Hi Rob,
The stacking of images has been done for a
long time in laboratories using electron and
light microscopes. We did this in the late
70s on an HP2000 system in my Institute. I
was able to entice a virologist from Holland
(despite Apartheid) to visit for a few
months. His work was mainly optical
transforms. We then developed a suite of
image processing programs using Tony
Crowther's (MRC Cambridge) ideas as a start.
But it took 8 hours or more to do a FFT, mask
and reverse FFT on a 512 x 512 matrix. I can
do it in a minute or two. And that time is
taken making the mask. The actual processing
is over before one's finger leaves the Enter Key.
But a straight answer is that it was not
possible to make images like the one you are
talking about outside very big facilities
like mine. Instead, people used dozens of
images at different levels and different
orientations of the specimen to show structure.
The colour fringes are not a problem unless
one is concerned with aesthetics. I think it
would be less work to fix the final image
rather than the individual frames. I'll look
into this. What do you use for the job? The
collector, the condenser, the objective and
the eyepiece all contribute to the problem
and a combination of optics to eliminate them
entirely would cost about £20 000 at a rough
guess. As it is the objective I used for this
lists at about £1200.
Don
Rob Studdert wrote:
On 16 Feb 2006 at 9:12, Don Williams wrote:
That must take a long time. CombZ will stack
100 pictures in five or six minutes. Helicon
Focus (less configurable) is faster, but in
the case of the blue flowers it produced
strange halos in some places -- around very
dense areas for example. Having both programs
is useful for the stuff I do. When things
move, both produce artifacts. But should I
start injecting microscopy into this group?
Perhaps it would be out of place? But maybe
not as far out as religion?
Please don't hesitate, I'd much prefer to read about microscopy.
As an aside I've found that automated applications which use stacked images at
different exposures to generate a single wide exposure latitude image also
exhibit some very noticeable halos and colour aberrations even if there is very
slight movement in in the subjects between frames.
Anyway here goes:
http://www.kolumbus.fi/mimosa/HOLD/401-452.jpg
this is a stack of 52 images in 2 um steps
taken at about 450X under phase contrast. The
object is the Lorica (siliceous skeleton) of
a small aquatic protist -- the name of which
escapes me for the moment but is irrelevant
anyway. This is a good example of how well
CombZ does it's job. The depth of focus
obtained is about 100 micrometres.
That's a pretty amazing image, how was this type of image achieved before
digital stacking was available? Also I noticed significant CA in the composite
image, do you apply any CA compensation to the individual images before
subjecting then to the combine process?
Not often would I envisage that I'd need to stack more than 4-5 images in the
case of 1:1 macro subjects, this is why it's not such a daunting task to effect
it manually. Also the control afforded by being able to decide which areas you
wish to extend the DOF in rather than it being a blanket treatment allows a bit
more creative freedom. I expect that CombZ would be of use to me occasionally
but I will need to experiment. Thanks again.
Cheers,
Rob Studdert
HURSTVILLE AUSTRALIA
Tel +61-2-9554-4110
UTC(GMT) +10 Hours
[EMAIL PROTECTED]
http://members.ozemail.com.au/~distudio/publications/
Pentax user since 1986, PDMLer since 1998
--
Dr E D F Williams
__________________________________
http://www.kolumbus.fi/mimosa/index.htm
http://personal.inet.fi/cool/don.williams
See feature: The Cement Company from Hell
Updated: Added Print Gallery - 16 11 2005
