Photo of Spoke Deflection?

"Luns Tee" wrote: (clip) Werehatrack seemed to be
describing the bottom of the wheel appearing flatter [than the top]. My
point is that the shape seen for the top of the wheel will be no different
from what's seen of the bottom of the wheel.
^^^^^^^^^^^^^^^^
You are right on both points. Knowing what to expect, I read that into his
wording.

On Mon, 11 Jul 2005 03:02:09 GMT, "Leo Lichtman"
wrote:


"Luns Tee" wrote: (clip) As these rings pass by the camera, there's no
difference between what happens at the top and bottom of the ring. (clip)
^^^^^^^^^^^^^^^^^
Your idea of masking the spinning wheel to eliminate rotational effects is
clever, but it is not on point. The "ovalization" is not caused by the
rotation of the wheel, but by its translation, so the ring masks would also
appear as ovals.

Lets do this thought experiment: Take a picture of a vertical line moving
past the camera. If you need a physical example, paint a vertical white
stripe on the side of a semi-trailer. Set up your camera, with a
focal-plane shutter, on the sidewalk, and make the exposure as the truck
passes. If it takes 1/50 second for the slit to move from the top to the
bottom of the film, and the truck is moving 60 MPH (88 ft/sec) the line will
advance 88/50 = 1.76 ft during the exposure, producing a distinctly sloping
line on the negative.

If, instead of a vertical line you have a bicycle wheel, it will be as
though you painted the wheel on the side of a tall stack of cards, and then
slid a tilt into the stack.

The reason the wheels are not oval in the OP's photo is that the film is
moving instead of the slit, so that the image and film are moving together.
So the film records the wheels in their true shape. Since the spokes are
not synchronized with the film, they are distorted.

There's no film. The image is recorded digitally, and it's done in a
series of vertical bars, all taken of precisely the same plane across
the path of the cyclists.

On further examination of the method, the wheels will come out round
if the speed of the object through the plane of the image capture is
just right, or if there's a correction done digitally after the fact.
The image supplied to the media was very low-res; given that the
margin of victory was reported as two millimeters, the resolution of
the original image had to be very high indeed. Since this makes it
explicit that the original image was at a different resolution, some
image manipulation to reduce the size of the file was definitely done;
correcting the apparent aspect ratio to de-ovalize the wheels would be
a simple matter at the same time. Given some of the artifacts I
observed in the media image, I'm pretty sure that an aspect ratio
correction was applied.

--
Typoes are a feature, not a bug.
Some gardening required to reply via email.
Words processed in a facility that contains nuts.

On Mon, 11 Jul 2005 04:20:12 GMT, "Leo Lichtman"
wrote:


"Luns Tee" wrote: (clip) Werehatrack seemed to be
describing the bottom of the wheel appearing flatter [than the top]. My
point is that the shape seen for the top of the wheel will be no different
from what's seen of the bottom of the wheel.
^^^^^^^^^^^^^^^^
You are right on both points. Knowing what to expect, I read that into his
wording.

And I was incorrect with that assertion. However, image manipulation
would almost certain;ly still be needed, given the image capture
method used, to keep the wheels from looking elongated or compressed.
Image manipulation was certainly present. The original image
assembled would certainly have been of a far higher resolution, and
most likely of a less lifelike raw form, than what was handed to the
media.
--
Typoes are a feature, not a bug.
Some gardening required to reply via email.
Words processed in a facility that contains nuts.

In article ,
Moe Kit wrote:

If you look carefully at this photo

http://www.velonews.com/images/details/8436.11841.f.jpg

the spokes appear to be bending in some sort of complex curve, like a
higher-order polynomial or a catenary.

At first I thought it was a shutter/moving object effect with the camera.
But if you look at the rear wheel of the foreground rider the top spokes
appear linear while the bottom spokes are 'bent.' If it was a camera/speed
artifact then all spokes should appear bent because wheel speed is constant
around the wheel.

Is this spoke bending real?

If it's real, then why don't the spokes bend under load while the bike is
stopped?

If it's real, and spokes bend and unbend to this degree with each
revolution, why aren't spokes failing under fatigue loading more
frequently?

That picture is not a photograph; it is a composite.
The line in the image is not the finish line; it is a movable line
drawn by a computer graphics system.
The picture is not an image of a physical entity.

All camera-lens systems have a field of view.
The camera at the finish line has a particular field of view:
a narrow strip at the finish line.
The width of the image of the strip is one pixel.
The camera makes images of the physical entities occupying this
strip many times per second.
Each image bears a time stamp and is stored in a computer graphics
system.
A progression of these images are pasted together in the
time sequence of their recording. Voila!
The CGS drawn line is associated with the timestamp of the image
upon which it lies.
I leave to you the pleasure of explaining the appearance of the
composite to yourself.

--
Michael Press

In article ,
Werehatrack wrote:
However, image manipulation
would almost certain;ly still be needed, given the image capture
method used, to keep the wheels from looking elongated or compressed.

This is a semantic question of what one calls manipulation.
The word 'manipulation' usually implies going from something which you
naturally have from a particular process, then adjusting it to something
that better fits some criteria not built into the process.

The pixels that are captured, have coordinates of vertical
position, and time. A 2-d image with coordinates of vertical and
horizontal position requires conversion of the time coordinate to
horizontal position. This conversion is simply multiplying the time
by some speed. But, what speed to use?

Using the speed of the riders gives you the natural aspect ratio
for viewing purposes. As far as the collected data is concerned
though, this speed is no more valid as any other - the camera doesn't
know or care what speed the riders go by at (and what if two riders go
by at very different speeds?), it only cares what time any particular
vertical stripe was recorded at.

If the image is captured onto a rolling film, there is an
inherent speed in the film's passage, and the image as found on the
film could very well be squished or stretched and would have to be
manipulated.
On the other hand, if the image is just captured
electronically, there isn't really a 'natural' horizontal scale
intrinsic to the data to begin with. Picking a suitable speed for
viewing is part of the display process - calling it a manipulation
is like saying that adjusting the volume control on your stereo is
manipulating the music you listen to. It is or it isn't, depending
on what you want to call the starting point for manipulation.

If the riders' speed is measured by radar or some other means,
and this speed, normalized to the vertical scale of the camera, is
then fed into the display process, the resulting images would have the
appropriate aspect ratio right off the bat. Is there still a notion of
manipulation in this case?

-Luns

On Mon, 11 Jul 2005 08:05:51 +0000 (UTC),
(Luns Tee) wrote:

In article ,
Werehatrack wrote:
However, image manipulation
would almost certain;ly still be needed, given the image capture
method used, to keep the wheels from looking elongated or compressed.

This is a semantic question of what one calls manipulation.

Coming from a background of photography, it is not uncommon for an
image to be unrealistic or unsuitable as recorded, and the recorded
image (whether captured on film or via other means) as originally
obtained is often manipulated to produce the final result. Sometimes
this manipulation is electronic (particularly today), sometimes in the
darkroom via a number of effects. But yes, anything that takes the
original data and renders it differently is a manipulation. And yes,
most digital imagery is, by that definition, manipulated in some way,
if only by resampling to reduce the pixel count to something that
requires less bandwidth. Throwing away data is the most basic form of
image manipulation in digital photography. In the case of the image
under discussion, the manipulations definitely included data
reduction, and some of them appear to have produced artifacts that
looked initially like zone remaps in the process. If we had access to
the uncorrected image, I suspect that it would have an amazingly long
aspect ratio; there's little need for extreme resolution in the
vertical plane, but the sampling rate is likely such that the raw
image rendered directly from the captured data would look like it was
stretched by a factor of ten or more horizontally. I'm sure they only
run the capture when there's actually an object approaching or in the
target area, and I would expect that they trigger the process using
the signals from the modules on the chainstays.

And in the world of the esoteric audiophile, yes, you will find that
there are those who hold, quite firmly, that adjusting the volume of
an audio signal is a manipulation. (In that case, however, my opinion
is "then so is having to listen to your diatribe over the music.") I
have occaisonally found hours of amusement by reading an audiophile
newsgroup or site to see the analog-vs-digital wars raging, the
tube-vs-solid-state screaming fests, and the discussions of the merits
of $2000+ speaker cables.

It's one thing to recognise and accept that changes occur from source
to presentaion. It's quite another to be a fanatic about trying to
eliminate the changes in every instance. Happily, because visual
media in general do not lend themselves to the same simplicity of
capture and recreation, the need for induced change (not always to
make the rendered image more like the original) has been accepted and
regarded as normal from the outset...but it's still there.
--
Typoes are a feature, not a bug.
Some gardening required to reply via email.
Words processed in a facility that contains nuts.