DIY Stack and Stitch System

[mawyatt2002]

Hallo,



Nice you to try extreme macro panorama. I’ve tried different way to x y axis step of the subject to photograph but the problem occurred to have stitching problems because the subject changes its perspective for the pano stitching is that not the right way. I found a easy way semi automatic.
Are you interested to know about ?

Best regards

Olivier

Hi Olivier,



Yes very interested, please note.

This problem of perspective change can be reduced by using lenses that are telecentric or close to telecentric. Often use the Mitutoyo 5X, 10X, 20X & sometimes 50X infinity corrected lenses, for larger subjects I use the Printing Nikkor PN105 F2.8A.



Best,

--
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

 

[Macrodesigner]

Hi, to avoid most of the stitching pano problems the telecentric lens is the right choice. I have also the printing Nikon 105mm 2.8 which is one of sharpest lens. I use it about a range of f4 which is a comfortable opening. For pano stitching I shift a modified GoPro hero 7 remotely connected to the controller. 12 mp is enough to make rows of stitched raw images. The solution is the lens remains still and the camera shift with a modified adapter bellow connect to micro tables to shift if to x y position. Why a GoPro? Because the chip has got a 5.7 crop factor that do not need too much magnification factor. This technique allows a perfect raw Stich without error. This technique works well with asp c sensor but because of the sensor uses also the edges of the printing Nikon 105 (some aberration occurs) that also occurs on the Nikkor m 1x 3x 5x 10x 20 Measure Microscope. The working distance is up to 100 to 25 mm. By using the middle Center of the lens the aberration is less. If I need more resolution, I use a Labor camera 1/2.5 “ camera but the quality of the (jpg) is compressed and the dynamic range is less. A last camera that I have good result is the new Nano1, but unfortunately there is no remote control, the Stiche image has to be done by time laps synchronised with the controller. It worked also well here a picture test.



it would be nice to have more automation but because each sample is complete different it is better to have some visual control, there is still post and manual correction to do.

What experience did you have ?
Regards Olivier

 

[mawyatt2002]

Hi, to avoid most of the stitching pano problems the telecentric lens is the right choice. I have also the printing Nikon 105mm 2.8 which is one of sharpest lens. I use it about a range of f4 which is a comfortable opening. For pano stitching I shift a modified GoPro hero 7 remotely connected to the controller. 12 mp is enough to make rows of stitched raw images. The solution is the lens remains still and the camera shift with a modified adapter bellow connect to micro tables to shift if to x y position. Why a GoPro? Because the chip has got a 5.7 crop factor that do not need too much magnification factor. This technique allows a perfect raw Stich without error. This technique works well with asp c sensor but because of the sensor uses also the edges of the printing Nikon 105 (some aberration occurs) that also occurs on the Nikkor m 1x 3x 5x 10x 20 Measure Microscope. The working distance is up to 100 to 25 mm. By using the middle Center of the lens the aberration is less. If I need more resolution, I use a Labor camera 1/2.5 “ camera but the quality of the (jpg) is compressed and the dynamic range is less. A last camera that I have good result is the new Nano1, but unfortunately there is no remote control, the Stiche image has to be done by time laps synchronised with the controller. It worked also well here a picture test.

it would be nice to have more automation but because each sample is complete different it is better to have some visual control, there is still post and manual correction to do.
What experience did you have ?
Regards Olivier

Hi Olivier,

Yes that is a good technique to use, move the sensor and not the lens. Some folks over on photomacrography have done similar techniques but with much larger lens than PN105, these lenses were used in semiconductor photolithography and had very large aperture, something like 60~100mm I recall. Of course these are massive lenses and require special mounting.

Your use of the GoPro seems like a very good approach.

We've done S&S sessions that produced ~ 20,000 by 15000 pixel images with the Mitutoyo lens, and soon will be doing 30,000 by 20,000 or more. These originally took days to capture, then when completely automated took almost a day, now we've got this down to a few hours. These images are of semiconductor chips, with features that are precise and orthogonal, so any perspective changes would corrupt the final rendering...thus the importance of using telecentric lenses.

Now we've developed extreme precision stages (nanometer levels) based upon piezoelectric devices, these will be utilized for very high magnification S&S sessions and later hopefully camera body pixel shifting (shift the entire camera body precise pixel dimensions). Here are some links.

Piezoelectric Stage Development Evolution, Part A - www.photomacrography.net

Piezoelectric Stage Development Evolution, Part B - www.photomacrography.net

Piezoelectric Stage Development Evolution, Part C - www.photomacrography.net

Closed Loop Piezo Stage Operation. - www.photomacrography.net

Really like your GoPro approach, very creative!!

Best & Happy New Year

--
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

 

[Luisifer]

I did close to 30x20 (18.9 x 28.4 after cropping and resampling down, stitched resolution was 21186 x 30037) but i found that ACR can open only 536.9 Mpx file. That is quite sad because i like to use ACR as filter on the result.

Btw i started with shooting at 10:48 and last frame was taken at 22:14 but i had to leave for a few hours and process waited for me - shooting of 36 stacks was not fully automated this time.

 

[mawyatt2002]

I did close to 30x20 (18.9 x 28.4 after cropping and resampling down, stitched resolution was 21186 x 30037) but i found that ACR can open only 536.9 Mpx file. That is quite sad because i like to use ACR as filter on the result.

Btw i started with shooting at 10:48 and last frame was taken at 22:14 but i had to leave for a few hours and process waited for me - shooting of 36 stacks was not fully automated this time.

Nice!

Just checked earlier in this thread, the one image was over a year ago and was 27500 by 21400 pixels for 2000 individual images, not 20,000 by 15,000 as I had noted above. Think it was something ~440MB also, big files indeed!!

Much effort has been put into reducing the image to image time, even at very high magnifications, now well under 2 seconds per image in the fully "hands off" automated mode. This allows more "reasonable" S&S sessions, hours rather than days.

The past few months have been devoted to developing the piezoelectric stages, these perform at levels even the excellent THK rails and superb Trinamic controllers can't approach, which are the best I'm aware of.

These piezo stages are completely "Closed Loop" systems without any backlash, stiction or mechanical limitations of systems that rely on moving mechanical surfaces. Also no load induced detectable stage movement or creep and even 1st order temperature compensation. Nanometer levels of precision and repeatability but able to transition 100's microns in fractions of a second. No gears, motors, screws or nuts to deal with. Actually there are no moving mechanical surfaces involved, just ceramic piezo element expansion (material force) and flexures (they bend instead of sliding or rotating).

With your mechanical skills & equipment you should look into these piezo stages, they are remarkable performers!

Note the hairdryer test notes here.

Closed Loop Piezo Stage Operation. - Page 2 - www.photomacrography.net

Using a high magnification (20X Mitutoyo and camera 16X) at 320X effective, a hairdryer on Hot and Max Airflow was blown on the piezo stage with no apparent movement in the controlling axis when viewed with Live View, the temperature was increased by well over 25C. When this was done to the THK K20 the Live View image quickly shifted off the screen due to thermal effects as soon as the hot air flowed, the piezo stage utilizes feedback to sense and control the stage.

You really have to experience these piezo stages to see just how good they are when operated Closed Loop!!

Best,

 

[Luisifer]

It should be interesting to play with it too. But i have to stay with my μGUN for now and just develop better SW for it or in case of extreme troubles (f.e. with backlash) add some precize linear magnetic metering system or expand to the more motorized axes.

Mainly because of money are gone and results are adequate for 1s interval shooting together with steady movement.

On the other hand i would like to start/continue to work on my 12x12" technical camera based on materials like titan gr5. For now i have just lens plate and few ideas how to join other parts with precize motorized movements.

 

[mawyatt2002]

It should be interesting to play with it too. But i have to stay with my μGUN for now and just develop better SW for it or in case of extreme troubles (f.e. with backlash) add some precize linear magnetic metering system or expand to the more motorized axes.

Mainly because of money are gone and results are adequate for 1s interval shooting together with steady movement.

On the other hand i would like to start/continue to work on my 12x12" technical camera based on materials like titan gr5. For now i have just lens plate and few ideas how to join other parts with precize motorized movements.

Nice setup. That's certainly a hefty stage for one axis control, is it stable on a tripod? The cheap plastic XY stages I have are not very good (have 2 or 3 of them), is the XY stage you are using for the subject stable enough for precision work, it looks like the ones I have?

If you develop your own piezo controller and get surplus piezo stages like we did, it's not very expensive, although the OEM controller and stages will cost over $5000, some much more.

The setup we have uses the Piezo Stage to move the subject and the THK KR20 or KR26 to move the camera/lens. This allows the use for conventional stepper motor stacking with the THK rail and ultra-precision stacking with the Piezo Stage.

Here's some images of the stages and custom developed controller/driver/amps. The controllers feature direct RPi interface and have 16 bit resolution DACs and ADCs capable of a nanometer resolution (you can command the stage with 4nm step resolution with the DAC and read the actual stage position with the ADC). Everything runs from a single 12 volt source at less than 1/2 amp.

This design was done to directly interface with the Automated S&S system and all the drivers, interface, and control code has been developed, including initial stacking routines. Various testing programs were developed to help get things setup and operational, including the ability to actually measure (not speculate) the rail performance at nanometer levels. These tests verified the Physik Instrumente specifications for similar stages (these stages were custom developed, the "K at the end, and PI won't release any data on such). All code is in Python for easy adaptation and use, and runs on the RPi.

Now I'm busy building a couple systems for folks, one based upon augmentation of the popular WeMacro Vertical Stage for precision use, machined components should be here this week.

See threads previously mentioned on PM for more details.



Best,






Physik Instrumente




PI P601K Stage with XYR micrometer stages for subject positioning




PI P601K Stage with XYR Subject Positioner




Custom Developed Piezo Stage Controller/Driver/Amp for RPi




Another view of Controller/Driver/Amp




Another ControllerDriver/Amp view




PI P603K Piezo Stage




Another P603K Stage view




Surplus OEM P603K Driver/Amp with Custom Developed Controller for RPi




Custom Controller mounted on top of surplus OEM P603K Driver/Amp













--
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

 

[Luisifer]

mawyatt2002 wrote: That's certainly a hefty stage for one axis control, is it stable on a tripod? The cheap plastic XY stages I have are not very good (have 2 or 3 of them), is the XY stage you are using for the subject stable enough for precision work, it looks like the ones I have?

It is three axis control (not only one).

Scene/subject is stable enough. It is only static setup (relatively provisional) for simplifying of preparation default scene position. And yes i was inspired by your setup and chose Kiwi FC-III.

 

[mawyatt2002]

mawyatt2002 wrote: That's certainly a hefty stage for one axis control, is it stable on a tripod? The cheap plastic XY stages I have are not very good (have 2 or 3 of them), is the XY stage you are using for the subject stable enough for precision work, it looks like the ones I have?

It is three axis control (not only one).

Scene/subject is stable enough. It is only static setup (relatively provisional) for simplifying of preparation default scene position. And yes i was inspired by your setup and chose Kiwi FC-III.

Ok, didn't see the other stepper motors!! Very nice indeed!!

What is the Kiwi FC-III? Is this the controller?

Best,

--
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

 

[Luisifer]

It is for raw moving (setting) of scene between ball heads:

Amazon.com : Kiwifotos Deluxe Macro Focusing Rail Slider for Canon EOS 6D Mark II 5D Mark IV III 90D 80D R6 R5 RP Rebel T6 T7 T8i T7i Nikon Z5 Z6 Z7 Z50 D750 D780 D850 D7500 D7200 D5600 D5500 D3500 D3400 & More : Electronics

Amazon.com : Kiwifotos Deluxe Macro Focusing Rail Slider for Canon EOS 6D Mark II 5D Mark IV III 90D 80D R6 R5 RP Rebel T6 T7 T8i T7i Nikon Z5 Z6 Z7 Z50 D750 D780 D850 D7500 D7200 D5600 D5500 D3500 D3400 & More : Electronics

www.amazon.com

 

[mawyatt2002]

That's the same as I have, just under a different name, they came with WeMacro Vertical Stands.

Both mine (may have 3 but don't know since I never use them) are not stable enough for my general use, yours must be much better.

Best,

 

[Luisifer]

Maybe later it will be replaced by T/R axes...

 

[mawyatt2002]

Maybe later it will be replaced by T/R axes...

The WeMacro XY stages are quite good, small at 40mm square, and don't cost any more than the Kiwi.

XY Stage.

Fine X-Y adjustment platform 2.0 – Wemacro Rail

www.wemacro.com

Here's the XYR Stage.

XYR stage with specimen holder – Wemacro Rail

www.wemacro.com

I use 3 of these stages, they are really accurate and stable.

Best,