Designing a new film scanner; need your help

[quejai]

Caleb and I are currently putting a fair bit of work on the motors, and how to make them fast and quiet, as recent experiments have shown the current speed and sound as undesirable. A new job has slowed down my progress a bit, but of course it is still being worked on - it's also funding some more items, such as a new microscope lens that could increase the maximum scanning speed by 2.5x.

hello quejai,

i joined rff a few weeks ago to help with an issue i've been facing for a number of years - scanning the 35mm photos (i use my family's leica m4) i've taken from my times as flying (i work in brooklyn for a private airline company and regularly bring my leica up with me). i've always been unhappy with the results of scanning my images a few weeks ago i realised a number of images have started to really fall in quality. so i came to the rff (a passenger told me about the rff - he had a widelux!) to find a solution- and alas!!!

i'm really excited with this project but i have one concern & its how long will it take you to build these? when can i purchase the kits & will they be offered first??? on what jockos said - wouldn't mind also being on the beta team!!!

😛

Hola! Well hopefully this will be the machine you're looking for. Regarding manufacturing rates, we're thinking that we could each manufacture at least one per week, perhaps even up to 3 or 4 per week. Any faster than that would be too expensive to pull off. When they become available to buy, to avoid the who-gets-theirs-first chaos I'm imagining that we'd specify a time period in which people can buy a scanner from the first beta batch (a couple of weeks?), and then after that time period closes, we'll randomly work the order out, and open up orders for the next non-beta batch.

hi quejai,
I'm very interested in your project, I hope there will be the chance to buy a kit in Italy too.

What Luuca said!! 🙂

Piano adagio!

The last couple of posts by Italian forumers remind me of the Ferrania approach and how this project matches it for its focus on "future" and sustainability; Modularity/adaptability and smaller scale with a focus on quality. It's great that you took an idea and optimised it for film scanning.

We're excited to hear more of course. Still, quejai, don't feel pressured.

Of course, we'll be happy to ship some to Italy. Postage might be a bit steep though...
Prest 400: The main motivation for working on the current scanner is how much other ones sucked, and it just so happens that modularity / the ability to be updated over time was a side effect of it. Hopefully it does work out, thanks.

Love all of the positivity of this thread and the DIY spirit.

Quejai, excellent work! I just wanted to mention, if you haven't heard of Hackaday, it's a great site that I think you'd be interested in. Lots of folks hacking electronics and building projects like these on it.

Anyway, they have a great write up about structural framing systems (like the kind you are using).

http://hackaday.com/2016/12/14/a-how-to-in-homebrew-design-fab-and-assembly-with-extruded-profiles/

Oh, and this could be of interest as well:
http://hackaday.com/2016/12/15/high-quality-film-transfers-with-this-raspberry-pi-frame-grabber/

They are scanning images at different exposures and combining them using OpenCV.

Keep up the great work!


Sent from my iPhone using Tapatalk

Super cool links, I know the site but those articles are new to me. I'm going to try the first link's suggestions! We were considering using a rapsberry pi camera a long time ago, but for some reason (can't remember which now) we decided to move onto something else. I have used the pi and pi camera before, and their approach is interesting in its elegance - just sub in the pi in place of the film.

Scanning a film roll like telecine machines or Imacon is best approach. Use ClingZ foil to attract dust with electrostatic force or ultrasonic rolls. When I was developing 16mm film in WFDiF Warsaw it came out very clean after ultrasonic cleaner. Projectors tend to scratch delicate Kodak Vision negatives.

That's pretty much what we're doing already - and this ClingZ foil sounds awesome! do you have any links for its use in telecine machines?

 

[jzagaja]

Got ClingZ samples will chcek usability. I'm planing use it for electrostatic speaker.

 

[jzagaja]

Similar comercial scanner?

http://www.hamamatsu.com/jp/en/C12000.html

 

[quejai]

Similar comercial scanner?

http://www.hamamatsu.com/jp/en/C12000.html

Interesting, it is similar. I was actually reading about an identical machine from another company yesterday. Seems overly complicated though.

So the 10x objective arrived, I did some tests, and it looks promising for that 2.5x speed boost by lowering the lower end of the resolution range. So with two 4x and one 10x and 20x objectives, we can modify the native resolution of the current sensor to between 2,600 - 33,000 ppi by changing which two objectives are in use.

Also did some proof-of-concept experiments for Caleb's idea for the gripper for a film loader that would allow pakon-like operation, and the idea is sound. There isnt too much more work to do before the dry loading holder should be ready.

I also played with the super bright backlight LEDs we have, and worked out a much better configuration if different LEDs are used. I'll send an order through for some tomorow. This will allow a more modular construction, ease of swapping out the LED module for, say, LEDs optimised for monochrome, and more even illumination.

So... getting there.

 

[Evanjoe610]

Sorry, I'm a late comer to this Scanner project. Read the entire thread and find it quite promising.

I would like to join in the final production when it comes to fruition.

I'm a former drum scanner and X-Y flatbed prepress scanner user.

At home I have an X-Y flatbed prepress scanner, Scitex Eversmart Pro II and the Imacon Flextight Precision III.

Both has performed excellent work for me based on the tight space I have allocated it to . The weight of the Scitex Eversmart Pro II is HEAVY...

Keep up the good job and progress.
An open source product that IS not proprietary is best for this community.


Evan

 

[rfaspen]

I'm excited about this!!!

I like the idea of a 3-tiered offering and pricing:
1. Fully assembled scanner $$$
2. Partially assembled scanner $$
3. Box of parts $

 

[CliveC]

To be honest, I didn't initially think that this project would succeed. I am glad to be proven wrong! Any chance we can get a short video seeing this thing in action?

 

[Archlich]

I'm ready to spend $$$$ on this. Existing medium format scanners are that expensive anyway!

 

[struene]

I follow the thread from the first starting and i am very excited. Just dont forget the most important thing:

there are plenty of more or less nice but useable scanners out there, you address most of their shortcomings. Just dont forget the strongest: its the software! A reliable and constant workflow in my view can only be achieved with the approach of exact densitometric calibration of the scanner and according paperlike gradation-curves. With a change of the light emmision after prescan, not by moving around endpoints of RGB-curves... Thats, as i understand the way noritsu/pakon/Fujitsu goes and they shine in that. Of couse the hardest thing stays still the white ballance for colour negative....

thanks for your efforts!!
schöne Grüße,
Johann

 

[fjolnir]

If open-source I would love to help write a driver/scanning application for this (programming is my day job).

 

[Tijmendal]

I honestly wouldn't want anything from the software: just something that lets me put out non-DNG RAW files would be perfect.

 

[Jockos]

I honestly wouldn't want anything from the software: just something that lets me put out non-DNG RAW files would be perfect.

Same here, except I'd like a invert colors function for negatives, because that feature is missing from Lightroom, and flipping the histogram messes up all other sliders..

 

[quejai]

fjolnir - super appreciated, thanks. Will try to get the first version working by ourselves but we'll know who to ask when we get stuck. Our style is to learn what we do as we go, so I'd say that you'll find a fair bit to improve on when we publish our code.

update:
We're working on finding a supplier for anodised aluminium panels for people who would prefer those over acrylic.
We've also ordered a whole bunch more parts (motors, motor drivers, leds, filmholder parts, ...) that should be arriving soon.
Present plan for organising initial sales is through a kickstarter-like service (We did rule it out earlier, lel 🙄 ), with options such as beta-testing, just parts, partial kits, fully assembled, and other options like aluminium vs acrylic panels, filmholders, etc. We will start work on the kickstarter and our website (designdata+forum) once the prototype is fully functional, as we don't need to spend time fussing over those quite yet.

 

[jzagaja]

Quejai design (XY microscope lens) is very clever - compare with single shot scanners used in cinematography. To get high resolution an expensive hi-end lens and monochrome ccd must be used:

http://digitalcinemasystems.net/?page_id=50

if only a laser beam (flying spot) and photomultiplier could be used as in Rank Taylor and Cintel scanners - last scanner that used this technology was Cintel Millenium II. Look for samples they are very analog looking.

 

[jzagaja]

An idea for making light source - use rgbw led and holographic diffuser instead of integrating sphere. It will hide emulsion scratches.

 

[quejai]

if only a laser beam (flying spot) and photomultiplier could be used as in Rank Taylor and Cintel scanners - last scanner that used this technology was Cintel Millenium II. Look for samples they are very analog looking.

Interesting, now that I think about it, to implement a flying spot approach would be very possible, just sub in a new light source/sensor. This is a great excuse to mess with PMT's, but right now the complexity : performance is such that I won't spend time on the design until the rest of stuff is done.

An idea for making light source - use rgbw led and holographic diffuser instead of integrating sphere. It will hide emulsion scratches.

Hadn't heard of holographic diffusers before, but I was reading up on diffraction diffusers, they are very nice as they allow the use of super bright and efficient lasers as opposed to LEDs. Super expensive though, so likewise, not something I'm presently pursuing. This would also be a sub-in light source replacement.


Re. the scanner itself: That stuff is starting to arrive. The new motors and drivers are very, very nice - these are black NEMA 14 1.5a models, and the driver is the DRV8880 which has a lot of good things going for it. Testing has verified that this motor and driver combo is faster, smaller, higher torque and quieter than the previous setup. This will allow a more optimal arrangement of motors and rollers.

 

[Jockos]

Re. the scanner itself: That stuff is starting to arrive. The new motors and drivers are very, very nice - these are black NEMA 14 1.5a models, and the driver is the DRV8880 which has a lot of good things going for it. Testing has verified that this motor and driver combo is faster, smaller, higher torque and quieter than the previous setup. This will allow a more optimal arrangement of motors and rollers.

What's the current price estimate with these new parts? 🙂

 

[jzagaja]

Consider an option for use USB3 astronomy CCD cameras that deliver true 16bits, as you use stitching it can be a small cheap sensor.

 

[quejai]

Consider an option for use USB3 astronomy CCD cameras that deliver true 16bits, as you use stitching it can be a small cheap sensor.

Would be good, but you'd need to do a bit of DIY to make it fit. i.e. remove the case if too large, build a mounting panel for it, etc. Definitely doable though.

What's the current price estimate with these new parts? 🙂

Well these parts are the new normal, so no different price for them. Basically horizontal space is more expensive than vertical space for the motors, so these motors are thinner and taller.

Regarding cost, the current idea is that the whole scanner will be between 1K and 2K USD. Once the beta testing phase is done, we plan to build them in batches of 10 for as long as there is interest. We're also thinking about doing test scans for prospective buyers so you can play with your own raw data yourself before committing to anything. If you're seriously considering getting one, you will need to start saving.

Redesigned and rebuilt the lower carriage level to fit the new motors last week. As usual I couldn't find the right bolts, so there are some (temporary) silver-looking spacers below the motors. The image sensor has been taken out to more easily link up to the computer for programming.

To redo the upper carriage level is the goal for this coming week. It requires about half the effort of this carriage level.

In the pic, the lower motor is the one used for focus control. This motor is much easier to spin by hand (no gears, unlike the last one) which means you can either do manual focus in software or by hand.

That lump of silver is the 4x objective without its case, which just screws right off. All the objectives I have basically halve their weight when you remove the case! Also, 35mm canisters are great caps for these objectives.

 

[quejai]

To clarify:

Focus works by spinning the focus motor. That spins a belt that moves around the whole perimeter of the carriage level; from the motor, along the sides, to the far pulleys you can see in the top photo.
As this belt moves, it spins other pulleys, which have been modified to contain nuts. As these nuts spin, they pull a bolt up or down, and that bolt is fixed to the holder holder, which moves up or down, focusing the image. Sort of like a DIY micrometer.

Compared to the last photos we put up, we've changed the design of these focus rollers too. They're much less prone to jamming now, and have much less hysteresis.

Super hard to explain, super easy to comprehend when we put a video of it up.