Strange streak discussion: 2004 Dec 7 APOD

[hawaiianmike]

I am not convinced that the flash is indeed a reflection of the camera flash off of the body of a bug---and right on top of the "pole, mast, exhaust pipe, whatever?" I don't think so--if a bug flew in front of the lens, very close up, it is just too much of a stretch that the trail ends right where it does---see the white smoke visible in the frame that has the pole flash event? It would seem that whatever made the flash also made the white smoke/mist/whatever--it is not in the "after" photo so it must have either been windy and therefore blown out of the frame, or it burned out quickly, like it would be if the smoke was part of the flash event--I had been saying that the trail would have been visible below and to the right if a bug had flown by at just the time the snap took place, but I accept that if the exposure was 1/20 of a second, it is possible that the exposure ended before the poor bug had a chance to get down to the right, lower corner of our picture---unlikely, but possible-

I STILL thgink that there was some sort of flash event in line with the pole, and it co-incided with some unrelated event in the background, perhaps even as far away as the landmass in the background, which caused the trail---2 separate events--an inspection of the light on the pole would be useful---or exhaust pipe, or mast, etc. And it seems out of sequence with the other poles shown, so it may be something off of a boat hidden from view---we don't really know if that flash was from the top of a lightpole or not---or whatever??--I can only hope that if a bug was involved, that it died that night, hopefully painfully--

[Guest]

Any test of a hypothesis needs to allow the possibility of disproving the hypothesis.

Any hypothesis that cannot be disproven is useless.

You are correct. I should have phrased the last sentence as "evidence for a periodicity of 200 Hz in the 'dark streak' would tend to disprove most rival hypothesis not involving flying insects, as those hypothesis would lead to predictions of no periodicity."

That would cover meteors, lightning, sub-atomic particles, dark energy, space time warps, CPBs, exploding light bulbs, toy rocket launches, discarded cigarettes, hot pixels, hairs on the lens, CCD degradation, reflected sunlight, planar contrail shadows and Elvis.

You could argue that it would not count against Moire patterns or CCD defects/artifacts, as you could make a case that some sort of regularity may be imposed by the very mechanism that generates these things.

[Harry Law (A)]

Big congrats to victorengel for the horizontal compression technique. I tried that on my own diff image, only I ran the horizontal compression to 10:1 -- there is NO WAY that line is straight.

In an earlier post, I commented that only the exceeding straight line made me nervous about the bug conclusion. But the line is not straight.

It's a bug.

I have made up a composite image of my diff, with an inset blowup of the bug outline, and another inset of the 10:1 compression on the streak. It looks like a nice presentation of some well placed image analysis.

The bug wins.

Even if the line is not completely straight. How do you explain it being there at all? The shuttle speed is so fast.. I can't imagine a bug can fly that quickly.

How about.. let's think simple? What if it is simply a drop of water? Have anyone mention the possiblity of that?

I'm not very technical on digital imaging or photograph. I do need someone's experties to common on the "Drop of water" idea. Will anyone care to enlighten me?

(The A means I'm waiting for my registration ID sent to me via e-mail.. :/ )

[Guest]

How about.. let's think simple? What if it is simply a drop of water? Have anyone mention the possiblity of that?

I'm not very technical on digital imaging or photograph. I do need someone's experties to common on the "Drop of water" idea. Will anyone care to enlighten me?

Yup, water droplets have been raised several times. The "water droplet " hypothesis would lead to several testable predictions.

One of them is that the path of the "droplet" is indicative of the direction and velocity of the wind in the vicinity of the camera. The large slant would indicate a strong wind from left to right (if I make the reasonable assumption that the drop is falling downwards). The photographer has stated that the conditions at the time of exposure were nearly calm. We know the location and time so we could check the weather conditions, both for rain and wind.

Other predictions involve the length of the path which must be constrained by the shutter speed, the maximum speed of a falling drop of water and distance from the lens. Distance is constrained by the width of the streak and size of the blob.

Still another is the shape of the caustic in the bright blob. If it is a flattened blob of water the shape must be explainable as a reflection/refraction of the flash.

[Phil]

The animated gif below shows three frames: 1st shot, 2nd shot, and the diff of the two, for 1 sec. each. It's very likely that whatever occurred was caused by the lamp turning on.



The three crops in jpeg format are:

http://members.cox.net/dr650/1.jpg
http://members.cox.net/dr650/2.jpg
http://members.cox.net/dr650/3.jpg

Exactly! When the lamp turned on the bug shot for it like all bugs do to lights in the night.

[victorengel]

Perhaps of interest is a picture I took of the Congress Avenue bats in Austin. There are some parallels to this picture and the insect theory. The picture was taken near sunset. The picture is of a flying object occluding a light sky and also illuminated by flash. The exposure is long enough to show motion blur. The flash illuminated version of the subject appears white. But note the bats have black wings whereas an insect would probably have transparent wings.


In this case, the camera was a Canon 10D with first curtain sync enabled. Shutter speed was 1/50 second. Aperture was f/2.3. The sensor on the 10D is much larger than on a G3, and the G3 picture was shot stopped down more, so there is more depth of field in the G3 picture. I'll post another picture shortly showing droplets in the air, since that's been in the conversation recently, too. Also notice how certain things, especially eyes and glasses show up very brightly even though hidden from the sun.

Also notice the difference in appearance between the bats in front of sky vs. the bats in front of the bridge. This clearly shows how the flash stops the action of the bat while the long exposure simultaneously captures blurred bats when they're sillhouetted in front of the sky.

[victorengel]

Here's the promised picture with the droplets. They look like out of focus pentagons because the iris in the lens I was using has 5 sections. The closer the droplet is to the camera, the bigger the blob and the brighter it is because it gets more illumination from the flash.


I don't know the source of the droplets. I could feel them on my arms, and they were quite small. One thought I had was that they were urine from the bats. Another was that they were honeydew from small insects flying in the air. Some of the blobs might actually be insects. I just don't know.

[Guest]

Even if the line is not completely straight. How do you explain it being there at all? The shuttle speed is so fast.. I can't imagine a bug can fly that quickly.

There have been calculations made on previous pages showing the possible distance speed etc of the bug.

skyglow1

[Guest]

I tried to look at the picture from the bug theory, and can't see any obscuration of the image by the bug body at the time of the flash.

I also can't see any distinct "centerline" by the bug body on the line (trail).

If the bug's body was the bright spot exactly, why it was not like a bug one? And what is the large halo around it? Super-large wings?

If it was just a bug (and bugs are common), why the photographer was so puzzled with the picture? Just coincidence?

Why the bright spot coincided with the lamp discovered to be non-operational later? Just coincidence again?


Aren't too many coincidences and "if" for the bug theory?

I think that unless additional data (try to find the people who were in the area) all theories are pure speculative (but some are unlikely already now).

[victorengel]

I tried to look at the picture from the bug theory, and can't see any obscuration of the image by the bug body at the time of the flash.

Please note that before the bug got to that position (assuming 2nd curtain sync) the camera's sensor was being exposed by that area of the subject behind the bug during the rest of the exposure when the bug wasn't there. So the "obscuration" would only amount to the percentage of the bug's body length to the entire trail length, which I think I calculated near 3%.

I also can't see any distinct "centerline" by the bug body on the line (trail).

The trail is the centerline. More specifically, the head, thorax, and abdomen of the insect (assuming it was one with these three distinct sections) obscure the background for an approximately consistent width. One would expect the lightness of the trail to diverge from even lighting only by how much the profile of the body is not a consistent length across the width of the insect. Put another way, measure the length of the body at one side of the insect. Measure again at the center of the body. The difference between these values is the amount of variation we should expect in the trail. For most insects, I think this would be negligible.

If the bug's body was the bright spot exactly, why it was not like a bug one? And what is the large halo around it? Super-large wings?

I'm guessing the bright, yellowish spot is the insects abdomen, possibly covered with pollen, as in the bee picture posted earlier. The whitish blob below and to the right would then be the insects thorax and head, which is black, in order to be colored the way it is in the picture (see my bat pictures). Alternatively, if the insect is flying in the other direction, it's the head that's bright and the rest of the body that's black.

If it was just a bug (and bugs are common), why the photographer was so puzzled with the picture? Just coincidence?

That is not a coincidence.

Why the bright spot coincided with the lamp discovered to be non-operational later? Just coincidence again?

It doesn't coincide. There is no coincidence.

[Ramunas]

'can't see any obscuration of the image by the bug body at the time of the flash.'

Object in motion attenuates the background image or, at the moment of flash, adds to it. Either way, the effect is 'translucent'.

I thoroughly enjoy this collective surge of mind power that will, ultimately, result in a burst of light complete with a dark streak.

Goodnight.

[zoltan]

I have two questions for the bug theorists:

1. If the flash comes from the upper left of the camera, then why does the bug seem to be illuminated from the lower right? The highlights trace an outline of the lower right side of the bug, not the upper left as would be suspected.

2. Why has nobody answered how the same trail appears somewhat wider and much fainter in the actual before picture? (not the picture that is linked "before") Look at this animation that Diffit posted and you can see a very faint trail quite clearly.

The animated gif below shows three frames: 1st shot, 2nd shot, and the diff of the two, for 1 sec. each. It's very likely that whatever occurred was caused by the lamp turning on.



The three crops in jpeg format are:

http://members.cox.net/dr650/1.jpg
http://members.cox.net/dr650/2.jpg
http://members.cox.net/dr650/3.jpg

Diffit: Could you do the same processing on the third image (actual after) that you did on the first two in the animation? I'm curious to see if the trail is there as well.....maybe this would add support to the camera artifact theory....the flash of the lamp simply brought it out more.

-z

[wombat]

Someone posted a question about looking along the "shadow" (for want of a better word) to see if there was going to be any periodicity to the darkness caused by the beating of wings. Now given that there seems to be general agreement that the shadow is that of the body and not the wings (as was highlighted by those magnificent bat images) this would seem to be an exercise in futility, but never mind, I put my hand up for the job.

What I did was as follows.

1. In Paintbrush (all I've got that will do this), skewed the picture 34 degrees to make the shadow as horizontal as possible.
2. Drew a dark line above and below the shadow to delineate where I wasgoing to measure from.
3. Cropped out all except this long thin rectangle. It ended up, conveniently, 32 pixels high.
4. Converted the picture to 256 colour greyscale. This is because the programs that I wrote to analyse image files have this limitation; I'm far too lazy to do full 24 bit colour. Now, this may well have reduced the information to the point where it's meaningless, but bear with me.
5. Ran my histogram program on every one of the 32 lines.
6. Combined these histogram outputs into a spreadsheet.
7. Finally, in the spreadsheet, I then averaged "across" the line and then graphed it.

There is a periodicity of 3 pixels in the resultant graph. It is quite clear and quite sharp, a sort of a "sawtooth". Given that there are about 1000 pixels along the path, that gives say 330 full waveforms, which in 1/20 second translates to a wingbeat frequency of 6600 Hz. Now, is that the sound of raucous laughter from the entomologists (spelling?) who have been following this saga?

Whether it means anything, of course, is highly debatable. I only hope that my posting here inspires someone with better tools that my homewritten ones to do an analysis on the original coloured 24-bit image.

Well, I'm going away for two weeks from tonite and will have no Net access during that time. I anticipate that this issue will be fully resolved by that time, so I look forward to reading the no doubt 600 pages that will be here when i return. It's been fun posting, and even more fun reading. Cheers to you all.

[Guest]

[quote="victorengel

I'm guessing the bright, yellowish spot is the insects abdomen, possibly covered with pollen, as in the bee picture posted earlier. The whitish blob below and to the right would then be the insects thorax and head, which is black, in order to be colored the way it is in the picture (see my bat pictures). Alternatively, if the insect is flying in the other direction, it's the head that's bright and the rest of the body that's black.

The bright spot does not have any blurring due to motion of the hypothetical bug during duration of the flash. But the bug had to move almost across the whole picture for 0.05s. So my estimate of "possible blurring/total path" ratio is <~ 0.001 . So in the bug theory this imposes an upper limit on the duration of the flash as ~0.05s/1000 . Could somebody find out duration of the flash (technical info)?

[Huff]

It appears to me, that lightning struck the high tension lines down line somewhere and this poor pole was the nearest ground.

1) it appears that the arc spans the ground near the pole and extends to the high tension wires.

2)the pear is lit up rather brightly , noticable just to the left of the high tension pole and to the right of the shorter pole.

3)shadow is from the lamp body

[Guest]

When looking for an explanation, it is not enough to make a guess and then start supporting your guess.
That is how people succeed at fooling themselves.

wise words

[caseyoconnell]

Entertainment value is important.

This whole episode is worthwhile for that reason alone.

The fact that there has been some very intelligent and otherwise talented posting on this topic has made me glad I found out about this place.

The bats photos are exceptional. Being a former Austinite, I particularly appreciate the Congress Avenue bats.

There are indeed some details remaining before we can write this off as a bug.

When we can identify the genus and species, and perhaps his last name and address, then we will have truly achieved something.

[hazeii3]

Note: In all the following, the original image has been rotated so that the trail is approximately horizontal. Larger versions are available as .tif images - either click on the lower-resolution images below, or visit the index page for these images.

The first image is an enhancement of the object and trail, creating by taking the difference image, converting to greyscale, and performing a 'stretch' on the trail (note that the object at the end of the trail has not been enhanced).



This is another version of the enhanced image shown above, with a 10:1 compressed version at the left.



In theory, if the trail is caused by an insect, the wings should have slightly darkened the area to either side of the obvious 'body' trail. To test this, I summed the rows of pixels in a direction parallel to the trail, then normalised the resultant value and replicated it along the pixel row. The expected effect would be to reveal the slight dimming caused by the wings - however, this is not readily apparent.



Note that the centre of the trail apepars to be somewhat offset from the object itself; this is because the object appears to have started to curve downards towards the end of the trail (as can be seen in the horizontally compressed image above), while the lowpass filter reveals the average centre of the trail.

All images are available as high-resolution uncompressed .tif's - click on the low resolution images above, or visit the index page for these images. Any further images will also be posted to that page, along with feedback received.

[Douglas]

There is a periodicity of 3 pixels in the resultant graph. It is quite clear and quite sharp, a sort of a "sawtooth". Given that there are about 1000 pixels along the path, that gives say 330 full waveforms, which in 1/20 second translates to a wingbeat frequency of 6600 Hz. Now, is that the sound of raucous laughter from the entomologists (spelling?) who have been following this saga?

Whether it means anything, of course, is highly debatable. I only hope that my posting here inspires someone with better tools that my homewritten ones to do an analysis on the original coloured 24-bit image.

Great idea, well done, _but_: A periodicity of 3 pxl sounds like the effect of the JPEG-compression (Chromacity(?) is only encoded for 2-pixel-blocks iirc, and as the trail is diagonal, that would make about 3 pixel).

--
Douglas <douglas at isja.org>

[fpj]

Note:
All images are available as high-resolution uncompressed .tif's - click on the low resolution images above, or visit the index page for these images. Any further images will also be posted to that page, along with feedback received.

Excellent analysis! At least some of us are trying to follow some sort of scientific approach.

Your idea to try to find a pattern for the wings is very interesting, but the path does not seam to be exactly horizontal in your images. I tried to make it more horizontal by rotating the image for about 0.5 degrees counter clockwise (from this image it obvious that the path is not exactly a straight line). I wonder if your last analysis could give different results with this additional rotation.

[hazeii3]

....
Your idea to try to find a pattern for the wings is very interesting, but the path does not seam to be exactly horizontal in your images. I tried to make it more horizontal by rotating the image for about 0.5 degrees counter clockwise (from this image it obvious that the path is not exactly a straight line). I wonder if your last analysis could give different results with this additional rotation.

I tried various rotations between 33 and 34 degrees, but they didn't make much difference - the main effect was to cause the centreline to become wider. The rotation is approximately correct for the first (left) half of the trail, after that the trail seems to curve downwards somewhat.

However, I decided as the remaining curvature is about 3 pixels vertically, and the object 'wingspan' is considerably bigger than that, it wasn't worth going to the next step (which would be to find the best fit to the trail, unwarp the image to make the trail truely straight, and repeat the low pass filtering).

Thanks for the feedback; I'll add this as notes on my page.

[hazeii3]

There is a periodicity of 3 pixels in the resultant graph....

Great idea, well done, _but_: A periodicity of 3 pxl sounds like the effect of the JPEG-compression (Chromacity(?) is only encoded for 2-pixel-blocks iirc, and as the trail is diagonal, that would make about 3 pixel).
--
Douglas <douglas at isja.org>

Agreed; I considered doing a periodicity analysis looking for the wings, but I doubt it's possible. Apart from JPEG artefacts (which will be above the frequency range of interest), there's the slow CCD response.

One possible approach would be to perform autocorrelation along the path, but we know the exposure was 1/20th of a second and if it's an insect its wing rate will be 50hz to 200hz, say. So there's between 2.5 and 10 periods along the track, which probably wouldn't give enough processing gain to pull the signal out of the noise (I'm assuming it's not a tiny insect with a >200Hz wing beat because it's far enough from the lens to show at least some shape).

[caseyoconnell.]

How about checking for the periodicity on a smaller portion of the flight path, maybe a portion corresponding to one or two wing beats.

[tonycc]

Here is a cropped image of a photo I took a few weeks ago and inadvertantly used the flash. It shows the flash bugged phenomenom.


Details:
- It was lightly snowing at the time, with winds coming from the left at an estimated 15-20 miles per hour.
- The image shows that my flash went off first, then the duration of the image captured the path of the flake. (The duration was about 1/400 sec, thus the trails would have been almost 20 times longer had I used the 1/20 as in the Australian photo.)
- In this case, since snow is white, the paths are lighter than the background.

Here are the camera details:
Nikon CoolPix 4500
FL: 7.8mm
Exp Mode: Prog Auto
Metering Mode: Multi Pattern
1/317.8 sec - f7.5
Exp Comp: OEV
Sensitivity: ISO 100
AF Mode: AF-S
Flash Sync Mode: Front Curtain

I realize this is late in the discussion. I really have enjoyed the logical give and take, expecially the recent use of data and the scietific method.

Thanks,
Tony

[victorengel]

1. If the flash comes from the upper left of the camera, then why does the bug seem to be illuminated from the lower right? The highlights trace an outline of the lower right side of the bug, not the upper left as would be suspected.

I don't think there's enough detail to make this determination. How is it that you can determine the direction of the light source? In reality, given the resolution we have here, the light source is essentially from the same direction as the lens. If you see a difference in the wing, I submit that the leading edge of the wing is thicker than the trailing edge, so one edge likely reflects more light than the other. But maybe your're seeing something else?

2. Why has nobody answered how the same trail appears somewhat wider and much fainter in the actual before picture? (not the picture that is linked "before") Look at this animation that Diffit posted and you can see a very faint trail quite clearly.

I have not seen this demonstrated except in this animation picture. It could simply be an artifact of the gif animation process or simply your eyes accommodating the animation. Let's see you demonstrate the trail in a still picture. I don't see it.