High Resolution With Pixel Shifting

Fascinating stuff.

If I have a particular feature I want to photograph I pull out the 10x microscope objective and shoot it. But if I'm documenting a new variety or other coin of interest, and I want to view details across the whole coin, I want a big image like the 156MP so I can look at all areas of the coin. The variety attributers like CONECA, Coppercoins, and Wexler shoot the individual markers when they publish, but there are often markers that are missed or overlooked, and the whole coin image will show these.

Hmm, I don't think so. It's really an either/or type of thing since it takes different optics and equipment to do what he is talking about. I'm using one system, and can change lenses to give different magnifications. If I want to do the 156/39MP type of image, I use one lens, while if I want to do the high magnification detail shot I change to a different lens. He mentioned shooting with a microscope with photo capability, which would not have the capability to shoot full coin pics but can do a good job on detail shots.

The difference between 156MP and 39MP is just how I save the file. The baseline is 156MP, but I can downsize it 2x and get 39MP. Viewing on a site like EasyZoom, the step sizes are factors of 2, so you can view at 156MP at full zoom (without "Max"), and when you zoom out one step you're viewing at 39MP. It's actually pretty cool, and makes me think that I'll publish at 156MP since 39MP is always an option.

Very cool! I like it; this has me considering an upgrade. The Nikon Z5 I currently have is unfortunately not capable of pixel shift.

I just finished an image of a Texas Quarter that strained my system due to the complexity. The requirement was a 24000x24000 pixel (576MP) final image. With "normal" stack and stitch this would have required a huge number of tiles, but with pixel shifting I was able to keep to 6 tiles, though even so it was a half-day effort. Main issue was having to move large amounts of data around, followed by the pixel shift processing itself. Stacking and stitching were relatively trivial in comparison. Here are the details:

6 tiles (3 tall x 2 wide)
Each tile has 20 stacked images
Each image has a group of 16 pixel shift RAW images
1920 RAW files, ~115MB each, total 220GB
Each RAW PS group renders to a 1.34GB TIFF file, total 120 files,160GB
Each group of 20 TIFF files stacks to a final 220MB tile, total 6 tiles
Final tiled panorama is 24000x24000, 576MP, 660MB

The final file is manageable using many programs, though not Digital Photo Professional, which is limited to 10000x10000 pixels. But intermediate files for sure caused me some strain. I was pleased though that Sony's Imaging Edge Viewer was able to batch process the RAW pixel shift files. It took an hour to process, but that happened offline without any input from me so no worries.

The final result is fairly impressive IMO. You can view it here on EasyZoom:

https://easyzoom.com/image/655297/a...?mode=manage

I think you'll find the coin interesting, as it has thousands of small divots and scratches from some unknown issue.

I can't keep this image up on the server for very long as it is using up a big % of my 2GB allotment. EasyZoom has not responded to my requests to increase storage so I'll need to delete it at some point.

Til then, enjoy!

The images are spectacular, of course. =)

I think there are separate use cases here. The super resolution seems better for examining new or disputed varieties. For example, I think that's the ideal tool to analyze something like a 1957-D 25¢ Misplaced D or your previous work on MMS-004/009 equivalence.

For basic variety identification or documentation, the extra level of resolution might not get the same mileage. Most consumers of reference images will not have equipment that can examine their coins in ultra-detail. I also find myself wondering if phenomena like Die Deterioration flow wouldn't obliterate such tiny markers.

I heartily endorse this idea. For well known varieties, I will often check the variety catalog sites, and then head straight to PCGS TrueView to examine the whole coin in high resolution color photographs.

It is also very true that the original listing puts you at the mercy of whatever the attributer found worthy as markers. I would assume that they often do not have multiple specimens to compare. The remaining information on the coin is simply lost.