When you get down to it - a scanner is "just" a really specific digital camera with a film holder. The problem is that quality is more about setting up the entire system consistently (film aligned on the scanning camera focal plane correctly, can shift from frame to frame, etc). As with all photographic setups, the chain is only as good as its weakest link.
It is probably possible for any single person to build, through sufficient time and effort, a "better" scanner than any model manufactured - but once you are spending that much time and effort on it you've "lost the plot" of the whole effort.
I guess what I would naively expect to exist here, is some kind of "kit" for turning a DSLR into a 2D flatbed raster scanner, in the vein of the Pi-based emulation portable "kits" — i.e. "here's a box with all the mechanical and structural stuff, flat-packed with foolproof assembly mechanisms that guarantee rigidity; just bring-your-own expensive electronic bit to slot into it, and watch it go." Where in this case the "expensive electronic bit" is the DSLR — probably with a Wiki-maintained list of known-compatible camera bodies + lenses.
If such a kit existed, the "sufficient time and effort" would amount to maybe five hours and $500. It'd basically be the same hardware + electronics as a cheap 3D printer — just without the need for one of the axes. (Mind you, that's presuming all the imaging is flowing directly from the camera to software on your computer, without any electronics in the device MITMing the signal to do computational photography stuff. If it needed to stitch the photos itself, it'd be a lot more expensive.)
Enough people seem to both have this exact problem, and have the ingenuity to be the one person who produces the kit to scratch their own itch but also the itch of everyone else in the community from then-on, that I would be somewhat surprised if at least an attempt at such a kit didn't already exist.
I mean, I agree that such a kit would be hugely popular and useful. You just don't understand how hard the problem is. There's a reason that people pay thousands of dollars for twenty year old hardware.
> I would be somewhat surprised if at least an attempt at such a kit didn't already exist.
Do you expect that an existing mid-tier consumer 3D printer company could make such a kit (with sufficient quality to be useful) as a byproduct of their 3D-printer production line — the same way that iRobot makes "robotics platforms" that are just Roomba vacuums, minus the vacuum mechanism, plus an SDK?
If so, would it be a good idea to go bug one of the 3D printer companies — most likely, one of the ones that frequently pumps out new models to address the problems customers complained about in previous models — to try doing this?
The problem is that you have a... thick 160μm plane (the 'target') that you need to orient exactly along another plane some distance away. You are going to need to mount things on the target and move them through with good speed and precision. Every link in the chain has the potential to throw off the alignment. Traditionally this is solved through a centralized manufacturing process where one group of people get really good at building (and maintaining) the devices such that they generally line things up reliably (i.e. a company making a scanner).
It's kind of like P=NP for manufacturing I guess? If you can really figure out how to instruct a random person to get an above-average scan out of their DSLR, you can instruct them to make almost anything.
Friend built a light box with some color calibration markers and DSLRed 20,000 slides in a few days and post processed the ones he liked. It's not professional grade but it worked fine.
When you get down to it - a scanner is "just" a really specific digital camera with a film holder. The problem is that quality is more about setting up the entire system consistently (film aligned on the scanning camera focal plane correctly, can shift from frame to frame, etc). As with all photographic setups, the chain is only as good as its weakest link.
It is probably possible for any single person to build, through sufficient time and effort, a "better" scanner than any model manufactured - but once you are spending that much time and effort on it you've "lost the plot" of the whole effort.