THE CRYSTAL COSMOS: JOHN's GIGAPIXEL MICROSCOPY
TECHNIQUE
John started doing photo-microscopy way back in 1984. He
used an Olympus BH-2 microscope to
photograph tiny crystals grown on a microscope slide by either cooling a melt
or evaporating a saturated solution. With polarized light
illumination such crystals show astounding colors, and, along with
somewhat unpredictable and sometimes astonishing structures, proved a
fertile ground for exploration of this niche of the micro-world.
His early photographs were made on 4x5 inch sheet film. Prints
were exhibited at local galleries and special exhibitions.
Many of these early film images were eventually digitized and some
were submitted to the Nikon Small World photo-microscopy contest,
starting in 2000, where several received various
awards.
Over the years he returned to the microscope from time to
time to make videos of Oil
Films and Soap Films,
and to do 3D conversions,
Skip forward to 2020+. Covid severely limited
travel to make stereoscopic movies, or to show videos at 3D
conventions and film festivals. So what can be done
isolated in the
lab with
a 35 year-old microscope? The resolution of a print made
from a photomicrograph (even off a sheet of 4x5" low-grain
photographic film) depends primarily on the microscope objective lens
being used. Something like 20 mega-pixels is overkill for most
all visual-light objectives. Thus a print made from a 2X
objective lens will appear decently sharp up to about 11"x14".
When you make a substantially bigger print for a gallery, and walk up
close to it, parts of the picture will start to look fuzzy. One
popular approach is to stitch together several digital images into a
"panorama". Thus one might make an 8x4 foot print from 32 of these 2X pictures. This can work for subjects that are relatively smooth at small scales.
However, many crystals, like
these single frame shots,
or these,
have intricate details that are revealed at higher magnification.
We want to see these fine features up close and fully sharp, while
also being able to stand back (or zoom out) and observe the entire
panoramic structure. A solution is to use a high-magnification
high-resolution (high numerical aperture) microscope objective to zoom
in on the details (e.g. 10X to scan a 1X original-sized picture, 20X
to scan a ~2X image, 60X to scan a 4-10X original image). The
idea is to shoot a lot of these high-magnification images on an X-Y grid, and blend them together in order to
make a large print with gigapixel resolution.
There is one significant issue. Many crystals, especially when
grown from a melt, and when observed at high magnification, exhibit
focus problems. Focus can vary either as you scan around the
microscope slide, or within each individual X-Y picture itself.
Even hand focusing at each X-Y grid point won't always work, because
the specimen may be too thick for the high magnification
being used. It is not totally in focus. Thus, in most
cases it was found necessary to do "focus
image-stacking" at each X-Y stitching grid point, often with a 10X
objective, and always at 20X
and 60X. In "stacking", a sequence of images is obtained as the
microscope's focus is racked
through the specimen. Then a program (like Zerene Stacker) is
used to composite, or blend together, all the in-focus parts of each
frame in the stack. At each single X-Y position one all-in-focus
image is thus obtained. The collection of these is then stitched (or 'tiled') together.
One practical difficulty
is the acquisition and processing of the huge number
of images so generated. For example, a 20 by 20 X-Y stitch-grid with 50 focus-stacked
images at each location results in 20,000 pictures to be:
1) Acquired by scanning the microscope stage in
X and Y, and taking a focus stack at each location.
2) Downloaded from the camera flash card to a PC
and batch converted from RAW to 16-bit tiff's.
3) Reorganized into folders, each containing
all the stacked pictures (e.g. 50) for an individual location (e.g.
20x20=400)
4) The folders containing the "stacks" are
batch blended into (e.g. 400) all-in-focus images.
5) These are then stitched into a "panorama".
6) The final panorama is adjusted in Photoshop as needed (using the huge .psb
gigapixel file obtained from step 5).
Focus drive stepper motor. DIY dual X and Y stepper motor drives
mounted on an Olympus BH2 microscope stage.
John built a stepper motor drive for his old BH2
microscope X-Y stage, and mounted a timing-belt geared-down actuator for the Olympus BH-2's fine focus. He wrote a
program in visual-Basic to control acquiring the X-Y grid, while
commanding a "Stack Shot" focus-drive to get an image stack at each
X-Y location. Thus the "acquiring" part is fully automatic, and
takes typically a couple of hours. Next the sequential file-set is downloaded from the
camera flash drive to a computer, converted from RAW to 16bit-tiffs, and broken up into folders (by
another DIY program), each containing the focus-stacked files for a
particular location. These folders are processed using Zerene
Stacker's batch mode. The 20x20'ish output is stitched by PTgui.
This whole process is summarized in
an off-site video.
All in all, when stacking is needed it takes a full day, with the computers
working hard, to make one high-resolution gigapixel photomicrograph. But it's worth it!
|