Do you have a pdf or any file that lists the parts needed for the construction of the film gate guides/rails?
I don’t really understand your question. Yes, I stabilize the perforations using the capture software and then in DaVinci Resolve. But the scanner already produces images that don’t jump around all over the place.
Do you have a 3D printer that would let you print these parts?
Creating a PDF with a parts list including the dimensions, and attaching the printable files, is a lot of work. I’m looking for another, simpler solution for me.
Maybe a single .stl file with all the parts for one module. Then the user would select the parts to print and also be able to calculate the dimensions of the parts to buy. I’m looking into that.
Dear Roland,
I’m finally at the point where I can confirm full operational status of my copy of your machine. Took me a while, learned alot. It wouldn’t be possible without your help, so God bless you.
Already doing updates, ordered a bunch of different LEDs, with intention to build narrowband mixing box. Also looking for updated transport system and improvements in software. Fortunately AI helps with coding significantly and scanner operates on custom designed application.
5 Likes
I’m really pleased to see that your project has come together, and just as glad to see that you adapted the scanner’s operation to match your own ideas.
Your way of handling the triggering of the flashes and the camera through a Basler system seems like an excellent approach to me.
Please keep sharing the improvements and modifications you’ve made — I’d love to read more about them. 
1 Like
After repairing the projector electronics and audio, while the audio works, it unfortunately is picking up nearby radio stations. I ordered the amplifier based on the TA7668.
From the initial testing it certainly works better than the old electronics.
Note that the output level is suitable for a small headphone (about line-level).
When using the amplifier, please note that the two wires connected to the audio head will go to the audio input GND and R (or L). The floating wire shield can be grounded to the power supply, but do not connect it the input GND.
Thank you for this information. In my first attempts, I had two problems at once: an amplifier board that wasn’t working and the sound head I took from a projector wasn’t working properly either. That made it quite difficult to get my homemade setup to work even roughly. I managed to get a few noises out of it, somewhere between the vague monologue of a drunk man and the mooing of a cow giving birth.
On the other hand, the idea of testing directly on the projector is excellent. It removes the head-positioning issues and will let me focus on the amplification first. As for the interference problem, I’ve already been able to hear that the stepper motors produce noise, and that will probably be difficult to solve, but I’m not at that stage yet.
And since I’m in the middle of moving house, I’ll have to take a forced break, but I’ll come back to it.
Many thanks for your valuable advice.
1 Like
There’s always the transmission line solution that the radio folks use: put everything in a metal box, including the wires!
If you used 75-ohm terminated, shielded coax (say, RCA cables) running between all PCBs (which should also be in their own metal boxes), that sort of interference should disappear more or less completely. You just need an extra 6dB (or 2.0x or 2V/V) gain op-amp after each terminated signal to bring it back up since properly terminating the transmission line cuts the voltage in half.
I would have liked to have heard that! 
I should have been able to get the old electronics a bit better, but it is time consuming without an schematic. And don’t want to remove the heads to replace the wire, to avoid alignment issues.
So at the end the new board is better and small enough to fit near the heads. And yes, a metal box is part of the plan!
1 Like
Hey Roland,
My first worthy upgrade which I want to share with u is narrowband lights for negative color film. I’ve ordered custom 3-UP chips of Cree XP-E2 LEDs (650nm, 530nm, 450nm) on a single plate (actually 9 of them) which are balanced and provide narrowband light output supreme for C-41/ECN films. For BW and Positive films I’m using 3000K BXRC LEDs as positive film requires warm light imitating the projector lamps. So yes, I need to switch light sources for different film stocks, but results are way superior than white light.
Just to illustrate quick snip comparison between white light capture vs narrowband capture of a negative film. Insane.
Very interesting. Great example! It shows that narrowband LEDs are a must for negative footage. I argued previously that with color reversal material, the opposite is true: you want to use broadband illumination in this case.
The situation is less clear if you have severely faded material. Here, a multi-spectral approach might be the best option.
How did you select your wavelengths for the color channels? There should be some optimal combination for a specific film stock. On the other hand, my color enlarger for analog photography is also using a generic combination of narrowband (interference) filters and that works fine for various film brands…
I’ve used information from this research - GitHub - jackw01/scanlight: A better light source for scanning color negative film · GitHub which seems to be accurate enough to pick perfect LED pairs. Also I was using his scanlight and compared with my Nikon Coolscan 9000 (which was industry standard) to verify that narrowband lights excel over white lights. The benefit is that the conversion process becomes way less complex, simple flip works amazingly good.
2 Likes
Nice link! Would be interesting to look at the setups of old analog equipment and how they solved this. Wonder what illumination setup an old optical printer used…
1 Like
Great information, thank you for sharing.
One observation. From the datasheet of the Basler A2600
When using white illumination… that is certainly a significant limitation on the red channel. The Basler falls sharply at ~650 nm.
Going back to the Swiss selection of narrow bands…
The sensor Red band captures the Magenta-Cyan overlay, and no doubt that plays a role.
Looking at a negative film dye density curves (for Kodak Vision3-200)…
The Cyan is centered above the falling edge of the Red channel…
While I can agree that narrow band may be a practical solution… would qualify that the issue may not solely be the white LED, but the particulars of the Bayer sensor.
The information is interesting and timely, @dswanson just posted some examples of different narrow band LEDs.
@cpixip Rolf, you are the forum resident expert on the subject matter, and in the context of scanners based on the HQ, I have a question…
I am working on choosing between 7 LED bands to build a PCB to add multi-spectrum to the scanner. The IMX477 Red Bayer rolls off to about 60%, and then the IR/UV filter cuts off anything above. I use an external IR-UV with a sharp falling edge at 700 nm.
When selecting the LEDs, the one I initially decided to keep out was the Far Red centered at 730nm.
Now that I look at the Kodak negative spectral dye density… I am reconsidering.
If I have to pick three out of four bands, and the options are 615, 625,660, and 730…
(with the IR-UV is removed)… I am rethinking if it is best to leave out 625 (very close to 615), and experiment with 730 (without IR filter).
Thoughts?
Hi Roland,
Thank you for posting information about your film scanner. Your design is simple and robust. Your 8mm film scans are amazing. My film scanner is made from 2020 T Slot. It is Raspberry PI 5 based using a Marlin board to drive 3 NEMA 17 steppers. I have one stepper on each side of the film gate with pressure roller. I use OpenCV to locate the sprocket hole. The steppers move the film back/forward until the sprocket hole is found. This is resource consuming and I can only scan about one frame per second. I like your solution better using a laser sensor.
Are you a mechanical engineer? Your 3D designs are brilliant. The film spool hub locks are genius.
Would it be possible for you to post the remaining .STEP or .stl files for your components. Specifically the LED light block and the laser holder.
Thanks
My move isn’t finished yet, but the biggest part of the work is done. Sorry I wasn’t able to get back to you sooner.
@Andriy_But : Very interesting, and the result for your negatives is spectacular. But I have to admit I’m far from understanding the technical details of Pablo’s explanations.
@JohnSmithYou : Welcome to the forum. I see you’ve already done a great job building your first scanner, so you shouldn’t have too much trouble making a second one, which should be simpler in my opinion.
I’m attaching a link to download two .stl files. The first contains all the press holder and lighting components in a single file. In your slicer, disable the parts you don’t want to print, such as the 40x40 rails, and keep only the components that need to be printed.
In the second link, you’ll find the complete laser holder. The rods are not meant to be printed, but made separately, either from metal rods or from carbon tubes, such as kite spars, for example.
1 Like
I am new on this forum and would like to know if anyone can give me a path to follow and build this scanner. I have read everything carefully but the instructions or where to start is still not clear to me.
@Roland your design is amazing and I would like to know what is the limit of the camera that can be attached to the scanner? 6K, 8K, 10K? How many global shutter fps is the minimum required for the scanner to work properly. I saw the YouTube video and I really love that it can scan in real time.
@Andriy_But you made one almost by yourself. Can you share with me the steps you took? I will be forever grateful. And I know this will help everyone.
GENERAL EXTRA QUESTION:
Does this scanner support the capability to export DPX?
Indeed, there isn’t a clear step-by-step guide. However, I’m happy to answer questions and provide information when asked. Writing a full assembly manual with all the details, the list of parts to buy, and all the .stl files takes a lot of time, and I don’t have the energy to do that, sorry.
You don’t specify which one — do you mean assembling the chassis?
You can choose whichever camera suits your budget. It will determine the formats you can get. Mine has a maximum resolution of about 4000×3000, but I never use the full resolution, even though I come close to it for 35mm films. Keep in mind that your computer also needs to handle that data stream, and for the vast majority of films, a higher camera resolution will not improve the quality of your final image.
There is no minimum transfer speed; the range goes from about 1 frame per second to around 16 frames per second. Beyond that speed, sharpness may decrease. But this also depends on the power of your lighting and the image resolution.
My camera does not produce images in DPX format. On the other hand, I always work with 10-bit .tif files. I find that image formats are easier to handle and less prone to problems.
2 Likes
Hey fabriccio,
welcome to the forum, you’re definitely in the right place if you’re trying to build your own scanner. I can agree that current design is superior in terms of ease of building and mechanical complexity.
also I can confirm that this thread has all the information needed to assemble similar machine (and me personally is a statement for that). While I was also pursuing step by step guide and wanted to simplify the process I strongly recommend to read closely our communication. Now with Roland’s help there are even stl files for all the parts. I’ve designed half of the parts just watching Roland’s videos)
I assume you don’t have engineering background, which also applies to me, so building exact version of Roland’s machine would be much harder than you think. Building my version is way simpler as it has laziest electrical approach and I simplified it significantly.
My transport/capture design is built on three basic components:
- Aluminum frame (40x40 profile) which can be sourced locally or through AliExpress.
- Basler SLP Light Control + Basler camera of your choice (global shutter). I’ve got mine on eBay for 1/4 of the retail price.
- Laser of your choice (I went with Keyence retro reflection model) which you can wire up to opto input of SLP so all capture is controlled via Basler Pylon Viewer. This means that it will capture images on its own just by running motor forward. That’s it.
- I was using cheap universal motor driver, but now I’m building my own version on custom PCB. You don’t really need it to work, it’s just improvements that suit your own vision.
- Finally I’m vibe coding simple app using Claude. Just ask it for help - it will do the rest. Magic times)
In the end of the day it’s a DIY approach and you can built it the way you need it. I strongly recommend using SLP as it solves so many complex things you need to face when going Roland’s route. And lastly - it’s all here already, all the steps. You just need to start and eventually you’ll have a machine. And we can answer the specific questions down the route.
3 Likes
The camera suspension system and the lens mount were still missing; these two component groups ensure that the camera remains free of vibration.
The .stl file does not include the linear stages used to precisely adjust the focus (vertical adjustment) and the depth (lateral adjustment of the image framing).
The linear stages are available in 40 mm, for example from AliExpress.
2 Likes