Welcome to the new 8mm Forum!
The forum you are looking at is entirely new software. Because there was no good way to import all of the old archived data from the last 20 years on the old software, everyone will need to register for a new account to participate.
To access the original forums from 2003-2019 which are now a "read only" status, click on the "FORUM ARCHIVE" link above.
Please remember registering with your first and last REAL name is mandatory. This forum is for professionals and fake names are not permitted. To get to the registration page click here.
Once the registration has been approved, you will be able to login via the link in the upper right corner of this page.
Also, please remember while it is highly encouraged to upload an avatar image to your profile, is not a requirement. If you choose to upload an avatar image, please remember that it IS a requirement that the image must be a clear photo of your face.
Thank you!
Since that Elmo telecine projector uses synchronous motor I guess there's literally no "drift" throughout correct? That should make the process quite a breeze then.
It works quite well at at least for 25 minutes of this film.
Here are the steps:
The camera was run at 48 FPS which is twice the projector speed. This way only one frame will always be good without ghosting.
The stills were then extracted. I used Virtual Dub for that.
Set format of extracted images to jpg. Tiff takes up much more space.
DOS script used to pick either odd or even stills.
The odd or even stills imported into DaVinci Resolve and the Resolve clip attributes were set at 24 FPS (this is important step).
Created timeline from clips.
The original video is then pulled into Resolve with the sound.
Now you have 3 timelines: Original Video, Original Sound, Image clips timeline
Delete original video timeline.
This way you end up with the odd or even video with the synched sound and no ghosting.
The DOS script is available here: https://drive.google.com/drive/folders/1pf...
Originally posted by Nantawat KittiwarakulView Post
For Super8 films I would use it in frame by frame capture method as per usual, at around 2.5 fps. For 16mm films I would do frame-accurate capture at (about) 24 fps. The audio was separately captured then combined later. The result would look like this.☺️
This is done by TEMPORAL OVERSAMPLING - setting the camera's frame rate to 48 fps. I had the projector's shutter blade removed, and (mechanically) "tuned" the projector's speed to darn close to exactly 24 fps. The resultant video would be "clean" captured frames interleaved with "blended" frames. The rest is just a matter of (patiently) selecting only "clean" frames in the editing program. A bit time consuming process, but still faster than frame by frame method.😎
Have you ever tried this solution with the Elmo 16CL or Elmo ST180? I assume the shutter wheel should be removed because the Lumix uses the rolling shutter.
Since I have no speed-accurate Elmo TRV-16, just a plain ordinary Eiki projector with inaccurate drive motor so there's not much progress on 16mm side. However Robert's mention about BLDC did give me the idea. Now noted that to my shopping list.
But on super8 side I've just made a MAJOR success - literally the biggest achievement in years.😎
Got this rather unique Fujicascope SD20 projector a while ago (I probably had posted about it back then too). Knowingly that for its paticular function - quartz controlled motor with dead accurate speed, it can be set to run at PRECISELY 18.00(000) or 24.00(000) fps. Initially intended for sync-sound editing/dubbing, but it will also be a great candidate for DIY film scanning setup too. However only recently that I managed to did some needed modifications - and eventually tried it out.
Replacing its original burning hot lamp with cool LED light source. Since it can be run from 12 vac source so I simply feed it with a direct tap from the main transformer - simple as that. Then adding a thin piece of white acrylic as a diffuser, as per usual.
Then having the shutter blade removed in the non-destructive way. A very involving & time-consuming major disassembly, but worth the attempt since it can be always reassembled back if needed.
The only irreversible damage done to the mechanism is widening its aperture to make the entire film frame visible. But since this projector shares the same chassis with some other make&model, so it's still possible to find original/untouched replacement gate parts from other scrap machine.
This is all for projector part. Now the camera part...
I can use either 1080 or 4K resolution provided that the camera's system frequency is set to TRUE 24.00 fps mode. In other words other cameras with only 23.976 fps will NOT work, as the timing will drift within less than a minute rendering it unsuitable (sorry to most Sony cameras).
Now with the variable frame rate mode: on, set the speed to 48 fps - twice the projector's speed. The 18/36 fps combo setting should also work too (turns out that it doesn't, more about that later).
Now run the projector as normal, and start recording on the camera side.
The captured file will usually contain "good" frames interleaved with "blended" frames (or ALL good frames if I'm lucky). If the "good" frames are even frames for example, it will be solidly consistent throughout from the first frame to the last frame. With some tricks & settings & playing around in the editing program I can cleanly filtered out the blended odd frames, keeping only good even frames. The rest is just a matter of setting the correct playback speed to 18 fps then start editing as per usual.
So yes, I can now scan super8 reel at best possible quality, and in real time (in fact even faster considering most films are shot at 18 fps). No more slow 2fps capture speed. No more tons and tons of still image sequence, just one complete video file per session to deal with.
The only gotcha is that the camera will NOT record sound when in VFR mode thus this can be an issue with sound film. But since super 8 sound film rarely ever shows up anyway I would just run a separate audio capture pass then match them together when editing, and call it a day. I'm unable to think of any other major improvements so this could be considered that I've reached my "goal". And would live happily with this setup well within foreseeable future.
PS for some reasons the projector's 18 fps setting isn't as dead accurate as 24 fps. At 18 fps it will "drift" away for about a frame after about 20 minutes. Although that could be corrected later on while editing - but better just work at 24 fps, and be done with that.
Great work. You can use the dos script I posted in one of my previous posts in this thread. The script works very well. It is a bit slow, but I mean who cares...My RPI solution with the TRV16 also works and the sync is right on but the issue is with resolution. It runs only 800x600 at 48 FPS. That is why I am considering to go with your solution or maybe try this camera: https://www.amazon.com/HD-100A-Imagi.../dp/B082J9376G https://cvp.com/pdf/hd-100a-aida-user-manual.pdf
Thank you for posting.
Actually that HD100A was a waste of time. cannot even do 24 FPS. I gave up on it and returned it. So the only reasonable solution for a high quality transfer without flicker and ghosting is the Panasonic GH5. I do have a solution with iPhone X with very good quality 4K capture with sound but it is a rather complicated setup. It has no flicker and very little ghosting. I will spend some more time on it and try to simplify it but it will take some time. The main issue is the custom lens mount that is not very sturdy and another issue is vibration. Has to mount the iPhone one a tripod and the alignment is a pain. Currently it is set up with the Elmo Cl16.
In the meantime I am interested in trying Nantawat's solution with the GH5. Too bad that it cannot record sound. But possibly the sound output can be routed to a cheap HDMI recorder such as: https://www.amazon.com/gp/product/B08LPSG8CY
The video from the GH5 then can be split into odd and even images sequences and one of the sequences can then be brought into the video editor.
Then pull in the HDMI video plus sound and then delete the video.
You end up with the resulting video and sound perfectly synched.
As I mentioned before I have a TRV16 + RPI solution. The RPI is running at 48 FPS and there is no flicker or ghosting but the resolution is low.
It is 800x600. https://github.com/vintagefilmograph...frame_by_frame
My solution for sound film is even simpler - just do 2-pass capture.😆
First pass at 24 fps to capture audio. There'll be loads of blended frame of course but that doesn't matter. As long as there's enough "good" frames for reference when editing that's OK. This might be a good chance to assest the image quality, and to do the white balance/exposure tweak.
Then capture the second pass without audio as previously described.
Then throw both clips onto the timeline of the editing program. Align the very first frame of both clips, then verify that the very last frame of both are still aligned. Have the video track of the first clip disabled/muted and boom - export the finished product as needed.
In practice any audio delay less than 0.1 second will be unlikely to be noticealbe. Therefore I wouldn't give a dime if there's a 1-frame audio delay - just go ahead and get the job done.😁
Regarding the camera to be used there're two critical catagories that come to my mind.
1.The camera must have exact 24.00 fps frame rate, or there must be means to precisely control the projector's speed to match the camera's framerate. I'm thinking about Elmo GS1200 - should be the perfect candidate, since you can set it to run at ANY designated speed according to the input sync pulse frequency.
2. The camera should have fast sensor readout speed. Typical super8 projector would have about 60° pulldown, or 1/6 of the cycle. At 24 fps each cycle would last about 41.67 ms, so 1/6 of a cycle would be around 6.94 ms. However since the camera runs at twice the projector's speed, so about 20.83 ms. Deducted with 6.94ms pulldown, then there's about 13.9ms left for exposure & sensor readout. With some safety margin added, I'd say sensor readout should last no more than 10 milliseconds.
Most Sony cameras do have pretty slow readout speed, sometimes 30-40 ms hence most of them are out of question. To my knowledge GH5's sensor readout speed is in fact 20 ms thus it should't be suitable for the task. However that figure is for total sensor heigh - 3,888 pixels. With Extended Tele Converter function on it will give 1:1 pixel readout from only center portion of the sensor, reducing the readout speed considerably. At 4K resolution the readout is supposed to be about 11.1 ms - just enough to fit in the standstill phase of the cycle. At 1080p it is even faster - around 5.5 ms, so you have more than enough leeway at this resolution.😁 The next model in the series GH5s with less pixel count should have even faster readout so even more suitable to the task than plain GH5, but I'm yet to find that exact figure as of now.
The GH5 looks like a good choice. My RPI 48 FPS solution works with the Elmo 16CL. Will try it with the ST180 but as I said before it is low rez (800x600) in this mode. It is a single pass.
Here the ST180 mod video run at 24 FPS: https://drive.google.com/file/d/1VSm...ew?usp=sharing
The resolution of this video is 1440x1080
The issue is ghosting.
I will try 48FPS with the same setup at 800x600 rez.
Here is the iPhone setup. Almost no ghosting at all but pretty complicated setup. Designing a new lens setup that should make it simpler. https://drive.google.com/file/d/1cnU...ew?usp=sharing https://drive.google.com/file/d/1r5E...ew?usp=sharing
Then adding a thin piece of white acrylic as a diffuser, as per usual.
).
I'm unable to think of any other major improvements so this could be considered that I've reached my "goal". And would live happily with this setup well within foreseeable future.
Tweet
Comment