Increasing resolution by means of closeup and/or better sensor

what do you think would be the limiting factor? Naturally I would also need a stepper motor with a very small degrees/step ratio, otherwise I would be getting oblongated voxels, but other than that are you accounting for the suposedly 13mp-something sensor to be so noisy that it wouldn't result in a real increase in laser line detection definition? Or what else?

Yes, I would definitely need thin lines on the lasers, and 0.9° step angle on the motor (or better a reduction gear assembly), yet as long as I can put together this little frankenstein of a scanner, and all I need to change are focus on camera/lasers and turntable assembly with reduction gear, but no low-level tinkering with the hardware and software, than I am all down for the experiment.

Thank you for your ideas, I first want to tackle the project of building a PET filament extruder, which would be more functional to my short term needs, and then enjoy myself with the scanner project. I saw also a laser line scanner that tilted the model in several positions with several successive scans. Structured light also makes more sense, being faster and catching probably more details, but these are too techy approaches out of my DIY reach. I will test instead a closer distance and, if I can manage, a higher resolution camera, not a pi camera obviously but a high end USB webcam should work fine as well with the software.
Regarding the law, I am not subject to fda as I live in Europe, yet in my opinion that should regulate only commercial machinery. As long as it's the doctor who deals himself with the preparation of the X appliance, none should be interested in how he does it since he's the doctor and is taking full responsibility for his job, I could be using black magic for all I care as long as I get the job done ))

Hello pirvan
I understand your opinion, and I am sorry if what I wrote gave you the impression of superficiality, or boldness, or, well, you name it
I am the doctor in this example, and I also have a passion for technology: I understand the certification requirements for medical equipment, especially the ultimate aim of it, which is obviously the safety of both personnel and patients, yet neither the 3d printer which I am testing in order to build transfer splints, or the custom atlas 3d which I might build in the end, are medical equipment nor will be used as such, since they will be off-office, and working on gypsum models. I, as the doctor, do not endanger the safety of my patients by putting them in direct contact with these contraptions (well, you and me know I wouldn't do that even if they were, and that 90% of the certificate business, is, well, just "business", which ends up making clinicians spend sometimes ten times the money needed to buy the same technology on the domestic market, obviously switching those costs on the end users), on the other hand with the help of computer aided planning on precise 3d models I am in fact offering a better service, without the cost overhead of "commercial" solutions.

I also understand you are a very proficient user of this scanner since before posting here I noticed your interventions in the forum here and there: did you have the chance to explore the possibilities of closing the distance to the camera in order to get better resolution, or have otherwise some tips specially directed to this end? Also, is the declared 0.25mm resolution a real-life expectation?

I'm unaware of any other camera module that has been interfaced using the RPi's camera port. It is not a trivial matter because of the restrictions on access to the software that interfaces with the camera chips (my understanding). USB cameras can be interfaced to the RPi but I think they have to conform to the Universal USB camera standard. Also, since the data rate is limited to the USB 2.0 standard, large image transfer is slow unless it is compressed. Changing the camera will, as I'm sure your aware, require significant changes to FreeLSS the software that drives the scanner.

You might find this YouTube channel interesting: https://www.youtube.com/channel/UCmjN3g … gUWV-mZRpw

To add to it, the amount of time required to complete a scan would be prohibitive.  A 1600 step scan @ 5MP currently takes about 75-80 minutes.  A 12MP scan might take 2.5-3 hours. 

I don't think FreeLSS would require any major changes to work with the data provided by the camera as long as the camera is supported, but the sheer amount of data generated might overwhelm the CPU and available resources. 

I was having a hard time completing 3200 step scans @ 5MP before I upgraded my RaspBerry Pi to v.2, which has double the RAM, but I'm pretty sure a 12MB scan would kill it in short order.

Thank you pirvan, I am probably underestimating the declared 0.25mm, since after all the intermediate data will be interpolated by creating mesh triangles, and there will be no "voxelization".
I planned up front to buy a raspi2, which I already ordered along most of the other needed components; time is not a limitation though, since unlike my 3d printer, I can safely leave this scanner working by itself even while I'm out of home doing other stuff. I plan to build a sort of dark room around it, since for the use I plan I don't care for surface color scanning, just shape.
On a side note regarding the scanning steps: since most of the nema motors have 1.8° angle for each step, a whole 360° revolution would be completed in 200 steps; I was in the market for a 0.9°/step motor just to increase the radial resolution... so what is the meaning of 1600 or even 3200 steps in this case? Am I missing something?

My understanding is that the stepper motor driver is capable of micro stepping the motor which allows it to divide each of the static magnetic steps, 200 in this case, by a factor of 4, 8 or 16. This effectively increases the number of profiles that the scanner captures to 800, 1600 or 3200. Each profile contributes to the calculations that are used to generate the point cloud/mesh that the scanner generates so a higher number of steps translates into a finer, higher resolution mesh. Of course this is within the limits of the camera resolution being used, depth of field, quality of focus and in combination FreeLSS's ability to narrow all of these values down to a coordinate point. Hope this helps.

Using a longer focal length lens would allow a more narrow field of view for the scanner using the same camera to object distance. Longer focal length lenses tend to have flatter fields but also less depth of field. The following links are for:

1) A site that provides extensive technical detail about the RPi camera sensor and optics and sells both c-mount and M12 adapters.
http://www.truetex.com/raspberrypi

2) Another site that sells M12 adapters and a range of compatible lenses. Note that the lens characteristics are defined in terms of 1/2" and 1/3" sensors. Adjustments to these values can be made by using information on the above site.
http://www.m12lenses.com/CNC-Machined-R … 020rpm.htm
http://www.m12lenses.com/category-s/64.htm
http://www.m12lenses.com/M12-Fixed-Iris … s-s/44.htm