis). When close together (local analysis) we can do a much better job.
Bins or Scope?
Here is an interesting question. If our eyes and visual system are adapted for binocular vision then presumably our vision, including our interpretation of perspective and lighting is much the same when we use a pair of binoculars. But what happens if we use a monocular or a scope instead? Is perspective and lighting more accurately gauged through a scope than through bins? Or, are our eyes simply not equipped to be able to take advantage of monocular vision when an opportunity presents itself?
And what of the other, 'lazy' eye? Most experienced birders tend to use their dominant eye for the scope and have somehow 'trained' their brain to ignore the image coming from the other open eye, which is usually staring at the ground or off into space. Those inexperienced in using a scope will usually close one eye, or place a hand over it, because they find it distracting and they are unable to give full attention to their scope eye. So, at some point our incredible brain has learnt to decouple the images from our two eyes, ignore one and focus our attention almost exclusively on the signal coming from the other eye. This is no doubt amazing, but how can we be sure that the other eye is still not influencing our vision and judgement, including our global and local assessment of subtle tones? Should a critical observer carry an eye patch with them just in case? ;) A camera might be more useful!
Dynamic Range and HDRI Typically high dynamic range comes into play on a bright, sunny day. A camera's dynamic range cannot cope with the full range of light intensity from details captured in deep shade to details in highlights under such lighting conditions. For those who are not very familiar with Dynamic Range and High Dynamic Range Imaging (HDRI) here is a really nice video that will quickly bring you up to speed.
If you open any field guide and look at the plates, the illustrator in almost every case has chosen to depict a bird as it might appear under ideal, neutral and relatively low intensity lighting. Most birders, starting out would find the challenge of bird identification made all the more difficult due to the ever-changing nature of normal ambient light. If ones first experience of birding were a day in the field on a sunny winter's day, it might well turn out to be ones last experience of birding, such is the added challenge posed by the often harsh, forbidding light conditions in winter!
The diagram above hopefully illustrates the potential use of HDRI for forensic image analysis. From the perspective of bird identification from digital images, we hope to use HDRI to bring the lighting in an image more in line with the ideal lighting we are familiar with from illustrated field guides.
HDRI Using Exposure Bracketing
Most HDR images are created by using exposure bracketing to make three or more exposures in quick succession with different exposure times, designed to capture three discrete exposure brackets within the dynamic range of a scene. When I photographed the subject of the next few images it was a wonderful day for birding, except that, due to the time of year, the sun was never very high in the sky and the light was very harsh at times. In other words, the ambient lighting exceeded the dynamic range of the camera. I grabbed three exposures of this Robin Erithacus rubecula sitting on a black-capped white pillar late late in the afternoon. These were created using exposure bracketing. Due to the camera's limited dynamic range, the light was too contrasting, and consequently none of these exposures worked out great. The bird was stood facing the sun, so from the point of view of the camera the bird was side lit. This image most closely matches the lower of the four images displayed above and is a good candidate for HDRI.
What would it be like to combine the three bracketed exposures, extract the good bits of each exposure and disregard the bits that are either over or underexposed? Well that is the whole basis of HDRI. The goal is to try and flatten out the contrast in the image, bring up the detail of the parts of the bird that are currently in shade, and subdue the brightness and saturation of those bits that are in full sun.
Using Adobe Elements Photomerge tool I have attempted to create a HDR image from these three exposures. But, it has not worked out. Why? In the milliseconds it took to create the three exposures I moved the camera and the bird also moved. This is a common problem with HDR images and the main reason why many people don't bother trying to create and use them. Well there is an alternative solution.
HDRI from RAW
We know that RAW format images contain a lot of hidden detail. Why not create multiple exposures using one RAW image file? The three images to the left below were all created from the same RAW image. All of their image settings are identical with the exception of exposure. I adjusted exposure to roughly match the exposures I had made in the field. I have now combined them in Photomerge to create a HDR image from them.
Here is a comparison between the original JPEG and the HDR image made from RAW.
The main difference between these images is a reduction in contrast in the HDR image. A HDR image is the scene's dynamic range compressed to fit within the dynamic range of the camera / display device. In turn, this more closely matches, or rather mimics the dynamic range of the human visual system. While this hasn't quite flattened the image to the point that it might appear ideally lit, there is certainly a marked improvement. It is much easier to appreciate fine details and subtle colours throughout the tonal range of the subject from highlights to shadows.
Could this HDR image have been created from RAW following the normal RAW work flow from a single image, without the need for all of this multiple-exposure merging? The short answer is yes! Further down I have done just that. But there is a bit more digging to be done first, so please keep reading.
HDRI versus a Contrast Tool
As we have seen, the major change brought about in a HDR image when compared with an original JPEG is in the contrast of the image, what is the difference between HDRI and simple Contrast tools? Here are a couple of images comparing HDRI and the Contrast Tools in Adobe Elements and Camera Raw.
I made the above comparison image using the 256 tonal grid I had used earlier for the Adobe Lighting Tools posting (HERE). The HDR image (created from the three images on the right) and Contrast Tool set to maximum contrast reduction (-50) look quite similar but the histograms show there is a bit more going on in the HDRI image. As noted in the posting linked above, the exposure tool in Adobe Elements doesn't treat brightening and darkening of an image in quite the same way. Darkening an image produces a flatter histogram than an equivalently lightened image, for whatever reason. This possibly accounts for the jumbled HDRI histogram. So, perhaps exposures created using Elements do not make the best example.
The HDRI and Contrast images below have both been created from RAW so that is possibly a more valid comparison than one made in Elements. The HDR image and the image created by merely flattening contrast certainly look much more similar and their histograms support that conclusion. The HDR image just looks that little bit more compressed.
This image emphasises the clear link between HDRI and simple Contrast tools. A Contrast tool can be used to create a rudimentary HDRI image. For a nice example comparing a HDRI image with one obtained using a simple contrast tool skip to the end of the page linked HERE. Clearly, if done right, a HDRI image produces much better tonal detail and vibrant colours.
HDRI versus a RAW work flow
As the image above illustrates, most of the work done by Photomerge could have been replicated by simply using the contrast slider in Camera Raw. So where is the added benefit in creating multiple exposures and combining them using a bespoke HDRI software? That is certainly a legitimate question and one I have sought to address below.
HDRI Software
Firstly, lets take a look at perhaps the best regarded of the HDRI software packages, Photomatix Pro.
There are quite a number of HDRI software packages and plug ins on the market, most intended for artistic/aesthetic image-finishing purposes. These are powerful editing programs. Could they offer us any advantages as image forensic tools?
I have come across another very good video using the same multiple exposure from RAW technique that I used above. While this video is much more about the aesthetic/artistic value of HDRI, hopefully it provides another interesting insight into HDRI and the powerful image editing tools provided by some of these programs. Clearly, there is more to HDRI than a simple merging of exposures.
Having seen what Photomatix Pro can do I first downloaded the trial version of Photomatix Essentials. Below I have compared the results using Photomatix Essentials with the results I previously had obtained using the simple Photomerge tool in Adobe Elements.
The upper two images compare the results obtained with the three bracketed exposures. In theory HDRI images from bracketed exposures should offer the best HDRI results because each exposure bracket is a proper standalone image, with minimal noise and maximum tonal range preserved. But the camera wasn't held perfectly steady and the Robin moved. This is the big problem with HDRI from exposure bracketing. The resulting 'ghosting' renders the Photomerge tool pretty useless for HDRI using bracketed exposures. Photomatix has a tool to remove ghosting but it does not work perfectly, and, not surprisingly introduces some artefacts of it's own. In this case there is a big improvement but there is still slight ghosting of the bill, and the rear crown is also affected.
Comparing the lower two images, the Photomerge Tool does a good job with the three copies from the same RAW image. Photomatix Essentials produces a slightly better, more natural result, with more detail but I am not convinced that Photomatix Essentials would so a better job than a normal RAW work flow (see below). Photomatix Essentials costs around €/$40.
Verdict:- possibly not worth the money unless working from exposure bracketed images.
Camera RAW V Camera RAW + Photomatix Pro
Photomatix Essentials' more expensive cousin Photomatix Pro obviously has a lot more useful functionality but at around €/$100 has the price tag to match. The question is, does this added functionality push it far enough beyond a normal RAW image work flow capability to warrant adding it to the forensics tool bag? Full credit to the manufacturers of Photomatix Pro - one can freely download and play around with the full package on trial and this is what I have done here.
For this comparison test I have created the lowest contrast, yet reasonable result I could manage in Camera RAW. The process of creating a single HDR-like image from a single image is calledTone Mapping so this is what I have done in Camera Raw. I have then taken that image and used the additional Tonal Mapping functionality in Photomatix Pro to try and bring out more detail and tonal quality from the resulting image.
Verdict:- Once again it is hard to justify paying extra money for arguably no real improvement in image quality. As with Photomatix Essentials the real benefit of having Photomatix Pro would only come about where in-camera bracketed exposures were obtained in the field.
In Summary
The high contrast light that characterises bright sunny days challenges and often defeats the dynamic range of digital cameras. The solution is High Dynamic Range Imaging (HDRI). However there are some practical difficulties in obtaining good bracketed exposures. One might be better off taking the time to shift position and get a better angle on the subject relative to the sun rather than trying to create steady bracketed exposures! As for HDRI software? There are certainly some high performance software packages out there but they don't appear to offer anything above the standard Camera Raw workflow, unless of course you have obtained good bracketed exposures, in which case Photomatix should easily outperform the Camera Raw work flow which is based on a single exposure.
Last video...a Tone Mapping example to show the HDRI capabilities of Adobe Lightroom from RAW. It might be time I upgraded from Elements!
%2BMike%2BO'Keeffe.jpg)
.gif)
