Ultimate Sharpness

It's funny that you mention the idea of attaching the diffusor to the lens instead of a platform like I've been doing; it's something I've been thinking about but haven't gotten around to (I've been busy with life stuff, which is why I have been on the forums less and inactive with coins lately.) I did the platform thing because it was easy to do with stuff I had laying around (and I could experiment with the height), but I think the idea of attaching it to the filter threads would be ultimately nicer to work with.

I actually was thinking I would play with that idea at some point in the future when I had saved up enough money for a 3D printer. I could "manufacture" different diffusors with in different configurations.

I would love to have copies of the RAW files. I can see visible JPEG artifacts in the images you've posted here and on Photobucket, so I don't think I can appreciate the true sharpness. I could also render them in Lightroom with different RAW profiles, which would allow for a nice compare/contrast between DPP and Lightroom. I expect the biggest difference will be in color, but since they use different demosaicing algorithms, the sharpness might also be affected. I also have a custom RAW profile made specifically for the Jansjos, which gives me more accurate color with copper (I had another thread about that).

One tip with stacking (this is with Zerene; it may not apply if you're using something else) - IMO, using DMAP gets more "natural" results than PMAX. PMAX gives a stronger first impression because it can exaggerate contrast and sharpness (which increases as you use finer steps), but it also "oversharpens" and will add noise to the image (noise appears as stronger local contrast, so for any given pixel, PMAX will pick the frame that is sharp and has the greatest noise). DMAP is a bit more fussy, and requires some manual intervention, but I think the results are usually better.

There is a third option: Stay out of diffraction territory but allow some areas of the image to be out of focus.

The problem is that with small coins we are right on the edge where it's not clear which is best. There is little benefit to open up the lens beyond the DLA limit, so for the T2i, with DLA of f/6.7, even f/4 is OK for Cents when you take the effects of AA filter into account. With extremely careful focusing on the surface of Lincoln's throat, and perfect stage flatness, you can pretty much get away without stacking at f/4 with the 75ARD1, and I was actually able to get an acceptably-sharp image at f/3.3 with the 105PN.

I need to try Zerene and Helicon to see if I have enough control over the process (as the 'Cap suggests) to make a better result. If so, then I expect the stacked image to be improved over the single.

Now, for other cameras with bigger pixels, such as the XS with a DLA of f/9.1, I can shoot at f/4.7 safely, and up to f/5.6 without a lot of degradation. This will give enough DOF that single images should be superior.

The lesson I've recently learned is that it gives little benefit to open aperture wider than needed to be within the DLA of the sensor you are using. This is a bit new to me but has been making a lot of sense.

I think you're using a T3i, correct? Same sensor as my T2i, so your DLA is f/6.8. For a Dime, with diameter 18mm, full field the mag is 14.9/18=0.83, so at f/5.6 you are at feff=f/10.2, well into diffraction-limited territory. Even at f/4.7, you are diffraction-limited with Dimes, and the higher mag has shallower DOF vs Cents. I think the cutoff between getting away w/o stacking and needing to stack may be between Cents and Dimes with the 18MP Canons.

The 75ARD1 field flatness degrades as you move away from 1:1 wide open. Lens degradations are unpredictable, so the aberrations you are seeing are likely due to whatever is affecting the sharpness (distortion, CA, etc).

I'm completely in agreement with you on not wanting to focus stack most of the time. I only want to go through that level of trouble on special shots, or of course as a demonstration or discussion topic. Most published shots are at 3-4x reduction, so you can technically shoot at a much smaller aperture and not see diffraction effects in the final image.

Of course they don't. I usually do frame my shots to nearly fill the frame, but for what I consider practical reasons. I want my final shots to be around 800x800. On the T2i, which is 3456 tall, I want the coin to be just inside 3200 tall, and then 4x reduced for publishing. The XS is 2592 tall, so I make the coin just inside 2400 tall, and reduce 3x. In the end you can't really tell which camera the shot was taken with. The reduced diffraction I get with the XS is offset by the smaller downsizing ratio...

edited to add:

Actually, the 105PN I'm using was built in the early 70's. We're talking 40 year old technology here and I literally, honestly have not identified a lens made before or since that can do what it can.

For ultimate sharpness, it would be possible to do a stack-and-stitch; both a focus stack and a panorama using a high NA microscope objective. This thread over at photomacography.net shows a configuration of a Nikon CFI 10X/0.25 Plan Achromat on a Canon 100mm macro. It's an infinity corrected objective (and is therefore intended to be used with a 200mm tube lens), so putting it on the 100mm means it's operating at about 5x. But it has a big enough image circle and still preforms really well at that magnification.

A NA of 0.25 has an effective aperture of about f/10 at that magnification; the photographic lens equivalent of about f/1.7. It would be a lot of work, hundreds of individual frames, but an enterprising VAM fan could produce a 2 gigapixel image of a Morgan dollar.

I actually did a 2 image panorama of an Eisenhower dollar at a time when my setup didn't let me get the whole thing in a frame. I was worried about the lighting creating a mosaic look, but you can't see the seam that easily. It would probably be worse with more images though.

You've mentioned before that the 5x MM objective is nearly telecentric, which would make it ideal. But even if you moved the camera instead of the coin, or moved the lights with the coin, the luster might still move because the surface is partly reflective and light/coin/camera angle will change. That wouldn't be a problem on a subject like a flower or whatever that is mostly scattering light. Maybe the effect wouldn't be TOO big...

I think that could be avoided by pivoting the camera/lens around it's "nodal point" (entrance pupil? I'm not sure what the correct terminology is) instead of panning it. But that would mean having much deeper stacks toward the edge of the coin because it wouldn't be flat to the lens.