DIY 16mm Film to 4K Digital Transfer System | Part 3
ฝัง
- เผยแพร่เมื่อ 19 ส.ค. 2023
- Here's Part 3 (the final part) of my Do-It-Yourself 16mm film transfer system. I've modified an Arri 16S camera movement for pin-registered, high quality film transport. In this video, we finish the build - troubleshooting along the way. The goal was to produce a system capable of frame-by-frame 4K transfer of negative and positive Super 16 and Regular 16 film. There's plenty of transferred final footage to examine.
- ภาพยนตร์และแอนิเมชัน
You remind me of an NASA engineer that designed solar panels to fold like origami, he was an engineer and an origami passionate. You are a master and an artist in what you do, people like you move this world.
Thanks so much!
Great video! Regarding the motor speed drifting, there's a few possible ways to get around this. In short, stepper motors themselves can't really drift. It's either in your speed controllers or stepper drivers. Given stepper drivers tend to work based off step + direction pulses from the controller, I'm more inclined to think it's the speed controllers.
1.) If you want 2 motors to spin at *exactly* the same speed and never drift apart, you can actually power both of them from 1 stepper driver. I can't tell exactly what drivers you are using, but based off their size they should have more than enough power for it.
2.) If the 2 motors have to spin at different speeds, you can use a microcontroller. It can precisely output step pulses to both drivers at precise and defined speeds, ensuring they always move at the same speed relative to each other. Something like an Arduino could do this with a fairly low amount of effort.
This is good advice.
On point one: Assuming you wan to keep two drivers for the current requirements (looks like you have a fairly large motor there) you can wire the step and direction pins from one speed controller to both stepper drivers so they are receiving the same signal. This will ensure they are driven at the same speed. You would wire the step pin in parallel to each controller. This is assuming both motors have the same step angle. Not all motors are created alike. It will be labeled on the bottom of the motor. The most common is 1.8 degrees, but even if you send the same pulses to motors with different step angles, they will move differently.
On point two: I agree that a simple microcontroller here would do wonders. You could program whatever ratio you would like. For example, two step pulses on one driver would trigger one on the other. 2 to 1 ratio. This is very simple. Feel free to reach out if you want more advice on this.
Thank you so much for sharing your creations with us! I tried to make a telecine machine when I was a teenager. An HP engineer down the road even helped me build the circuit to run it all. Your videos are absolute chicken soup for this soul and a walk down memory lane.
That's great - thanks for watching!
Another brilliant installment to this series. I’d like to make one, but for use as an optical printer, with a Bolex as the camera, and the whole machine in a vertical orientation. Well, one can dream…..
There used to be a company in the 1970s and 80s that made the exact optical printer you're describing - I think it was JK Cinema, or JK Camera? I looked for one, but was never successful.
This build was really fun to follow and the results are wonderful! Should you ever consider rebuilding it, to couple the sprocket wheel loop former to the main film movement, you could CAD it up and order a big sheet metal plate with some bends and ribs for not much money.
Thanks - I did consider that!
I look at the ISO on your camera and it is 640. When I made the film digitizer, I checked the ISO noise and image distortion and made the LED brightness so that the camera ISO was no more than 100 with minimal frame distortion. I also reversed the image right in the camera so that I could see the positive image. I also limited the amount of light reaching the side of the film that is captured by the camera so that there are no possible highlights of the film.
but your device is made very well, bravo 👍
On the Lumix S5, 640 is the native low ISO - meaning the dynamic range scale above and below medium grey is equal. I could choose a lower ISO and get slightly less noise, but this also moves the dynamic range so there is less information in the highlights, and more in the shadows. By using 640, I can also stop the lens down to F5.6 to f8 for maximum depth of field and sharpness. Thanks for watching!
Thank you for pioneering a new aerial printer, this can bring film back for lower budget productions. My only suggestion is the addition of replaceable 'felt cleaner pads' to remove any dust particles before entering the gate. I certainly want to purchase an unit with an UTRA 16 gate (which is better than Super 16 for 4K transfers). Hopefully you will be able to get a manufacturing deal to sell these globally, maybe a company like Proaim would be interested, they make and sell good lower cost production gear. Either way, let me know when you go to market.
Thanks - we'll see what happens!
Been excited for this video! Brilliant work as always
Glad you enjoyed it!
It should be relatively trivial to take a control board that's designed for a 3D printer and reprogram it to drive those 2 steppers (or even all 3) in sync. There also would be inputs available for loop-size sensors, either microswitches with rollers or optical ones. However, this would require someone for whom programming those things is within their skill range...
A side effect of this approach would be greater stability and faster speed. The latter because the controller could drive the motors at a variable speed instead of one that's dictated by how long the frame needs to be exposed. It could speed up for transport and pause for exposure. And that pausing could extend to all motors, providing greater stability as the picture is taken while all movement has ceased.
It might be worth it to reach out to the 3D printer community; it's full of people who love these kinds of projects. That's how 3D printing at home started, after all...
Excellent video. Really enjoyed the buildup. Fantastic work. Very inspiring.
Thank you very much!
A fascinating series of videos. And with only limited engineering tools available. You were able to create, improvise, and solve each engineering hurdle as it arose. And in some cases reusing simple existing everyday objects. To create a precision film transport deck from existing redundant camera parts combined with new off the shelf items. Very creatively done indeed, with excellent end results.
PS As you demonstrated in the videos, a file is a very simple tool but it is also a very powerful one.
Thanks, I'm glad you enjoyed it - as did I!
Great video and congrats on the finished scanner. Since my last comment we managed to finish our scanner project aswell (the one utilizing an old Pathe projector).
There's just one tiny detail I wouldn't be happy with: It seems like you need to shoot at ISO 640 for optimal exposure thus introducing some sensor noise to the original grain of the footage.
It's not an issue if you clean up the images afterwards anyway which you seem to do but I personally wouldn't like that. Maybe a stronger LED would be beneficial.
Great idea to use topazlabs software by the way, I wouldn't even have considered it.
And I have another suggestion: It might make sense to change the mechanical switch to a magnetic switch sooner or later. I feel like it has less points of failure in the long run. Then again, never change a running system. If your switch eventually breaks, just add a magnet to the pulley and another one to enable contact in a box with contact plates below the pulley.
Some questions though: Do you get perfect registration or does the image move slightly in the gate? (I presume you stabilized the footage in post right?)
How do you deal with dust and scratches?
Cheers!
Thanks for watching! I did not stabilize any of the footage. If you look closely on the few static shots, there's some jitter up and down, but I think it was mostly from the camera itself. 16mm is so small, that any movement in the best system is magnified. My goal overall was to not use any digital stabilization. I used different cameras over the years, and there's a slight difference in the amount of jitter, but it's hard to say whether it's unique to the mechanical condition of that particular camera, or to the design in general.
Honestly, these videos are so inspiring it makes me want to create my own 16 mm digital converter, but as I can tell by your videos it takes a lot of time and work and the know how of all the parts and what not, but the footage you captured looks incredible!!!! I did wince when you said you used AI upscaling but I actually agree with you conclusion about it being sharper. Again great video I’m so excited for the next instalment/update :)
Thanks for watching!
Love it! Personally I think AI upscaling is not well suited for this job. Specifically, the AI models are not trained on this type of film stock, and it creates weird artifacts and halo effects around the natural blooming present in old 16mm. I'd love to see someone train an upscaling model specifically to 16mm footage to retain some of the natural grain and hazy edges.
Thanks for watching!
The end result is amazing, I had no idea 16mm could look so good. I have reels of time-lapse lying around, probably getting wrecked in the Australia heat. This has made me want to take better care of them, and make my own contraption! I wonder where where I could find a film gate
The "Scrubs" series has been shot on 16-mm first to last season, with the same Aaton camera. They released a single HD episode for season 5, then starting from season 8 they released it in HD. 16-mm film was originally considered not good enough for HD programming, but most broadcasters have since relaxed their requirements, while the HD version of "Scrubs" as well as this build show that 16-mm can very well be used for HD shows.
@@ConsumerDV interesting! I watched scrubs. I remember it was grainy now and then, but it fitted the vibe
Yes, I think 16, especially Super 16 easily can meet HD requirements. One of the key things I'm not sure I mentioned was the quality of the lens used - I shot all my footage with primes, mostly Mark 1 Zeiss Superspeeds - not the best ever made, but better than most. Some of the zooms from the 60s and 70s were pretty bad.
I watched your number 1 and 2 build and the effort you put in has paid dividends. This episode has convinced me to consider a 16 transfer machine, after I have finished my dual 8 machine. Excellent production Fresh Ground! 😊
@@brianmuhlingBUM Thanks! Depending on the footage you have, using a projector as the base would also work - maybe not quite as stable, but the camera mounting and LED light source are the most critical.
I also prefer the upscaled result, looking real good 👍
Nice solution, and excellent results. As others have noted, steppers dont drift, so you could replace your pulse generators with an aduino, and once you tune your two pulse speeds, they'll never go out again.
I bought an old film projector as the basis for my (as yet unrealised) DIY Telecine, figuring that they already provide the syncronisation between the gate and the take up reels, with only one motor needing to be replaced/ splowed down. I take it you choose the Arri film gate becasue it has more accurate register, and you owned them? Over here, those parts are pretty expensive.
Yes, that's why I used the Arri parts. On my channel you'll see I also made a Super 8 device from a projector - that was fairly easy. Then I made a 35mm transfer device using an old Arri X-Ray camera - again, that wasn't too difficult because of the built-in sync. Thanks for the input, I just need to study-up on Arduino, as that seems the best solution.
@@FreshGroundPictures I also just did my first ever project with an aduino. Turned out to be pretty easy. Excellent help from the aduino forum. Go for it.
Always nice to see another great TH-cam channel in the comments @RotarySMP . Glad to see it’s not just planes and lathes that tickle your fancy 😂
So far the best DIY Scanner I saw on youtube. Great job!
Another good video.
Thanks again!
That is I-n-c-r-e-d-i-b-l-e scanning quality!
Amazing build and even more
Amazing how well you were able to shoot 16mm . I think the section about upscale vs varioys other processes is interesting. With how far ai is moving now i would be very interested to see more tests such as scanning 4k vs scanning 1080 and upscaling.
This is a fantastic project. Congratulations on a job well done.
gosh i would love to have one of these, but dont have nearly the amount of required engineering talent...
It's possible to convert a 16mm projector like I did for the Super 8 transfer machine - maybe not as steady, but much simpler in theory. Thanks for watching!
As all your projects, excellently presented, and fantastic results.
You should drop the loop, and use a spring loaded guide roller, with a switch that stops the motor on the uptake side when triggered. That would give you constant uptake and reduce the tension on the gate.
I've considered that - thank you!
Curious why the after effects pass was necessary. Oh, probably the raw files wouldn't open in Resolve. I generally scan 16 mm at 2K and edit on a 4k timeline in Resolve for 4K output. Does the Topaz pass add anything to your quality?
Yes, because Resolve doesn't support raw still images as a sequence. The Topaz software does wonders with image sharpness and clarity, but it does tend to reduce the grain, which can be good or bad depending on your preference. I found that processing 4K all the way through did not substantially improve the image beyond 2K. With Topaz however, you can make a susbstantial difference in clarity from 2K to 4K, without the computer-intensive 4K processing all the way through.
@@FreshGroundPicturesMy yt channel is pd films,mostly old sd transfers. I haven't had good luck with the Standard Def interlaced transfers in Topaz. I will try it again on the Resolve output prores files. Maybe starting with the interlace already removed will improve the output. I built a scanner about 10 years ago. If you have questions,please reach out.
@@VintageFilmChannel I checked out and subscribed to your channel - looks very interesting! If you have interlaced masters, standard def, and if they were transferred on a fairly good machine - like a Rank or similar telecine, you can remove the interlacing and return the file to a 23.976 cadence simply by loading the file into After Effects, then right click on it and go to "interpolation". It should allow you to choose "Guess 24P 3:2 pulldown" , which will remove the interlacing and produce 24P frames. This only works if the 3:2 pulldown is consistent throughout the file, and I don't think it works if the movie was transferred on some old projector-type telecines. If you can get the file back to 24P, Topaz will work the best that way. You can even reduce, or dehalo excessive sharpening done in Standard Def. If you can't return a file to 24P, Topaz has a couple of deinterlacing codecs that create 60p files, which can the be reduced to 30p. I've found the deinterlacing to 60p, 1080 looks pretty good provided you tell the software if the dominant field is top or bottom. Of course there are many issues which could arise depending on how the initial video was edited, converted or otherwise altered.
Great video, if you convert the RAW files to DNG RAW files with Adobe DNG converter you can import them to Resolve and skip the destructive step to ProRes.
hello, I am following your work with interest, great work, I plan to make such a browser in the future.
The shutter mechanism you made for the camera shutter is very clever and practical.
as an alternative method:
How do you think it sounds like moving the reels with a stepper motor and making short movements that correspond to 1 frame and sending a shutter signal to the camera with an ardunio?
thanks
That sounds like it might work, however the Arri gate mechanism I used requires a 360 degree rotation to place each frame. If you could time the take-up reel to move after the shutter snapped, that might work, however the amount of film taken up changes consistently as the reel rotates, as more film is accumulated. Thanks for watching!
thanks for these !
Glad you like them!
you can use casptor to connect it to the motor it will give you speed without any problem
What are the AI enhance ment done and software used
I use Topaz AI to uprez a 2K file to 4K. You have to experiment a bit with the settings - use the algorith that allows manual input of progressive frames. I usually tweek the sharpness up to around 30, and apply a lower number noise reduction. You could output a 4K file directly color corrected from Resolve, but as I demonstrate, it's sometimes not worth the extra processing time - I use Topaz on a Mac M1 for the AI uprez and it's pretty fast.
You are a beast
Timing your motors.. that's a raspberry pie chip job.
Fabulous work! I built a Super-8 scanner 2 years ago and encountered many of the same issues, so I greatly appreciated the thought you put into your solutions. I'm contemplating building a 16mm scanner, but I have 16mm magnetic audio film which needs to be transferred and I don't have a good solution. Do you have this issue or have you given any thought to solving it?
I love this series, and being a stills film only shooter, I can offer no help. I do have a question though, what about sound? Are there separate sound recordings? Do these have to be dubbed in later?
As in most film production, sound is recorded on a separate recorder - digital recorders today usually all work fine as they run at constant speed. The key to recording sound is having a film camera that has a crystal-sync motor system - the film then runs at a very exact speed, so it can be matched to the audio recording. A slate is film at the front of a sound take, and the clapper image is synced up with the audio "clap" sound in an edit system. Most digital edit systems have simple ways to then lock the image and sound together as a clip. Years ago, when 16mm (and some Super 8 film) was used for news recording, there was actually a magnetic stripe on the film edge, and the sound was recorded in the camera, thus synced with the picture. This allowed fast editing, but the quality was not as good, and it created flexibility issues in the edit - to my knowledge, you can no longer buy 16 or Super 8 film with a magnetic stripe.
@@FreshGroundPictures thanks, that's very helpful. I'm constantly amazed by the quality of video you make with 16mm.
A question, what is that lens on that camera and where should one source it if trying to build one
It's a 25mm Macro lens from Laowa - you can buy in different mounts through BH Photo, Adorama, etc.
Greetings to you, it is a very ingenious system and the results are great, congratulations.
I was wondering if it is possible to help me with a project I have, in my country there is no one who can digitize 16mm film and I have films that were filmed in the 50's and maybe you can help me, thanks for reading.
My opinion on grain reduction: You don't!! Either you shoot on film and deal with Filmgrain as brush strokes on a oil painting like an adult or you can just shoot digitally in 4K. I can only afford Video for the most part but HATE the clean sterile look of it and would rather shoot on Film. Especially 16mm. And i hate going through all sorts of lengths to make video look like film in Post.