Wednesday, April 16, 2008

With much trepidation...

...I hitch up my boots and wade into the subject of Mike Collin's latest video (scroll down to the bottom of his Winter '08 field journal). I'm not sure why I do this...

First, the ability of us outside reviewers to interpret this video is severely hampered by the fact that we haven't seen the whole thing. I have found the individual frames and one brief sequence very confusing, removed from their context as they are. Viewing angles and geometries, when we are seeing bird and when we are seeing reflection of bird, this is all quite difficult to sort out without seeing the entire sequence intact. It also makes it hard to judge whether the single frames show typical, recurring features or are single-frame aberrations. Plus some of the stills have been "enhanced," which may or may not show real information captured in the originals, and may or may not introduce additional artifacts that can mislead. Bicubic interpolation, used on the latest sequence, adds no new information at all, but gives a false impression of higher resolution by smoothing out the pixel edges. This is akin to doubling your sample size in a statistical survey by just mathematically making up new data that matches the pattern demonstrated by the original data set. These are the same issues that have plagued interpretation of the Luneau video. So, until the whole, unprocessed sequence is posted it's hard to be very definite here. I should mention as an aside that de-interlacing is not an enhancement of the sort that fabricates new information (if it is done properly). It in fact is the opposite: it undoes a processing "enhancement" and separates data out to a form closer to the original raw captures.


Mike has just added stills and a video clip of a second bird, which appears in the stills at least rather clearly to be a duck. In the blurry video it looks quite a lot like his "Third Ivorybill Video" bird. Which leads me full circle back to concluding that both birds are, in fact, ducks. It occurs to me that in the "Ivorybill" video, what appear to be wings tucked tightly against the bird during the upstroke might really be wings rendered nearly invisible by motion blur and NOT in fact tucked tightly against the body but about half extended in typical duck fashion.


Okay, to the video:

The flight indeed does not look correct for a duck. The principal issue here is, as Mike pointed out, that on the upstroke the wings are folded tight in to the body. Scanning through stills and videos of ducks of many species in flight, you do not see this taking place. It is typical of the flight of woodpeckers, both definite Pileateds and whatever the Luneau bird is. But the question is, what other birds also show this type of flight? Parrots, doves, and pigeons do. Not saying this video shows any of those, just listing the three I know of off the top of my head. How many others? I don't know. I couldn't even guarantee you that no ducks do that, either. This survey of flight styles of large North American birds would take a long time, but would otherwise not be difficult in the Internet era. The size, rapid wingbeat, and high ground speed (35 m.p.h. is actually quite fast for powered bird flight, isn't it?) would help limit the search significantly.

The three frames presented that Mike believes show white traling edges to the wings are troubling. First, given the issues I mentioned at the start, I'm still not entirely clear on the viewing geometry here. I guess we are looking down on the topside of the bird, not at its reflection? Regardless, my first thought is, "Sheesh, here we go again..." Mike believes these images, especially the middle one, show a white trailing edge and a black leading edge. ONCE AGAIN, we're dealing with artifacts here. Look at the bright branch to the left of the bird. Notice its prominent "black leading edge." Notice how much it resembles the "black leading edge" on the left wing of the bird in the middle image. Look at the smaller bright branch to the right of the bird, and notice its "black leading edge" also. Additionally, notice that the "black leading edge" on the left wing gets especially prominent in the middle image, when that wing is passing in front of a darker area in the background. It's Luneau all over again. These are EDGE ARTIFACTS of the sort always created by jpeg (and jpeg-like) IMAGE COMPRESSION. Whether or not there is a real black leading edge there cannot be resolved through the artifacts.

Beyond the artifacts, I note that the bird's back does not appear especially dark. There's no clear break between the two white wing areas, just a fuzzy sort of darker area. This indicates that either the bird's back is not dark or the resolution of the video is too poor to tell. Either way, we're left considering all birds that have a lot of white in their wings (just about anywhere in their wings) and a back that might range from medium gray to black to boldly patterned, or just about anything else. Many ducks fit that description, which makes the issues of flight style all the more critical. So do grebes, Anhingas, quite a few shorebirds, and probably others that don't immediately come to mind. Could any of these match the flight style? Couldn't say, personally. But at this point making the jump from Wood Duck to Ivorybill seems extreme.


Post a Comment

<< Home

Site Meter