The backstory here is that several months ago the basil plants in my kitchen developed an infestation of whiteflies. To go with the whiteflies, there was also a thriving population of small parasitic wasps, body length about 0.8 mm. Some stacked pictures of recently deceased specimens can be seen HERE, in-the-kitchen pictures of live specimens HERE.
Like all wasps, in flight these little critters link their front and rear wings together so they operate as one, using a set of small hooks called hamuli on the hind wing, that catch a fold along the back of the front wing.
The hamuli that we usually see are on large subjects, where the hamuli take the form of a long row of robust hooks; see for example HERE, HERE, and HERE.
Of course we would expect that a smaller wasp would have smaller hamuli, but I did not anticipate just how small or how few might appear. So I was surprised when I saw the mechanism, in one of my specimens that had expired with its wings still linked. Being surprised, I also got curious, and I took it as an interesting technical challenge to get a decent picture of those tiny hooks.
First try was to simply crop from the 40X NA 0.80 image that is posted at top of page:
That was pretty good, but the wasp was shot from the top, looking through the wing membrane to see the hooks which point upward. I wanted better.
So then I flipped the wasp over and shot from the underside:
That was not much better, because the hooks were then largely hidden by the folded section of front wing membrane that they were hooked onto. (Remember that this is with an NA 0.80 objective, which accepts a cone of light spanning 53 degrees off axis all around any focused feature. Blocking half of that cone essentially reduces the objective to only NA 0.40, losing half the resolution to diffraction plus a lot more contrast loss from the interfering membrane.)
Third try was to select a different specimen, attach a hair to the hind wing to use as a handle, then remove the wing from the wasp and mount the handle on an insect pin. That gave me a set of hooks that were not obscured by membrane, plus more freedom to orient the specimen to get a good point of view.
This image shows the 0.5 mm wing, stuck on a hair, in front of the objective. Just out of view on the left is a hemispherical diffuser, providing about 120 degrees arc of diffused light when the several flashes behind it went off.
And here is the final result, best I could get:
In addition to showing a fairly good view of hamuli, this last image also illustrates some of the vagaries of imaging at high NA. Along the edge of the wing, on the right side of the image, there are a couple of bristles that appear sort of split, as if the bristle had cracked longitudinally and spread slightly apart. That's an artifact. What actually happened is that as each part of those bristles went slightly out of focus, the out-of-focus rendition looked split, perhaps due to that insect pin splitting the illumination. The stacking algorithm decided that the out-of-focus split version looked sharper than the in-focus unsplit version, so of course it preserved the wrong appearance. I thought the artifact was interesting, and it would have been troublesome to fix anyway, so I decided to leave it as rendered.
Speaking of "leaving it as rendered", I decided to sharpen using good old-fashioned USM (Photoshop's unsharp mask) rather than any of new Topaz "AI" tools. The reason for that decision is that I was not comfortable with some of the artifacts that Topaz was introducing. I expect that their training set does not include many images with characteristics like these. Anyway I did not want to wonder later whether some feature was real or produced by an AI, so I went with the basic non-pattern-matching approach used by USM.
Methods & equipment: Nikon BD PlanApo 40X NA 0.80 objective with its collar cut off, Canon T1i camera, multiple flashes at low power. Focus step 0.5 micron for the first two stacks, 346 and 79 source images. For the third stack I dropped down to 0.25 micron, 246 source images. The nominal DOF at NA 0.80 is about 0.69 micron, but I used the much smaller value hoping for some improvement. That was only modestly successful. Comparing the results from processing every frame (0.25 micron spacing) with every other frame (0.5 micron), the finer spacing did produce slightly better clarity of some features, but I could only see the difference in flash-to-compare, not in side-by-side view.
--Rik
