meloV8
Established
Snake holder looks like film rail in my Pakon F235 scanner. Nice!
Designing a new film scanner; need your help
- Thread starter quejai
- Start date
- Latest activity Latest activity:
- Replies 338
- Views 15K
Bandiahegyrol
Member
Hi there,
Anything to publish? How is the refine-beta testing going?
Anything to publish? How is the refine-beta testing going?
quejai
Established
Hi, yep I should have mentioned this earlier. Currently in uni exams, so progress has stalled recently, however from around the middle of November the scanner project will be my main occupation.
Our intermediate deadline/goal is to have a technically working demo at the next Sydney RFF meetup (usually in mid december), followed by sending out some beta units.
pyeh
Member of good standing
Quejai, so cute - next Sydney RFF meeting usually in mid-December! We'd better arrange one then! Scanner is looking real good, have to say.
Good luck with your exams.
Good luck with your exams.
Bandiahegyrol
Member
Thanks for your reply. It seems to me youre at the final stage, looking forward to see the beta test results.
Good luck with the examms.
retinax
Well-known
I applaud you guys for making this machine! That's an awesome thing, meets a real need and seems to be executed very well. Unfortunately I won't be able to afford one in a long time...
Bandiahegyrol
Member
Normally i cannnot afford but i will.
pyeh
Member of good standing
Show us!
Show us!
Quejai, see this thread for an attempt at corralling cats for a Sydney meetup:
https://www.rangefinderforum.com/forums/showthread.php?threadid=163166
Hope you can come and show us the progress.
Show us!
Quejai, see this thread for an attempt at corralling cats for a Sydney meetup:
https://www.rangefinderforum.com/forums/showthread.php?threadid=163166
Hope you can come and show us the progress.
quejai
Established
Thanks all
@pyeh I'm there
I've spent the last few days playing around with making software that will turn the output from the scanner into the resultant scanned image.
Here's a real-time video of the software operating with a simulated scanner's output. A scanner simulator was chosen for now while complexities are being ironed out - I'm spending time now making the data from the simulator less reliable (ie more realistic), and the processing software more robust to cope with it. This will be similar to the live-output monitor in the final scanner software.
https://www.youtube.com/watch?v=NjdCzKp_Om4&feature=youtu.be
What should be clear from the video is the general operating principle of the scanner; which is that when the scanning head has moved a specific distance, a new 'tile' is captured from the scanner. The tiles outputted by the scanner loop through a specified set of color channels that correspond to the chosen film, illumination and sensor used. I am anticipating that the rate of tile acquisition is going to be limited only by the frames-per-second of the imaging module in use. I'm really not pushing it for speed in this demo.
For the technophiles, this functionality exclusively uses OpenCV in C++, meaning it's compatible with Windows, Mac and Linux.
source image attribution: http://www.publicdomainpictures.net/view-image.php?image=82107&picture=small-fishing-boat
@pyeh I'm there
I've spent the last few days playing around with making software that will turn the output from the scanner into the resultant scanned image.
Here's a real-time video of the software operating with a simulated scanner's output. A scanner simulator was chosen for now while complexities are being ironed out - I'm spending time now making the data from the simulator less reliable (ie more realistic), and the processing software more robust to cope with it. This will be similar to the live-output monitor in the final scanner software.
https://www.youtube.com/watch?v=NjdCzKp_Om4&feature=youtu.be
What should be clear from the video is the general operating principle of the scanner; which is that when the scanning head has moved a specific distance, a new 'tile' is captured from the scanner. The tiles outputted by the scanner loop through a specified set of color channels that correspond to the chosen film, illumination and sensor used. I am anticipating that the rate of tile acquisition is going to be limited only by the frames-per-second of the imaging module in use. I'm really not pushing it for speed in this demo.
For the technophiles, this functionality exclusively uses OpenCV in C++, meaning it's compatible with Windows, Mac and Linux.
source image attribution: http://www.publicdomainpictures.net/view-image.php?image=82107&picture=small-fishing-boat
pyeh
Member of good standing
QJ, that scanner simulation certainly looks the goods to this inexpert witness.
Looking forward to seeing you on Saturday. Please note though the change in meeting time to 4.00 pm. Hope that doesn't affect you.
Looking forward to seeing you on Saturday. Please note though the change in meeting time to 4.00 pm. Hope that doesn't affect you.
Jockos
Well-known
Did you get the Tech demo together? Is the beast alive? 🙂
quejai
Established
Regrettably, not quite. It will purr soon though...
A lot of prep was done (decoding the unusual raw format, developing the motion equations, setting up the fpga, ...) , the weak link was being unable to communicate control signals reliably over USB. I’m partway through implementing a better approach, let’s see how that goes.
Caleb’s been on the hardware and optics as usual, presently investigating the viability of a smaller lower-cost model 2 that still uses the same software and electronics of the bigger model 1.
Lets see how the christmas rush goes.
A lot of prep was done (decoding the unusual raw format, developing the motion equations, setting up the fpga, ...) , the weak link was being unable to communicate control signals reliably over USB. I’m partway through implementing a better approach, let’s see how that goes.
Caleb’s been on the hardware and optics as usual, presently investigating the viability of a smaller lower-cost model 2 that still uses the same software and electronics of the bigger model 1.
Lets see how the christmas rush goes.
Jockos
Well-known
Any updates for the new year? 🙂
quejai
Established
Well there has been a bit of design turbulence going on over the past few weeks, so it's worth clarifying where we're at. The USB issue is solved though!
Our investigations into a smaller model 2 revealed an alternate architecture for the mechanical construction of the chassis, which we could apply even to the larger model 1.
So far, it appears fundamentally superior in that the new approach has fewer parts / bolts / panels. Its far easier to put together and far easier to get each part aligned, to such an extent that the previous hardware seems somewhat silly. I'm very glad that we haven't sold a fleet of scanners made with the older design.
At the most recent RFF meetup in Sydney, we were throwing some ideas around and one consensus was the value of having multiple 'lanes' of film going into the scanner wasn't percieved as being particularly high.
Admittedly, a workflow involving loading the scanner with, say, 6 strips of film at a time, and of dealing with the scan files they produce, could be confusing enough that some users might just use a single lane at a time out of simplicity. There are advantages to having multiple lanes on the scanner, such as a larger maximum quantity of film scanned per load-unload cycle, but at the cost of significant design complexity.
Based on this, Caleb and I have decided that the first model of scanner we produce 1) will use this new hardware design approach and 2) will nominally have a single film lane. ie we will not initially produce the larger model that has featured in our updates so far, instead we will go for the smaller model. These are the primary alterations so far, but already we have a design for a smaller, simpler to use, more affordable unit ready to go. Caleb is working on an alternate optical system, but I'm going to use the same sort of approach as before to start with. Parts have been ordered for the prototype.
This unit will be less risky for us to produce and sell, which is super important as our first product.
Yes it is a bit of a back-to-the-drawingboard scenario, which is of course inconvenient for people who like schedules, but once we're ready to sell them I'm sure everyone will be glad that the better design was used.
As you might notice we try to avoid CAD renderings... so photos of the new prototype should appear over the next few weeks.
There is of course the risk that this new approach carries as-yet undiscovered design flaws, but those will soon be dealt with if necessary.
A whole lot of technical material on control systems, linear algebra, color profile handling, computer vision etc has also been reviewed recently.
So to summarize: We have dropped the previous scanner hardware design , and are testing out a totally new, smaller design that still uses the same sort of internals and general operating principles. This is because the previous hardware approach was unsatisfactory, and because we have deprioritised the importance of multiple-lanes over simpler operation and lower costs.
Our investigations into a smaller model 2 revealed an alternate architecture for the mechanical construction of the chassis, which we could apply even to the larger model 1.
So far, it appears fundamentally superior in that the new approach has fewer parts / bolts / panels. Its far easier to put together and far easier to get each part aligned, to such an extent that the previous hardware seems somewhat silly. I'm very glad that we haven't sold a fleet of scanners made with the older design.
At the most recent RFF meetup in Sydney, we were throwing some ideas around and one consensus was the value of having multiple 'lanes' of film going into the scanner wasn't percieved as being particularly high.
Admittedly, a workflow involving loading the scanner with, say, 6 strips of film at a time, and of dealing with the scan files they produce, could be confusing enough that some users might just use a single lane at a time out of simplicity. There are advantages to having multiple lanes on the scanner, such as a larger maximum quantity of film scanned per load-unload cycle, but at the cost of significant design complexity.
Based on this, Caleb and I have decided that the first model of scanner we produce 1) will use this new hardware design approach and 2) will nominally have a single film lane. ie we will not initially produce the larger model that has featured in our updates so far, instead we will go for the smaller model. These are the primary alterations so far, but already we have a design for a smaller, simpler to use, more affordable unit ready to go. Caleb is working on an alternate optical system, but I'm going to use the same sort of approach as before to start with. Parts have been ordered for the prototype.
This unit will be less risky for us to produce and sell, which is super important as our first product.
Yes it is a bit of a back-to-the-drawingboard scenario, which is of course inconvenient for people who like schedules, but once we're ready to sell them I'm sure everyone will be glad that the better design was used.
As you might notice we try to avoid CAD renderings... so photos of the new prototype should appear over the next few weeks.
There is of course the risk that this new approach carries as-yet undiscovered design flaws, but those will soon be dealt with if necessary.
A whole lot of technical material on control systems, linear algebra, color profile handling, computer vision etc has also been reviewed recently.
So to summarize: We have dropped the previous scanner hardware design , and are testing out a totally new, smaller design that still uses the same sort of internals and general operating principles. This is because the previous hardware approach was unsatisfactory, and because we have deprioritised the importance of multiple-lanes over simpler operation and lower costs.
pyeh
Member of good standing
Quejai, sounds like good progress to me. Here's to success!
CMur12
Veteran
I'm following this with great interest, too!
- Murray
- Murray
Geaux
Newbie
This is very interesting! Keep up the work guys, I am sure it will pay off in the end.
Sent from my SM-G920V using Tapatalk
Sent from my SM-G920V using Tapatalk
Tijmendal
Young photog
I think this is very desirable. Great stuff!
Gregm61
Well-known
I’ll believe this all when we actually see something.
Arbitrarium
Well-known
This thread has several videos of the prototypes...
Similar threads
