So I met up with calabachie (another rff member) (not sure on the spelling) a couple of times over the past few weeks and we've made a bit of progress. Some brackets for the chassis have been redesigned, and we rebuilt the frame based on this new approach. There's still more to do in this regard, so the rebuild is not yet final. We've shifted our focus to the design of the filmholder system, so that should be featured in the next update. We do have a super-sketchy temporary design, but that will really only be useful for giving early sample scans if everything else suddenly starts working.
Further, the 'carriages' (the things that slide up and down the rails) have been built. The illuminator one is basically a simulator of a kohler light source (with ~90% of the functionality for ~2% of the design effort), and will be replaced in time with the proper design. The camera carriage is a sketchy preliminary version of the final design, and features the back-to-back lenses and an image sensor panel. And 14 bolts.
Unfortunately the wheels and bearings that these carriages use have slightly more friction / resistance to motion than I was hoping. I'll think about how this could be resolved, if it turns out to be an issue. Further the carriages are a bit bigger than I thought they were going to be, so I'm going to lengthen the scanner frame by about 2 inches to compensate.
By the way, I've swapped the positions of the carriages: Originally, it was the camera on top, illuminator on the bottom. But we can get a smaller scanner and a lower center of mass if we put the camera on the bottom and the light source on the top, so that's what the plan is as of now.
Software is getting there. I've found a whole bunch of highly relevant sample code, relating to controlling the image sensor. And it's nuts. Assembly programming is fun.
Just a general point on some previous posts in this thread: This scanner essentially *is* the DSLR scanning approach, optimised. It's also got a film transport mechanism.
It uses a small image sensor, which means that we can use smaller lenses, if we accept that we will need to take several images of each film photo and 'stitch' them back together. This is easy, because of the film transport mechanism. The small sensor size also means we can jump straight on the bandwagon of microscope lenses, and hence outsource much of the optical design.
The very controlled color of the backlight of this scanner allows us to use either the bayer grid in the sensor, or bypass it alltogether and make several passes for each colour channel, and hence bypass a common complaint of DSLR scanning, which is that the colors suck.
The main disadvantage of this approach compared to using a dslr is that, as has been pointed out on other threads, the dslr approach allows you to use that camera for shooting too.
Anyway I do have uni exams coming up, so there's a good chance that this thread will be fairly quiet for the next month. On the other hand, the only thing that's stopping some sample scans is that the software is not quite working yet, and - due to exams - I'll have lots of time to procrastinate by working on this. So yet again, I'm not sure when the sample scans will be ready.
The photos below look terrible. They were clearly shot on a potato, and the little camera module has far too many bolts on it to ever be photogenic. The lower panel on that module is the 'digital back', and yes we ran out of nuts today, so it's not quite as complete as it will be soon. We also haven't reinstalled the stepper motors or belts yet, but they will go back in much like in the previous build.
