June Electronics Swap - Eureka!

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

June Electronics Swap - Eureka!

Post by ray_parkhurst »

Saturday was the 4th electronic swap of 2019, and I decided to get out there very early. It opens at 5:00, and I was there around 5:45. About half of the eventual sellers were there, a few still setting up, so this is about as early as I'd like to arrive. It forced me to make multiple passes since sellers were arriving continuously, but that was OK as I ended up finding a few things on 2nd/3rd pass that I missed on the first.

On 2nd pass, a seller had just put out a bunch of boxes of laser optics. Most of the stuff I was clueless about, but the guy had a few small XY tables that looked interesting. One had Lin steppers and some really nice teflon ball screws, and there were 3 other complete low profile XY's plus some extra parts. I couldn't pass them up for $20 for all, and was glad I had showed up early!

Turns out these tables are almost exactly what I was planning to build for my Sensor-pan SnS system, except for the drives. They have Faulhaber micromotors, with integrated teflon ball screws, and 141:1 reduction gear. I am not sure how to drive these, so I ended up removing the drive system from one of the tables. This particular table had a broken drive nut, so was a good choice for experimenting. What I found after removing the drives was a high precision, 100mm travel XY table! Eureka!

Below are pics of a complete table with the micromotor drives, and then the XY table alone with drives removed:

Image

Image

apt403
Posts: 37
Joined: Mon May 06, 2019 5:29 pm
Location: Yelm, WA

Post by apt403 »

Nice score! Wish I was still in that neck of the woods for events like that.

What do you mean by "sensor-pan" stack and stitch?

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

apt403 wrote:Nice score! Wish I was still in that neck of the woods for events like that.

What do you mean by "sensor-pan" stack and stitch?
Rather than move the subject in XY under the lens (subject-pan), I plan to move the camera/sensor in XY across the image circle (sensor-pan). This should eliminate lighting and perspective issues but does require a lens with large image circle.

edited to add: there is a much bigger electronics swap in SoCal, and I've been trying to arrange a trip down there to coincide, but so far no luck.

apt403
Posts: 37
Joined: Mon May 06, 2019 5:29 pm
Location: Yelm, WA

Post by apt403 »

Oh, gotcha. Literally the panning the image sensor. Excited to see your build as it progresses! I've been curious about that sort of setup since I saw Lou's work w/ the Ultra Micro Nikkors.

What lens/sensor combos are you planning on using?

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

apt403 wrote:Oh, gotcha. Literally the panning the image sensor. Excited to see your build as it progresses! I've been curious about that sort of setup since I saw Lou's work w/ the Ultra Micro Nikkors.

What lens/sensor combos are you planning on using?
I have not upgraded my camera for quite a long time, and am still using a Canon T2i which had its AA filter removed. I call it my "HRT2i".

My primary photographic subject is the Lincoln Wheat Cent, and my goal for stitching is to create large, high resolution images which can be used for documenting the surface characteristics of variety coins (doubled dies, repunched mintmarks, etc). I have been subject panning at 2x magnification using a 95mm Printing-Nikkor (95PN) (see an example here: https://easyzoom.com/image/130646/album/0/4?mode=manage), but am interested (obsessed?) with going higher in magnification.

Lincoln Cents are 19mm in diameter, so at 3.5x I need a 67mm or larger image circle to do sensor-panning. I own a 105mm f4 Inspec.x L 3.5x (105IXL3p5) which has large enough image circle, but its f4 aperture limits sharpness to not much better than the 95PN. I started a search for a better lens in the 3.0-3.5x range, and when a Rayfact 3.5x came up for sale on eBay, I grabbed it! Initial tests show its image circle is at least 82mm, and that it shows excellent performance at f2.8, giving it NA of 0.14. The 95PN has NA 0.1 at f3.3 and 2x, so the Rayfact should give a sharper result.

This all needs to be proven of course, and I'm getting closer now to making that happen.

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

If someone only needs small sensor shifts, there are shift adapters for FF lenses on MFT cameras, and for medium format lenses on FF or APS cameras. You mount the lens and shift the camera. You can also do this with digital camera adapters on view cameras.

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Lou Jost wrote:If someone only needs small sensor shifts, there are shift adapters for FF lenses on MFT cameras, and for medium format lenses on FF or APS cameras. You mount the lens and shift the camera. You can also do this with digital camera adapters on view cameras.
My plan for the APS-C sensor is to stitch a 4x6 panorama. If I were using a FF sensor, I'd do a 2x3, and this is a big reason I'm considering FF, but I'd also need to drop the magnification below 3.5x to cover the Cent with some space around it and some overlap. I am not sure how far I can push the Rayfact 3.5x below 3.5x and keep a big enough image circle. Will be part of my experiments once I get the XY table going.

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Ray, how many pixels would you like to put across that cent?

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Lou Jost wrote:Ray, how many pixels would you like to put across that cent?
Right now I'm doing ~100MP with the 2x lens. This is 10k x 10k, so if it were a full sensor shot it would be 150MP (or 133MP for M4/3). With the 3.5x I will have a bit more overlap, so perhaps I'll be at ~250MP, or 375MP for full-sensor shot.

I'm expecting based on your question that you are thinking of the resolution increase of pixel shifting as a potential solution. Keep in mind that I am looking for sharp results at 100%, and I don't expect the pixel shifted images to hold up well to scrutiny vs optical magnification, though I'd be pleased to be proven wrong. I am still hoping that you can do a comparison shot of the detail you see with pixel shifting vs the real detail using optical magnification, only now I'd like to see that for the S1R. If I were to use pixel shifting on an S1R, I would likely plan to downsize the image 2x (back to original size) and use the pixel-shifting essentially as a sharpening tool to get better sharpness (and color resolution) at the original size. This is how I use SR now.

Edited to add: I think the best way to do the pixel shift vs optical comparison is to use a 2x teleconverter. Do a 2x pixel-shift zoom, followed by a 2x teleconverter zoom without pixel shift, and compare. I would even think a 2x teleconverter zoom WITH pixel shift, then downsized 2x, would be useful to reduce the potential issues with demosaicing.

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Ray, I was mostly wondering just how many tiles you really need, though you are right that shifting was also in my mind.

10k pixels is less than two FF tiles wide, even at nominal resolution. Using pixel shift, and downsampling to nominal resolution, you might be able to cover your coin with the simple manual shift adapters I mentioned above, though of all the people on the forum you are one of the best prepared to automate the process. I mention it for others, who may be looking to obtain that kind of resolution from a simple, cheap manual sensor-shifting set-up.

ray_parkhurst
Posts: 3413
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Lou Jost wrote:Ray, I was mostly wondering just how many tiles you really need, though you are right that shifting was also in my mind.

10k pixels is less than two FF tiles wide, even at nominal resolution. Using pixel shift, and downsampling to nominal resolution, you might be able to cover your coin with the simple manual shift adapters I mentioned above, though of all the people on the forum you are one of the best prepared to automate the process. I mention it for others, who may be looking to obtain that kind of resolution from a simple, cheap manual sensor-shifting set-up.
Lou...I've contemplated various ways to do the stitching, including (per your earlier suggestion) a simple 2-tile pan like you're describing above. For example:

Cent diameter: 19mm
Desired frame coverage: 90%
Effective tile width: 21.1mm
FF width: 36mm
Required magnification: 1.7:1

I could push a 105PN up to 1.7:1 using a 1.4x teleconverter but Feff is f12, so my current 95PN solution would probably still beat it.

What I'd end up with using a S1R in single-shot mode would be an 8368x8368 image similar to what I have now, possibly a bit less sharp. Doing pixel shifting to get a 16k x 16k image would increase the IQ of the original image, but still not sure about the quality at 100%. Certainly it would be limited by optics.

Post Reply Previous topicNext topic