Bandiahegyrol
Member
Hi guys,
Any news there?
Any news there?
Designing a new film scanner; need your help
- Thread starter quejai
- Start date
- Latest activity Latest activity:
- Replies 338
- Views 15K
Jockos
Well-known
Any updates? 🙂
quejai
Established
Hey all,
Apologies for the delay in this response. We've got a plan to avoid delays like this in the future, will get to that in a moment.
We've made significant progress with the scanner. We're currently investigating in incorporating some emerging technologies as well as experimenting with a range of different overall scanner designs. What we've recently contemplated is the discrepancy in our designs on paper and our designs that we've built, and it's primarily due to our shoestring budget. We have had to be so careful with what we spend money on, which is limiting how much we can get done and the speed of progress. We recognise that if we could buy the necessary things quickly and without rapidly deplenishing our budget we'd be able to make progress more rapidly and be less sensitive to productivity losses if we break stuff. Let's just say that with some better 'financial lubrication', we'd be able to get through this prototyping phase much more quickly, moving toward ultimately giving you the opportunity to buy one sooner than otherwise. We'd also feel more justified in providing more regular updates to this thread if our progress were at a higher rate.
The sort of things we want:
The mechanism to acquire funds that we'd like to try out first is donations. (I do hope that disussing this isn't against RFF rules) To do this, you can either join in to our Patreon, or bypass that and send a donation to our Paypal account directly. These are linked in our signature.
We realise that this is a big ask, but we're pretty sure that everyone would rather we succeed, and succeed quickly. And if you can help us have more-than-a-shoestring budget, all the better. The past few years have proved that we are dedicated to this project. Hopefully with this funding, we'll be able to post regular updates in the future with photos of our progress. As always, feel free to comment below to discuss anything on your mind that relates to this project.
Technical information:
Quejai and Caleb
Apologies for the delay in this response. We've got a plan to avoid delays like this in the future, will get to that in a moment.
We've made significant progress with the scanner. We're currently investigating in incorporating some emerging technologies as well as experimenting with a range of different overall scanner designs. What we've recently contemplated is the discrepancy in our designs on paper and our designs that we've built, and it's primarily due to our shoestring budget. We have had to be so careful with what we spend money on, which is limiting how much we can get done and the speed of progress. We recognise that if we could buy the necessary things quickly and without rapidly deplenishing our budget we'd be able to make progress more rapidly and be less sensitive to productivity losses if we break stuff. Let's just say that with some better 'financial lubrication', we'd be able to get through this prototyping phase much more quickly, moving toward ultimately giving you the opportunity to buy one sooner than otherwise. We'd also feel more justified in providing more regular updates to this thread if our progress were at a higher rate.
The sort of things we want:
- image sensor modules
- development kits
- test equipment
- general prototype fabrication
- mechanical parts
- optical components
- etc etc
The mechanism to acquire funds that we'd like to try out first is donations. (I do hope that disussing this isn't against RFF rules) To do this, you can either join in to our Patreon, or bypass that and send a donation to our Paypal account directly. These are linked in our signature.
We realise that this is a big ask, but we're pretty sure that everyone would rather we succeed, and succeed quickly. And if you can help us have more-than-a-shoestring budget, all the better. The past few years have proved that we are dedicated to this project. Hopefully with this funding, we'll be able to post regular updates in the future with photos of our progress. As always, feel free to comment below to discuss anything on your mind that relates to this project.
Technical information:
- Converted to new optical approach using single microscope objective. This allows for a simpler optical path and mechanical design, yet allows much more flexibility and resolution potential.
- Performed some preliminary testing of the new optical setup, currently resolving over 8000DPI with mid-range objectives. The resolving limits of this new approach are yet to be quantified.
- Investigating new solid state focusing system capable of sub-ms response times. Will simplify muechanical design/durability/precision, as well as allowing advanced masking and compensation features. An entirely novel backup mechanical focus mechanism is also in an advanced stage of design.
- Working on two holder designs, one with an internal 120/135 roll feeder, and a universal holder for plate, batch, strip and multi-format usage with integrated dust removal. Rough prototypes being built.
- Researched colour conversion approaches and worked with some local focumers. Looking at hardware based approach with specialty mv sensor.
Quejai and Caleb
Jockos
Well-known
To be honest, I don't think it's a good thing that you're back to prototyping your idea again after three years. While your original design may not have been optimal, maybe it'd have been better to go for a MK II in the future.
The phrase "investigating in incorporating some emerging technologies" also makes it sound a lot like vaporware. I'd rather see that you make a product, and then start experimenting. The image quality from previous samples seemed really good as is, and that was in 2015!
If the scanner is modular like you said it'd be earlier, a lot of improvements should be able to be added later on as well.
Don't get me wrong; I've really been hoping for this scanner for a long time now. But if it's going to be three more years, I'm going to look for other options.
The phrase "investigating in incorporating some emerging technologies" also makes it sound a lot like vaporware. I'd rather see that you make a product, and then start experimenting. The image quality from previous samples seemed really good as is, and that was in 2015!
If the scanner is modular like you said it'd be earlier, a lot of improvements should be able to be added later on as well.
Don't get me wrong; I've really been hoping for this scanner for a long time now. But if it's going to be three more years, I'm going to look for other options.
quejai
Established
Thanks for the constructive criticism. We’ll get back to you over the next day or so.
quejai
Established
Yes we were able to get some test images back in 2015 or so, however that was really representative of the sort of results to expect if we ended up successfully designing the rest of the scanner infrastructure/hardware to support it; and although we've gone through a range of ideas and prototypes in an effort to find a design that is even partially acceptable, we are really yet to find one. However, as always, we're close to significant progress for a very satisfying design. But - if anyone would like the design files of our earlier versions, send us a message and we'll help you build something like where we got to before realizing it was a dead end. There are just so many metrics of acceptable design - rigidity, simplicity, manufacturability, part-interdependence and alignment requirements, ruggedness, whether each of the innumerable sub-modules actually works, the design inflexibility behind committing to a given mount for modularity, etc etc.
The reason we build prototypes is to test and get feedback, we have taken this onboard and what we are currently working on is just the next step, which includes a few redesigns but no radical departures from what you know.
Regarding emerging tech, we agree it did sound a lot like vapor-ware when we were writing it. This was us trying not to disclose (and thus becoming sort-of committed to) which new technology we may be experimenting with.
We are also disappointed with the speed of progress but there is only so much we can do as a small team with limited funds and time.
But we don't need to winge, we are legitimately onto what might just be the final real design iteration.
On another note, I (quejai) have recently taken a part time job designing the electronics/computer/interface for obscure research lab equipment. I'm planning to use a very similar approach for the electronics/computer stuff in this scanner as, really, it's quite similar. My starting point at the job was the architecture I developed for this scanner, so to be quite honest, progress on this front is blazing (the arduino and raspberry pi combo in the pictures are just for this prototype/mockup, they are not part of our final computer system). Coincidentally, one of the mechanical design people in that company has great contacts for getting aluminium parts manufactured, which will come in very handy later on when we start replacing the acrylic in our prototypes with aluminium.
I'm also doing a uni thesis on dust/scratch correction methods. For this, it is becoming increasingly important to get some more image data, and hence there's a bit of a rush to get some sort of partially-working scanner that can produce image output yet be legit enough to provide repeatable, controlled results. This was taken as an excuse to take the latest scanner design and build a mockup with the parts we currently have (ie we can't afford all the proper optical elements yet, so things aren't quite as specified in our main designs - but close enough for now). Our latest chassis designs combine the focus and image-carriage level from the last prototype into an extremely compact setup, with so little to go wrong it is vastly easier to build, and really ought to never break or jam - our previous approach was otherwise. Again, this particular build is a prototype, but it is essentially what the latest chassis design is.
A fun fact about this new focus system: each end of the carriage's motion can be focused near-independently. This avoids the scenario where focus is correct at one position along the carriage's motion but not at others. In effect, it may appear as a slight 'tilt' - this image gives an example of an unrealistically extreme scenario to point out the capability. We'll have videos of this adjusting itself soon. This is a very good thing!
Now as we're using the older optical setup, we need a bit of extra clearance below the chassis for some cabling. We also need a stand-in filmholder, as we haven't quite finished designing / can't quite afford the parts for the latest universal filmholder design, we're going for a super simple non-mechanized flat approach. We also need a light source, so we're experimenting with a softbox-inspired diffuser. It's huge, it's also easy to make, so we're taking this as an excuse to try out the idea.
Yes it is a bit of a timing coincidence, but the primary rush for this is so it produces image output asap for a uni thesis. Over the next week or so the wiring up will be completed, and we'll keep this thread posted with progress. Watching the focus system moving is quite revealing of how it works, expect a video when we've got that going electronically.
The next work steps for this scanner is to order the parts for the more compact optical setup, and to finish off designing/ordering parts for the film holder.
The final-product scanner that I've got in mind is essentially what we'll have once those are done, just tidied up and neatened. From the outside, it will look quite a bit different than this prototype though.
The reason we build prototypes is to test and get feedback, we have taken this onboard and what we are currently working on is just the next step, which includes a few redesigns but no radical departures from what you know.
Regarding emerging tech, we agree it did sound a lot like vapor-ware when we were writing it. This was us trying not to disclose (and thus becoming sort-of committed to) which new technology we may be experimenting with.
We are also disappointed with the speed of progress but there is only so much we can do as a small team with limited funds and time.
But we don't need to winge, we are legitimately onto what might just be the final real design iteration.
On another note, I (quejai) have recently taken a part time job designing the electronics/computer/interface for obscure research lab equipment. I'm planning to use a very similar approach for the electronics/computer stuff in this scanner as, really, it's quite similar. My starting point at the job was the architecture I developed for this scanner, so to be quite honest, progress on this front is blazing (the arduino and raspberry pi combo in the pictures are just for this prototype/mockup, they are not part of our final computer system). Coincidentally, one of the mechanical design people in that company has great contacts for getting aluminium parts manufactured, which will come in very handy later on when we start replacing the acrylic in our prototypes with aluminium.
I'm also doing a uni thesis on dust/scratch correction methods. For this, it is becoming increasingly important to get some more image data, and hence there's a bit of a rush to get some sort of partially-working scanner that can produce image output yet be legit enough to provide repeatable, controlled results. This was taken as an excuse to take the latest scanner design and build a mockup with the parts we currently have (ie we can't afford all the proper optical elements yet, so things aren't quite as specified in our main designs - but close enough for now). Our latest chassis designs combine the focus and image-carriage level from the last prototype into an extremely compact setup, with so little to go wrong it is vastly easier to build, and really ought to never break or jam - our previous approach was otherwise. Again, this particular build is a prototype, but it is essentially what the latest chassis design is.
A fun fact about this new focus system: each end of the carriage's motion can be focused near-independently. This avoids the scenario where focus is correct at one position along the carriage's motion but not at others. In effect, it may appear as a slight 'tilt' - this image gives an example of an unrealistically extreme scenario to point out the capability. We'll have videos of this adjusting itself soon. This is a very good thing!
Now as we're using the older optical setup, we need a bit of extra clearance below the chassis for some cabling. We also need a stand-in filmholder, as we haven't quite finished designing / can't quite afford the parts for the latest universal filmholder design, we're going for a super simple non-mechanized flat approach. We also need a light source, so we're experimenting with a softbox-inspired diffuser. It's huge, it's also easy to make, so we're taking this as an excuse to try out the idea.
Yes it is a bit of a timing coincidence, but the primary rush for this is so it produces image output asap for a uni thesis. Over the next week or so the wiring up will be completed, and we'll keep this thread posted with progress. Watching the focus system moving is quite revealing of how it works, expect a video when we've got that going electronically.
The next work steps for this scanner is to order the parts for the more compact optical setup, and to finish off designing/ordering parts for the film holder.
The final-product scanner that I've got in mind is essentially what we'll have once those are done, just tidied up and neatened. From the outside, it will look quite a bit different than this prototype though.
pyeh
Member of good standing
Keep at it, Quejai and Caleb. You're doing good.
jpk
Newbie
Please keep going on, it's such a good project. Maybe we could buy the scanner in advance so you have enough money for R&D...?
As for film colors: I used "colorperfect" for scanned negs, its a very good thing and provides exact colors (I havn't come across anything better), see here: http://www.colorperfect.com/colorperfect.html?lang=en
There are youtube videos about it here: https://www.youtube.com/user/ColorPerfectPlugin/videos
Best whishes!
As for film colors: I used "colorperfect" for scanned negs, its a very good thing and provides exact colors (I havn't come across anything better), see here: http://www.colorperfect.com/colorperfect.html?lang=en
There are youtube videos about it here: https://www.youtube.com/user/ColorPerfectPlugin/videos
Best whishes!
CMur12
Veteran
Quejai and Caleb, I support your process, and I look forward with great interest to the final product. This is a helluva project and I greatly admire your expertise and your determination.
When the scanner is ready, a lot of us will be, too.
- Murray
When the scanner is ready, a lot of us will be, too.
- Murray
ErikT
Newbie
I've been following this thread with great interest for a very long time now.
What's the status for the design of the film holders? To me, keeping the film flat (or at an exact curve for that matter) is one of the bigger challenges. Especially if you—like me—think it's really important to be able to scan the entire film surface (that is, including the film rebate/edges).
I've searched for computational algorithms to "flatten" an image with a known distortion (curve), but I have not been very successful. If the curved solution is the one you're aiming for, have you found an algorithm for this (research papers et cetera)?
I know you've shown demo videos of a film holder design, but I wasn't sure if this is the route you're going for at the moment.
What's the status for the design of the film holders? To me, keeping the film flat (or at an exact curve for that matter) is one of the bigger challenges. Especially if you—like me—think it's really important to be able to scan the entire film surface (that is, including the film rebate/edges).
I've searched for computational algorithms to "flatten" an image with a known distortion (curve), but I have not been very successful. If the curved solution is the one you're aiming for, have you found an algorithm for this (research papers et cetera)?
I know you've shown demo videos of a film holder design, but I wasn't sure if this is the route you're going for at the moment.
KM-25
Well-known
I sold my Nikon 9000ED film scanner and all of it's accessories a couple months ago, not because I did not want it or use it anymore, but because of what I discovered in using my D850 for scanning film.
With a budget of $1,500 and with some minor modifications, I now use a Sinar P2 4x5 monorail large format camera with precise geared movements for shooting 4x5 film, DSLR scanning and mural enlarging. In scan mode I put on a utility standard that has a LED light panel mounted to it and in enlarger mode replace that with one that has a Heiland VC LED light source. And to mount the D850 I simply swap frames on the rear standard from 4x5 to one that has a Arca style mount plate for the D850...this literally takes seconds to do.
I use a modified negative carrier from my Omega / LPL 4550 enlargers and put the perfectly flat negative carrier into a modified frame on the other standard and again, in mere seconds. This setup can scan up to 4x5 as a single frame or come in at 1:1 and stitch the negative at what amounts to over 8000 dpi.
The files are then brought into Lightroom CC that stitches the NEF files automatically into one large and highly flexible DNG file. The scan quality is a significant gain over even a drum scan let alone the 9000ED.
And the best part, unlike a scanner, this DSLR scanner setup is not a one trick pony but master of several. Obviously the D850 is a scanner component and a camera used for taking photos but the Sinar P2 is that AND a mural enlarger! The geared movements make aligning it for each custom job a joy to use and the output tightly controlled. Using live view and the electronic shutter on the D850 make the final output top notch.
The modularity of it all and the fact that the "sensor" is so easy to upgrade makes a dedicated film scanner completely obsolete for me. I have no doubt in my mind that DSLR scanning is the way digitizing film is going and as more software and even in camera inversion algorithms appear this will spell that out pretty clearly.
I applaud the OP for his efforts but I think he is too late, I think by the time this dedicated scanner is ready to sell, several brilliant new ways of using DSLR's to scan *and* export a right viewed raw file for any use imaginable will have hit the market and all but decimate the market for such a device.
I don't say this as a critic of the efforts the OP made but more of a heads up that his team may want to course correct before it is too late.
With a budget of $1,500 and with some minor modifications, I now use a Sinar P2 4x5 monorail large format camera with precise geared movements for shooting 4x5 film, DSLR scanning and mural enlarging. In scan mode I put on a utility standard that has a LED light panel mounted to it and in enlarger mode replace that with one that has a Heiland VC LED light source. And to mount the D850 I simply swap frames on the rear standard from 4x5 to one that has a Arca style mount plate for the D850...this literally takes seconds to do.
I use a modified negative carrier from my Omega / LPL 4550 enlargers and put the perfectly flat negative carrier into a modified frame on the other standard and again, in mere seconds. This setup can scan up to 4x5 as a single frame or come in at 1:1 and stitch the negative at what amounts to over 8000 dpi.
The files are then brought into Lightroom CC that stitches the NEF files automatically into one large and highly flexible DNG file. The scan quality is a significant gain over even a drum scan let alone the 9000ED.
And the best part, unlike a scanner, this DSLR scanner setup is not a one trick pony but master of several. Obviously the D850 is a scanner component and a camera used for taking photos but the Sinar P2 is that AND a mural enlarger! The geared movements make aligning it for each custom job a joy to use and the output tightly controlled. Using live view and the electronic shutter on the D850 make the final output top notch.
The modularity of it all and the fact that the "sensor" is so easy to upgrade makes a dedicated film scanner completely obsolete for me. I have no doubt in my mind that DSLR scanning is the way digitizing film is going and as more software and even in camera inversion algorithms appear this will spell that out pretty clearly.
I applaud the OP for his efforts but I think he is too late, I think by the time this dedicated scanner is ready to sell, several brilliant new ways of using DSLR's to scan *and* export a right viewed raw file for any use imaginable will have hit the market and all but decimate the market for such a device.
I don't say this as a critic of the efforts the OP made but more of a heads up that his team may want to course correct before it is too late.
Kamph
Established
Can you show us some comparison scans? I've heard this claim many times for many years, but noone has been able to provide any direct comparison that actually show said advantage as of yet.
Corran
Well-known
The simple fact of the matter is DSLR scanning is a hands-on, frame-by-frame affair. Stitching exponentially increases complexity and slows throughput. Of course there is/will be accessories or workflows that help mitigate this but a traditional scanner that can in one sitting and with no user input scan a whole roll of film (or multiple rolls of film), or several sheets of LF film (and large ULF film), still has use and benefits. I'm not necessarily talking about this scanner, but more generally high-end flatbeds and autofeed-type scanners. DSLR scanning is novel but practically has enough drawbacks that I doubt it would ever be "universally" used.
KM-25
Well-known
Since I am out of town on a shoot, the only samples I can show are these two of a dark Kodachrome slide taken NYE 2010 in Times Square. On the light table it has wonderful tone but in the drum and 9000ED scan, the lower mid tones into the shadow detail just block up, drum only a bit better than 9000ED.
But the D850 version being a NEF file is so much better in dynamic range alone that it is actually the opposite of the above as I see things in the scan I can't easily see on the light table! The color cast is also *much* easier to control in the 850 output.
For the record, scanning is a small part of my workflow as I use a darkroom to print my black and white so I always digitize the prints rather than the negatives so I have 100% accuracy in how I show my clients the images online. So I really only use negative and slide scanning for the small amount of color film work I do and this system easily takes care of that, is highly upgradable and maintained. Once I am done with my Kodachrome book, that use will get even smaller so if I decide to divest of film scanning, all the hardware is used for other purposes to begin with so there is essentially zero net loss.
I'll post the two Kodachrome scans but that is about it as I am a working photographer and don't do a lot of show and tell of tests because of the time that takes away from more important things.
I wish the OP luck but I think they might want to get on the DSLR scanning bandwagon and quick.
Attachments
KM-25
Well-known
Funny thing, this kind of thinking is what Kodak did when digital photography began to gain steam. Through both demand and innovation expect it to get a lot easier, faster and better. It takes me less time to stitch a 4x5 neg than it did for my 9000ED to do a single frame of 35mm. Then my iMac Pro blasts out an 800MB DNG file of that stitch via Lightroom in mere seconds.
I am willing to bet you $100 cash that in less than 10 years DSLR scanning will account for nearly all enthusiast scanning and when one considers the scan houses that now already use the X, Y, Z staged Phase One system ( used for my 12 foot murals from 4x5 negs in Denver ) a large majority of pro service bureaus will have made the same switch.
I have a $100 bill stuck in my night stand drawer just for you in case I am wrong Mr. Garris...🙂
Scapevision
Well-known
I fear this project will end up on the no-longer-developing-project-after-thesis-done bandwagon.
Corran
Well-known
I don't know what you mean by "enthusiast scanning." I don't doubt it will become more accepted (it already has come a long way since I first experimented with it on 4x5 myself, back 5 years ago or so).
But a system like yours, for 4x5, is certainly far and beyond the needs, or wants, of most 4x5 shooters, whom I would call "enthusiasts," or hobbyists. It also is much more expensive than a V700 (of course the Epson is garbage, but again, for hobbyists it is more than fine).
Add in larger formats and you have a problem. Now, LF is a medium which lends itself to small-batch processing. For 35mm, and lots of it, most DSLR scanning will be cumbersome and require lots of hands-on work, while an auto-feed scanner will be way faster in every way. But, for those shooting a couple rolls of film a year, sure DSLR scanning is great. A very high-end institutional system for faster camera scanning is not going to be in everyone's home.
I have no interest in making it a bet, nor does it really matter to me what others choose to do. I've used a lot of scanning gear over the years and I think there's a lot of life left in those systems, though to be fair it is unfortunate that much of it requires older machines and such. I was interested in this project in some ways but at this point I can't justify it anymore as my current Screen system is fine.
Not sure why you seem to want to intimidate me by using my real name.
KM-25
Well-known
LOL, it was actually more a long the lines of a pat on the shoulder, all in fun hence the smiley.
The batch scanning, that is where the innovations will come in and it will end up being much faster than a scanner because the DSLR is. As far as enthusiast scanning, yes, that means hobby film shooters who don't have access to a darkroom or prefer hybrid process to both share online and print from film originals.
To everyone else and that includes the OP, it is because I applaud and respect the OP's effort that I want to openly share both what my direct personal experience is and what I am hearing from industry types regarding scanners and DSLR scanning.
The OP still has time to shift attention to making an outstanding DSLR scanning rig with great software to boot that could seriously net some sales.
Innovate folks, this is a ripe peach to be picked!
Corran
Well-known
I don't know, I don't see a big market for innovation in batch scanning, unless the supposed "film resurgence" really makes it big, for the same reasons we haven't seen dedicated film scanners improved much for a decade. We'll see.
As a side-note, I thought this scanner was using small sensors like from cell phones that basically is similar to DSLR scanning anyway - just without the S or R part. As opposed to single-line sensors that scanners have. I could be mis-remembering though.
Another side-note - supposedly one big "negative" on DSLRs is the Bayer sensor. I have never seen an in-depth comparison of camera scanning between a traditional Bayer sensor and one of the Sigma 3-layer sensors, that some say have "better" color. Would be an interesting comparison wrt color.
As a side-note, I thought this scanner was using small sensors like from cell phones that basically is similar to DSLR scanning anyway - just without the S or R part. As opposed to single-line sensors that scanners have. I could be mis-remembering though.
Another side-note - supposedly one big "negative" on DSLRs is the Bayer sensor. I have never seen an in-depth comparison of camera scanning between a traditional Bayer sensor and one of the Sigma 3-layer sensors, that some say have "better" color. Would be an interesting comparison wrt color.
brbo
Well-known
In 10 years film hobbyist will scan with sensors in their phones.
And the more serious rigs for hi-quality hi-speed batch scanning won't be any less complicated that the OP's concept.
And the more serious rigs for hi-quality hi-speed batch scanning won't be any less complicated that the OP's concept.
Similar threads
