This is basically how an optical CMM (OMM) works. There are tricks to determining and calibrating out alignment issues and determining snapshot viability, and you're touching on both of them. Very well done!
I am also just about to attach my microscope to a CNC system, but not for taking pictures of two dimensional manufactured objects but rather for Z-stacking use in biological light microscopy. Probably some horizontal stitching too.
The prompt was that someone gave me a broken project which they had never finished (small-scale benchtop CNC system which is highly rigid and heavy but had no wiring, an ancient power supply and a busted spindle motor). I considered finishing the project but the overall size was too small to be very useful for any CNC work I would want (we used to have a large scale commercial 4-axis!). Therefore I am retrofitting a new control system [acquired], rationalizing the power supply [done], and clamping on a USB industrial camera based microscope [clamp acquired, intermediate mounting plate TBD].
On my current microscope mount it's really annoying to zoom a little, take a shot, zoom a little, take a shot since there's a high chance of bumping the sample or some slight vibration affecting a shot, and it's very hard to move a tiny amount due to high-friction macro adjustment interface. I was part-way through designing a fix, with some axis modifications for a motor mount, but then realised it would be easier to just redesign the mount from scratch rather than retrofit a one-off modification. Before allocating time to get that done, the CNC fell in to my lap.
So you wanted to take pictures of cdroms with a microscope, going on a decent detour and the other way you can make a microscope from a disc player which is a really cool project [0]
Somehow that seems less fun. Take a thing that was already designed to image data pits on a CD, and use it for exactly that, but in the wrong way. Might as well just plug a CD player into a PC and write some code to generate a bitmap image of the CD from the data it contained.
I'm currently doing something similar to build a photographic film scanner. I will say that I've found that moving the optics is generally much more error and vibration prone than moving the target. I'm actually using a 2 axis microscope stage as the basis for my scanner, ironically enough, and CNC spindle z-axis for focus.
I'm shooting to do 35mm and medium format. 4x5 is a stretch goal. Even illumination is definitely a challenge, though I use the same technique you do, which is generally refered to as flat field calibration, where we capture the field of light without any target and use it to calculate the offset to apply an even field. One of the trickier aspects is finding affordable lenses with appropriate magnifications that can focus evenly out to the edges of the frame. There are various lenses pulling from scanners or that were specifically used for copying that are useful if you can find them.
I started with this but ended up designing my own system around a vertical 4040 aluminum extrusion post and various 3d-printed components including an XY stage (often costs $10K or more) and a Z stage (for focus). It was quite challenging to get everything to line up in a single optical axis and keep the dust out but the results have been quite good- I wrote my own software after trying to get micromanager to work, and can do large acquisitions (25x75mm, AKA a single microscope slide) as well as real-time object-detector based tracking.
It's remarkably hard to equal or best a $150 scope from Amscope in terms of optical quality, it's automating their stages that is tricky.
Nice. Try to image a vinyl record. Maybe someone can write something to play such images. The Library of Congress has such a system, called IRENE, but it was very expensive.[1] This might be a low-cost approach.
I just added an extra image where you can see contours being drawn around the 'pits' of the CD-ROM. I need to improve the speed of the autofocus etc. before I can image a whole CD heh.
Heh yeah, I split the area to be imaged into squares, and run autofocus in the middle of the square. Then use a 'snake' like XY pattern across that square to create images.
But I need to dramatically increase the speed of this process to use across a whole CD etc.
As long as you're scanning over a perfectly flat, if tilted (relative to the microscope's focus plane) 2d plane, you don't need to autofocus much. Just measure the focus at a few points and fit the plane, then run the scan and apply the focus parameters from the fit.
This is basically how an optical CMM (OMM) works. There are tricks to determining and calibrating out alignment issues and determining snapshot viability, and you're touching on both of them. Very well done!
Thanks, I assume https://en.wikipedia.org/wiki/Coordinate-measuring_machine is what you're referring to, which look interesting, hadn't come across them before.
I am also just about to attach my microscope to a CNC system, but not for taking pictures of two dimensional manufactured objects but rather for Z-stacking use in biological light microscopy. Probably some horizontal stitching too.
The prompt was that someone gave me a broken project which they had never finished (small-scale benchtop CNC system which is highly rigid and heavy but had no wiring, an ancient power supply and a busted spindle motor). I considered finishing the project but the overall size was too small to be very useful for any CNC work I would want (we used to have a large scale commercial 4-axis!). Therefore I am retrofitting a new control system [acquired], rationalizing the power supply [done], and clamping on a USB industrial camera based microscope [clamp acquired, intermediate mounting plate TBD].
On my current microscope mount it's really annoying to zoom a little, take a shot, zoom a little, take a shot since there's a high chance of bumping the sample or some slight vibration affecting a shot, and it's very hard to move a tiny amount due to high-friction macro adjustment interface. I was part-way through designing a fix, with some axis modifications for a motor mount, but then realised it would be easier to just redesign the mount from scratch rather than retrofit a one-off modification. Before allocating time to get that done, the CNC fell in to my lap.
So you wanted to take pictures of cdroms with a microscope, going on a decent detour and the other way you can make a microscope from a disc player which is a really cool project [0]
[0] "Hacking CD/DVD/Blu-ray for Biosensing" https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6066758/
Somehow that seems less fun. Take a thing that was already designed to image data pits on a CD, and use it for exactly that, but in the wrong way. Might as well just plug a CD player into a PC and write some code to generate a bitmap image of the CD from the data it contained.
If you photograph and stitch a panorama of a CD at a resolution with the data visible and distribute it under your copyright, is it piracy?
Piracy is not a technical definition but a legal one.
If you zip/print/scan/ocr/train ML on/restore from ML its still piracy even though none of the pixels are directly transferred.
The ML one seems to not be piracy from a legal POV.
Like distributing photocopies of a book in braille
Yes.
I'm currently doing something similar to build a photographic film scanner. I will say that I've found that moving the optics is generally much more error and vibration prone than moving the target. I'm actually using a 2 axis microscope stage as the basis for my scanner, ironically enough, and CNC spindle z-axis for focus.
Cool, that sounds really interesting, what size film are you using that for? Is illuminating the film evenly hard too?
I'm shooting to do 35mm and medium format. 4x5 is a stretch goal. Even illumination is definitely a challenge, though I use the same technique you do, which is generally refered to as flat field calibration, where we capture the field of light without any target and use it to calculate the offset to apply an even field. One of the trickier aspects is finding affordable lenses with appropriate magnifications that can focus evenly out to the edges of the frame. There are various lenses pulling from scanners or that were specifically used for copying that are useful if you can find them.
I started with this but ended up designing my own system around a vertical 4040 aluminum extrusion post and various 3d-printed components including an XY stage (often costs $10K or more) and a Z stage (for focus). It was quite challenging to get everything to line up in a single optical axis and keep the dust out but the results have been quite good- I wrote my own software after trying to get micromanager to work, and can do large acquisitions (25x75mm, AKA a single microscope slide) as well as real-time object-detector based tracking.
It's remarkably hard to equal or best a $150 scope from Amscope in terms of optical quality, it's automating their stages that is tricky.
Nice. Try to image a vinyl record. Maybe someone can write something to play such images. The Library of Congress has such a system, called IRENE, but it was very expensive.[1] This might be a low-cost approach.
[1] https://en.wikipedia.org/wiki/IRENE_(technology)
Would be really cool to read a whole CD this way. Actually, that wouldn't sound any different. You could read an LP though.
For vinyl, there's always the laser player (used for preservation, too)
https://www.elpj.com/
I just added an extra image where you can see contours being drawn around the 'pits' of the CD-ROM. I need to improve the speed of the autofocus etc. before I can image a whole CD heh.
greaseweazle for CDs
See also the EnderScope:
https://royalsocietypublishing.org/doi/10.1098/rsta.2023.021...
https://github.com/Pickering-Lab/EnderScope
ironic in some way?
>we want to detect plastic
>microplastics bad
>we built an open source masheen to detect evil plastic
>why yes, to build it you must unleash the amazing power of plastic
Getting that Z Axis perfectly level across the run is going to be fun at this scale.
Heh yeah, I split the area to be imaged into squares, and run autofocus in the middle of the square. Then use a 'snake' like XY pattern across that square to create images.
But I need to dramatically increase the speed of this process to use across a whole CD etc.
As long as you're scanning over a perfectly flat, if tilted (relative to the microscope's focus plane) 2d plane, you don't need to autofocus much. Just measure the focus at a few points and fit the plane, then run the scan and apply the focus parameters from the fit.