Friday, 17 March 2017

Colour - Saturation as an Image Artefact

In a summary HERE I described an artefact as the distortion of an image.  There are various environmental, optical and processing parameters which bend and distort the properties of an image in different ways.

I was recently given an interesting puzzle by a member of a rarities committee.  Photographs of a rare bird appeared to show colour-tinged fringes to a feather tract where one might not expect to see them.  By contrast, another feather tract showed pure white fringes as expected.  After some deliberation I concluded that the effect must be due to image saturation boost.  It was only through saturation boost that the colour fringes had become apparent.  In some of the captured images the saturation was boosted during processing, enhancing the colours.  In others the images were more natural looking, and the fringes were so subtly coloured as to be virtually unnoticeable.  The fringes did indeed have  subtle colour in them, but interestingly this had not been recorded by the observers in the field.

The immediate question becomes, well are these fringes really coloured or not, if after all they were not readily observable in life?  An interesting perspective I think!  I imagine that if this bird had been observed in the hand the fringes would have been rather more apparent, albeit still quite subtle.  In the images where the colour saturation was boosted, so that they became prominent, one would have to describe their unnatural saturation boost as due to an image artefact. In other words, they were a distortion.  Saturation boost has it's pros and cons as I described in my posting HERE.  One of the advantages is in being able to differentiate and name colours which may have become subdued due to lighting or environmental factors.  But one has to be able to sort through the noise and understand exactly what is going on in a saturation-boosted image to interpret it correctly.  

Saturation Boost is an Artefact while Desaturation may not always be so
What do I mean by this?  There are many image quality parameters which can leach or desaturate the colour from an image.  Some of these can be described as artefacts, but not all.  Moisture and dust for example can distort an image as it passes through the air by lowering contrast and saturation.  These could be described as environmental artefacts. Glare from sunlight hitting the lens is another example. On the other hand, I wouldn't consider low-intensity light by itself as an artefact, because the image is formed from this light not distorted by it, if you get the distinction.  But some of the image quality reduction caused directly by the camera's reduced ability to deal with insufficient light intensity could be termed artefacts (eg. noise, motion blur etc).  There are also of course many natural forces which bleach colours naturally.  The images below help to illustrate these points.

I believe these two images refer to the same individual tristis-type Chiffchaff.  And yet, the colours appear quite different.  Why?  The image on the left was taken in fairly low, diffuse mid-day light in mid-January.  The right hand image was taken about two weeks later in much lower light, approaching dusk.  I observed the same bird again once more, in better light a week or so after that, which allowed me to conclude a couple of things.  The bird's yellowish primary fringes had actually faded a bit since I had first seen it, so this discrepancy between these images had at least a partially natural explanation.  The large white balance correction required for late evening light may also have been a factor affecting colours in the right hand image.  Lastly, I believe the level of ISO amplification needed to form an image from such low intensity light meant that there was surely a loss of colour fidelity, including an impact on colour saturation.  So, this is an example of both natural and unnatural, artefact-driven colour desaturation at work.

Perhaps the most interesting revelation from these recent experiences with colour is that forensic tools like saturation boost may actually help uncover some hidden colour traits that we may not currently be aware of, simply because our eyes are not tuned to them.  I wonder if the subtly coloured fringes of that rare bird are actually a common but little observed trait which may become more apparent through the forensic analysis of photographs.  Indeed, perhaps there are similar 'cryptic field marks' waiting to be discovered.

Friday, 13 January 2017

Colour - A Bird In The Hand

I was recently asked if it there might be software available that would restore perfect white balance to jpegs of birds captured in the hand.  I am not aware of any software guaranteed to do this accurately every time without the aid of a white balance card when the images are first captured.  If there were, cameras would surely already be using this in place of current auto white balance algorithms, which, let's face it, can be rather hit-and-miss.  This got me thinking about typical in-the-hand photos and wondering, why ringers (banders) and photographers don't take a moment to properly cater for exposure and white balance when capturing these images.  After all, a lot of work has gone in to catching the bird.  Surely having a mini in-field studio setup nearby is not a major impediment, assuming of course the bird is not stressed and there is a moment or two to take some record shots.  In the time it takes the ringer to process the bird a photographer could quickly rig up a decent neutral mid-grey backdrop in an area with suitable lighting. But this is rarely the case.

Yellow-browed Warbler Phylloscopus inornatus - a real treat to observe and photograph in the hand. Unfortunately, images like this are often less than perfect owing to a contrasting backdrop and unavoidable lighting conditions.

I gave this problem some thought and concluded that there must be a tool and a simple process ideally suited for ringers/photographers in the field which would help improve the quality of images captured and help get the most of these images in terms of exposure and white balance control.  After a bit of internet searching I came across a product by Manfrotto called Lastolite Ezybalance.  I decided to purchase one to test out it's handling and, in particular, to compare it's neural white balance with the professional, and far more expensive, Colorchecker Passport.

The Lastolite comes in three sizes ranging from 30cm up to 75cm.  The 30cm one costs about €25 and should be more than adequate for photographing passerines at least.  The unique selling point for our purposes is that the product folds down to one third of it's working size to allow easy transport.  Unboxed, the ezybalance comes in a nylon pouch to keep it from unfolding, plus a useful user guide (HERE).  As soon as the ezybalance is removed from its pouch it automatically expands to full size.  It seems really durable, with a nice matt finish to minimise reflection.

One side of a the ezybalance is an 18% reflectance or mid-grey card with a white and black surround.  This is intended for exposure capture but may also be used for white balance correction.  The other side is a paler approx. 60% reflectance, neutral grey card more specifically intended for white balance correction than exposure metering.

Compared simply with the white-balance card from my X-rite Colorchecker passport I detected a very slight difference so I suspect the ezybalance which I purchased is not 100% neutral.  In fact I am pretty sure I can see a slightly warm yellow cast in it with the naked eye.  But for €20 its pretty good.  The 18% reflectance card is not too bad either and, given its usefulness for exposure metering I recommend using this side of the card for both exposure and white-balance purposes.  Okay so strictly speaking this method may not give 100% perfect white balance, but coming from a low base where few people introduce any white balance tool to their photographs, this is a good starting point.

Recommendations for using this card in the field would be to clip the card onto a pole mounted using a couple of clamps or clothes pegs, oriented vertically.  The card needs to be located in the shade if conditions are sunny.  It can be located out in the open if conditions are overcast.  In order to avoid lens and perspective distortion, and to minimise stress on the bird I recommend using a long lens 200mm plus and hanging back a bit from the subject.  Ideally place the camera on a tripod with everything already framed.  Before the bird is introduced, having already decided on an appropriate shutter speed and aperture check the exposure based on the 18% grey card, filling the frame with the card as recommended.  Based on the available light you might want to play around a bit more with ISO, aperture and/or shutter speed.  Just to be on the safe side I would recommend some exposure bracketing and of course shooting in RAW.  Now with the studio all set up it's just a matter of waiting for the model to arrive.  As the camera and settings are already set, it may take as little as a minute or two to obtain a perfect photographic record of the subject which can be reasonably corrected to a high standard for exposure and white balance.

On the left we have the original image warts and all.  On the right, a mock up of how the same image might have looked with a neutral backdrop such as the ezybalance card, positioned in a more controlled location in terms of diffuse lighting.  I hope this short note prompts some ringers and photographers to consider improvements in the capture of in the hand images come this year's ringing season.