Designing a new film scanner; need your help

[quejai]

I think Imacon had the right idea - putting the film in a cylindrical holder makes it completely straight in the axial direction.

A couple guys on rff shared with us a way of holding the film with two magnetic sheets, with appropriate holes cut in them.

Thanks for the ideas, interesting stuff.

By the way, I think the Imacon approach has actually been patented to some extent, probably more for marketing gravitas than functional ingenuity - although being straight in one dimension is definitely a plus. Just checked, the patent has conveniently lapsed.

Magnetic holders sounds like a great way to flatten the image, or even tame curled film. I hadn't heard of it before.

 

[Scapevision]

Magnetic holders sounds like a great way to flatten the image, or even tame curled film. I hadn't heard of it before.

that's very strange, as even Lomography made magnetic holders, but none of the Epson nor Canon ones are.

 

[majid]

Magnetic holders sounds like a great way to flatten the image, or even tame curled film. I hadn't heard of it before.

Contax RTS III cameras had a ceramic film plate with a vacuum system to keep film flat.

 

[brbo]

that's very strange, as even Lomography made magnetic holders, but none of the Epson nor Canon ones are.

Why is it strange? Curly film doesn't care, it won't magically become flat in neither (plastic or magnetic holder).

 

[Phil_F_NM]

This is a bit off topic but...

This is a bit off topic but...

You know what's worth $3000 to me? A subscription service, one-time, up front fee for reliable, continuous support for the already existing excellent scanners out there. Case, in point, the whole line of Nikon film scanners. I'd save up for a few months and gladly pay for that insurance policy.
Sorry to kick it off topic a bit but everything in scanners has already been covered in the thread. I'm willing to bet most of us who have scanners just want ours to continue to work, not to buy new ones.

Phil Forrest

 

[Tijmendal]

What's not important to me is 4x5 format and bigger. The current flatbeds out there do a good job of scanning film of that size. It will make the scanner *much* bigger I presume - which in itself isn't that big of a problem - but I also can't help but to think it will somehow compromise the quality of scans for 135/120 film (correct me if I'm wrong here).

What's important for me (in order):
Ease of use/loading (I HATE using flatbeds for that reason; ****ty holders)
Color accuracy
Speed
D-Max
Resolution

 

[DominikDUK]

Flatbed scan are not good enough for LF film, why do People shoot LF amongst other things to get the biggest possible enlargement with the biggest possible Quality. Hp5 in 35 is the same Emulsion as Hp5 in 11x14 so in order to get the best Quality out of it the film has to be scanned at the same resolution as the small film (I want a big print not a print in the same size as from a 35mm neg I want big). Film flatness is just as critical with lf as it is with 35mm film (lf usually doesn't curl but still) if I want to get the best possible result out of it.

Currently the best tools still are Drum Scanners flatbed like the Epson V series are so far away from best possible results that it isn't even funny anymore.

I would still advise the OP to leave LF out (too small a market at that Price) and plan for a future dedicated LF scanner (which is designed for the high end fine art and Museum market and not the prosumer market like the first scanner should be)

 

[quejai]

Hi again!

This is a general update to clarify where the project is up to.

First off, I've just finished my second year of university. finally! So for a few weeks there I had the constant and mind-numbing distraction of exams and group report writing for my various subjects. I found out, once again, that engineering maths exams are rather hard, and that all my other exams are just applied maths.

That meant that I wasn't making much progress on this project for a couple of weeks, however that period has passed; and as of this week I've been able to give it essentially my full attention.

So what progress is there to report?

Firstly, let's talk about the sensor cooling. While procrastinating for exam study, I thought about the behavior of noise. I realised that the kind of noise reduction we want (high ISO noise) is not the kind of noise (long exposure noise) that cooled sensors address. In other words, cooling the sensor will only produce noticable results when exposure times are large. As the scanner utilises short <1/30 sec exposure times, it seems that sensor cooling is unnecessary. This development will result in lower manufacturing costs, and also lower rates of wear on some parts.

However, the cooling element (a 'peltier' device) is currently in the postal system, and although it's probably redundant I'll still do some experiments to conclusively see whether there are any noticable benefits to sensor cooling for our purposes.

http://www.andor.com/learning-academy/te-cooling-of-scmos-importance-of-te-cooling-to-scmos

The other main thing to report is that, due to prioritising my exam revision, I was unable to finalise the total scanner design before the exams. That means that there may be the risk of being delayed by long shipping times.

However, this is not as problematic as it may appear. Firstly, you can be sure that the design wasn't rushed. Which is a good thing. Secondly, I was able to identify the design-independent off-the-shelf parts with long shipping times (CMOS sensor, microprocessor, optical elements) and they were ordered a long time ago. They should all be here within the next week or two, and shortly after that I should be able to squeeze out a sample scan. Although earlier I mentioned that it would be good to have the samples ready by late November, according to the projected shipping times this may slip back into the first weekend of December.

Another point to make is that everything else can be sourced and manufactured locally, with negligibly short shipping times of several business days. That means I can spend the rest of the week, and perhaps into early next week, finalising and reviewing the design before submitting the orders for the prototype's final parts.

But fear not, time spent otherwise inconvenienced by extended shipping times can be spent designing the software. I have some pretty cool plans for this, and I've found some nice sample code for the functionality that we need. I also have some programmer friends who I'm going to consult with regarding the ideal software architecture.

I'm not sure if I've mentioned this yet, but earlier I was thinking of publishing the kickstarter within the first week of December. I don't think that's probable, instead that's probably the time I'd start preparing the kickstarter. A fair bit of work is required for that, such as: videos, warranty policy development, contingency planning, pricing, etc. I've got a few friends who are keen to get involved with this, regarding legal / graphic design / video, so there would be many hands on deck to help out at that point. I'm well aware that tech hardware has a tragic track record on kickstarter, so I'm also in the process of analysing the previous failures and successes in an effort to ensure our success.




By the way, a 20x microscope lens should be here by this time next week! That means we'll have a 5:1 magnification ratio, quite a step up from the 1:1 of the current setup. I'll rescan that little crop from the Mamiya 6 photo I uploaded a month or two ago, and upload the results here to show you what the maximum resolution could be like. Honestly, I think that really would be resolution overkill in actual usage, but we'll see.
(Just to clarify, I can do some image tests at the moment, but they more represent tests of the optical elements, and do not yet use the final illumination, color, sensor or hardware mounting available in the actual scanner. Further, just like a microscope, the actual lenses used can be swapped for varying magnifications)

Also, the CMOS sensor (and it's backup) has arrived! I'm keen to test this out, but I'll need to wait for the high speed microprocessor to arrive before I can use it.

 

[quejai]

It really is great to have so many of you contributing ideas and suggestions, so thanks for that. I've read and considered each one.

I'm trying to work out the ideal dimensions for the 4x5 dry film holder at the moment. I don't actually have any 4x5 film to measure, so I'm trying to compare measurements from a variety of sources. If you have any laying around, would you be able to measure its dimensions (and thickness, if possible) and post them here, preferably to as much accuracy as you have?
Also, how far in from the edge of the substrate does the image begin?

As has been mentioned several times, it might be a good idea to leave out 8x10 functionality.
This is not an unjustified approach, and my thoughts at the moment are to make two versions; one with standard sizing, and one slightly larger version with a larger imaging area that can handle 8x10 for those who really want it. They would be quite similar, but the larger would use slightly different materials as necessary.

Thanks!

 

[oftheherd]

I don't recall seeing this before. Fascinating idea. I wish you luck with it and will be keeping tabs.

 

[Jockos]

Also, how far in from the edge of the substrate does the image begin?

0mm, scan the whole sheet 🙂

 

[Larry Cloetta]

For the OP; Having read this entire thread, but without going back to find the specific post, I think you mentioned it would not be able to use Silverfast. One of my personal priorities would be that you would reconsider that, or come up with a solution which would equal the results one can get with the Silverfast HDR scanning. And please understand well exactly what kind of benefits one can get through understanding how to use that software before assuming that results "just as good" can be obtained some other way. I'd personally never go back to accepting the limitations of DSLR scanning. Personally.
Everybody wants inexpensive perfection. Sub $1,000 seems wonderful, and who would not want that, but seems unrealistic. My Nikon 9000 will break someday, would love to have something of equal quality available to replace that, which could also do 4x5. I'd love it to cost less than $2000, but can't reasonably expect that. My opinion only.

 

[Asprine]

Here another potential buyer.
Most important for me, find a solution for the dust. I clean regularly but my v700 is just a dustmagnet after the first scan. Post processing is time consuming. Fix this and I'm onboard.
Resolution, dmap, speed, and something to keep the film flat are ofcourse high up the list.
Reliability is very important, hardware wise since In the end it is allot of money and we cant expect life long support.
Reliability in output. In the end I gues even more important then speed. If You need to tinker with settings and scan 2 or 3 times to get it right, You will loose even more time, scanning and post processing.
Rol/sheet feeder would be an nice extra for me.

 

[quejai]

Hey again! Here's another update on the development of the prototype since my last post:

Overall: Things are going to plan - but slower than expected. We're about 2 weeks behind where I wanted to be, which means that there aren't any sample scans to release this weekend, and starting the kickstarter is similarly delayed. This all sucks, but it could be worse.

I know we really need photos of the actual scanner itself, but I haven't got quite enough parts to be able to assemble (and thus take photos of) it yet. Stay tuned, though...

The final order for electronics (mainly for the strobing kohler backlight system) has been ordered, and arrived last week. This includes some really fast switches, which will enable sub-millisecond strobe periods. This is good!

[MOSFET switches above, R/G/B/IR LED's below]

The lens elements for the backlight system arrived. They're also a different shape than I was expecting (convex-convex instead of convex-planar). This has resulted in a slightly more complex design, but it's not much of a problem. Just for kicks, and because the price was barely more, we're using two aspheric elements here. To clarify, we're using conventional microscope optics between the film and the sensor, and a custom optical system built from these two lens elements between the LED light source and the film.

[The two aspheric lens elements for the light source system]

The 20x lens has arrived! Currently, I have the microscope lenses hooked up to a nikon J1 for some testing, but the depth of field is so narrow that I really can't get any useful images from this lens at the moment. Once the scanner is assembled, it'l be a different story.

Parts for the prototype's housing have been obtianed, this is a lower priority but a good deal presented itself and I couldn't resist the price for these parts.

One of two lots of hardware has been ordered, and too has arrived. This batch is the small-tolerance industrial stuff that ensures accurate alignment between moving parts, and also provides the mechanical 'skeleton' for the scanner. It's lighter than I thought, which is a plus, and it's really nice to work with. Beams, motors, bearings, bolts, etc.

[assorted bits]

The second of the two lots of hardware is the problematic, delay-causing one. It's the most complex order, and requires about fourty parts to be very carefully defined prior to the order. Long story short, it's frustrating and unsatisfying - every design decision leads to redoing another, and then optimising another, and so on. Another 20 hours of work on this part and it should be completed, but I've been too busy with other things this week to really knuckle down on it.
A positive thing is that this will be the final parts order; everything else is just assembly, testing and programming.

Earlier on, I mentioned cine-film holders. I've decided that, at the moment, this would have a very small user base; so for the moment I'm not going to work on these holders. If there is enough interest in the future, then cine holders could definitely be released later on - but the design difficulty just isn't justified at the moment.

Current rough schedule plan:
10/11th - submit order for the above mechanical parts
15/16th - mechanical parts arrive
18th onward - sample scans!

Once we get a working prototype, instead of starting the kickstarter soon after, instead I think it would be a better idea to spend a couple of weeks ruthlessly testing and noting down design improvement ideas, so that the model you'll actually have the chance to buy is as good/repairable/reliable/etc as possible. This is also the point where I'll try to get a team together, as there's only so much a one-person-team can acomplish. I know about 20 people who are keen to get involved, but I'll try to whittle that down to a more reasonable 4 or 5 main people. So around mid-late January, we'll probably give kickstarter a go, and see where that takes us.


By the way, check this out!
http://www.cmosis.com/products/product_detail/chr70m
It's a 35mm-sized 70 megapixel sensor. Now it sounds pretty ridiculous, but regardless I'm pretty keen to see what new cameras come out using this.

 

[Asprine]

Don't rush to bring out samples we will follow your progress anyway.
Any thoughts on how to tackle dust issues?

 

[Tijmendal]

Don't rush to bring out samples we will follow your progress anyway.
Any thoughts on how to tackle dust issues?

Agreed. Don't rush anything.

I'm like the anti-engineer so I have no clue what you're talking about, but I've got high hopes! I'm really curious what's going to happen and if this is going to revolutionize scanning - I sure hope so!
Can't wait for some samples!

 

[thereabouts]

I only really use black and white film. I see black and white film photography as an art format in its own right. For colour I use digital, as I don't see any advantage in shooting colour film - for my work. Perhaps if I used medium, or large, format I might think differently.

So, that being my motivation:

Numbers 1-5 are dealbreakers for me.

1. Dynamic range
2. Batch previewing/scanning (whole roll and/or strip of 6)
3. High enough optical resolution for A4 print at 300 DPI - A3 even better.
4. Software just needs to capture flat maximum dmax as TIFF files - everything else I do in Photoshop and no scanner software is going to be better at this. It's crucial to me that this is the easy to use default. For example, I hate SilverFast because it is so hard to disable all the auto 'enhancements' - which are awful. And its horrible icon-based interface is a dog's breakfast.
5. Reliability - obviously. Guarantee and service options need to be easy and reliable also.
6. Flatness of film and ability to scan right to the border.
7. Size – the Plustek 35mm scanners are great, because they are so compact. This is not a deal-breaker for me though.
8. Below $800 for 35mm. If medium format, I suppose you would be going over $1,200 and upwards. But that's too much for me, at the moment. Not sure what the market would support. But if it gets too expensive, I'd be looking at the established brands.

Good luck with it.

 

[PICHA]

Nikon coolscan V ed was perfect! 🙂

 

[quejai]

Asprine, regarding dust:
For color film, there is an infrared LED to enable a digital-ice sort of thing, however ideally there would be a method to physically remove the dust from the film. There isn't a glass base for dust to collect on, like a flatbed, so the vast majority of dust to be concerned about is that on the film itself. Currently, I think the overall best way to go is to pull your film between folds in a microfiber cloth before loading the scanner.

However, your comment inspired me to do a bit more researching and I found this goldmine, NASA's plans to get rid of lunar and martian surface sand/dust from solar panels on future interplanetary landings:
http://www.electrostatics.org/images/esa_2008_o1.pdf
(This would also be really useful when asteroid mining takes off, perhaps as a way to move dust-like ore around. Or as a way to move about *on* dust-like ore, kind of like tank treads.)

Anyway, that technique would be absolutely perfect. The catch is that you need about 4,000 volts to operate a dust system like this, and I'm just not sure if the benefits are worth the extra safety issues this brings up. However, I'm confident that if safety issues were adequately addressed, the current scanner design would definitely support drop-in dust collection pads based on this paper. I hope it's not patented, I haven't checked yet.

To summarise; this method is overkill for the standard scanner kit, but like the cine film holders, if the kickstarter is successful, accessories based on this are likely to become available.

Tijmendal:
Same here - and thanks!

thereabouts:
You raise some good points. I had kind of accepted that it would be impossible to scan right to the border, but I've thought about it again and I've realised that it is possible to design filmholders that do scan right to the edge, at a consequence of being 20% slower or so if scanning large batches of film. I'll make both types available.

Regarding your points about price, I'm very confident at this point that I won't be able to sell the scanners for sub-$1000. You know what? I couldn't even afford one of these scanners. I picked up a new job to cover the prototyping costs, but I can't expect other people to do the same. I have had an idea that might make it more affordable: Buying one between two people. If it sells for $1300, then this would bring it right down to the more palatable number of $650.

Any thoughts on this? You'd need a good, trustworthy friend to do it with; but it might actually work for some people. Even if you shoot everyday, you probably still only develop once a week at most. If you scan at that rate too, then there's easily enough time to let someone you trust use it. If you both are later able to afford the other half of its cost, then you each end up with the whole thing to yourself in what are essentially 'interest-free instalments', or whatever the correct financial jargon is. Just a thought.

PICHA:
Thanks! My design should be a bit faster than the Nikon, that's the plan. We'll find out soon if it is.

Thanks all! We're getting there.

 

[michaelwj]

I have been following this thread for a while. The project looks very exciting but I hope you don't end up cutting corners. Most people already have a scanner that works, but is somehow frustrating. If yours isn't perfect then it doesn't matter how cheap it is, we all have scanners that aren't perfect already.
My 2c. If $1300 is good and $1600 is perfect, then it costs $1600, or $2000 or whatever it is. I don't shoot a film rangefinder because it's cheap.
Good luck!