Having Trouble with DIY Telecine (8mm) System

[ronypony]

First time I have visited the forum in a while....wow new developments....I really like the sound of the 'software' approach using VirtualDub and Avisynth...very clever..

So basically you simply capture the entire film running at 3fps and use the software to process the captured 25fps captured footage...is that correct guys....I would really like to give this method a go and many thanks for sharing your script with us Vanberg...

Would it be possible to give a step by step instruction of how you set everything up to save all the trial and error for someone who is attempting this method of capture for the first time...it would be much appreciated.....I can't believe how this thread goes on and on lol

Mike

[jbd5010]

I don't have time right now to read through this whole thread, but I'm glad it's been bumped up so I could catch it... I got a new full HD Canon Vixia camera and a Bell and Howell Lumina MX60 Regular/Super 8mm projector a few months ago to start doing telecine on my dad's home movie super 8 films from when he was younger.

Even with my camera able to do 24p, 30p, and 60i framerates, I couldn't get a good result with an older B&H yard sale projector (Filmosound 485 or something). I'm hoping the MX60 with 3/6/9/18 fps settings and the resources on here will help.

I'm still focusing on VHS tape captures for now but I'll be back to contribute any findings

[VanBerg]

So basically you simply capture the entire film running at 3fps and use the software to process the captured 25fps captured footage...is that correct guys....

That's right, but I found out that the script as I posted it sometimes gets duplicate frames. There is an option in FDecimate to correct this (threshold) which you can set after running it with metrics = true. At the moment my telecine script looks like this:

#DLL calls============================================= =============================
LoadPlugin("d:\videos\Avisynth Restore\script_and_plugins\Decomb.dll")
LoadPlugin("d:\videos\Avisynth Restore\script_and_plugins\Fdecimate.dll")


#PROGRAM========================================== ==================================
Source1 = AviSource (film).ConvertToYV12().Trim (StartFrame, EndFrame)
# YUV: Y = luminence (helderheid)
# UV = chrominance (kleurcomponenten), U blauw en V rood
Source1 = Source1.AssumeBFF()
# DV: Bottom Field First
Source1 = Source1.Telecide(guide = 2, post = 2, back = 1)
# guide = 2, PAL
# post = 2 Deinterlacing with best fitting field from the same or the next frame
# back = 1 If result is stil combed then look back a frame
Source1 = (test == false) ? \
Eval ("""
Source1 = Source1.FDecimate (rate = TelecineRate, threshold = Telecinethreshold, protect = true) \
# Remove surplus of frames
# rate = projectorspeed, though film itself is 25FPS we just need 3 good frames per second
# protect = true to avoid bad frames after scene changes
Source1 = Source1.AssumeFPS (InputFPS) \
# assumefps only sets frame rate parameter
Source1 = Source1.FlipHorizontal().AssumeFrameBased() \
# After deinterlacing: consider film to be progressive
Source1
""") \
: Source1.FDecimate (metrics = true)
Source1

Then when I run it with test = true I get to see the threshold-values between the frames. I then determine which threshold to use and set that in the second running of telecine using test = false.

In the meantime I have received the LED-lamp and am still trying to get the best out of the scripts. There are many filters that clean film, but I couldn't find a filter that removes the larger spots that occasionaly occur on 8mm, so I made my own .dll and I'm testing it right now.
But the main problem to me is still the fact that film has a much wider dynamic range than DV. I capture with a fixed aperture on the camera (9.6 in my case, but that will depend of the lightsource) so that all whites are bright but not washed out.
After capturing I reduce contrast so the film gets very dark and then I stretch the levels. But, as I said, I am still testing: the darker spots stay too dark.

Does anyone have a solution for this problem?

[ronypony]

Hi VanBerg

Many thanks for your reply...I am going to try to use your method of telecine this weekend. I have attempted to run scripts using Avisynth in the past without success ....I could not work out what I was doing wrong...but hey I will give it another try. Once again thanks for sharing.

Mike

[wadiigo]

I have semi-completed my transfer unit. I have modified it, eliminating the condenser lens and using a Raynox DCR-250 macro lens on the camcorder and filming directly off the gate. I have posted a sample video on vimeo; link below if anyone is interested. The film is pretty boring but I'm fairly satisfied with the initial results. Now to figure out Avisynth and further improve the images with filters. This could take some time as avisynth is not sinking in.

http://vimeo.com/20970468

[ronypony]

Hey not too bad at all...maybe a little soft...could you sharpen the focus at all ?

You could use a sharpening filter in VirtualDub...but I think its best if you can get as sharp a capture as possible.

I achieved some half decent results using a demo version of 'Neat Video'...here http://vimeo.com/3256250

Download a demo here...http://www.neatvideo.com/

Did you use CineCap for the frame by frame capture ?

Could you possibly post a photo of your setup...I'd like to see it.

Good job though.

[sjelavic]

Hi,
I also capture directly from the film by using canon 100mm macro lens with extentiuon tubes. I run at 3fps into the EP1 Olympus camera with 8gbyte sd card in DH video mode. Then I decompose the video into tiff files by using blaze media sw. I run my own tiffdiff program to remove the duplicates and then combine the tiff files into avi by using Magix Pro. Here is a clip:
http://dl.dropbox.com/u/5667638/sample2.wmv

[millerandmiller]

Hi, this is something I have been working on for a couple of years. It's a passable realtime telecine directly captured to the Canon 5Dmk2 sensor. (work in progress, as they all are)

Projector: Eumig Mark 501
Camera: Canon 5DMKII



See video description for method.
http://www.vimeo.com/20950590

Results: Videos I transfered using this method:
http://www.vimeo.com/20900718
http://www.vimeo.com/20871186
http://www.vimeo.com/groups/84271/videos/21086292

Cheers, James

[ronypony]

These are fantastic results James...you have developed what looks to me to be a very reliable and speedy method of transferring your cine films. Nice job and thank you for sharing with us.

Mike

[VanBerg]

Hi,
I also capture directly from the film by using canon 100mm macro lens with extentiuon tubes. I run at 3fps into the EP1 Olympus camera with 8gbyte sd card in DH video mode. Then I decompose the video into tiff files by using blaze media sw. I run my own tiffdiff program to remove the duplicates and then combine the tiff files into avi by using Magix Pro. Here is a clip:
http://dl.dropbox.com/u/5667638/sample2.wmv

That is a wonderful result. You don't seem to have any of the dynamics-problems that I encounter: The light parts are not overexposed and you can still see everything in the shades. Did you do some postprocessing or is it the camera?

[sjelavic]

Thanks, nice to get some positive feedback after working on the setup for 2-3 years now. No, I do not do any special post-processing except for duplicate frame elimination, which I do by using my own program (willing to share it). The camera does a pretty good job. I am using Olympus EP1 pen camera in HD video mode with manual focus and ISO set to 100. The apperture is set around 4 I believe but will double check. The lens used is a 100mm macro lens with 200mm extention tubing which gives 3X magnification. 3X is needed since the film is 7mm diameter and the camera sensor is 18mm. The led light I used is one of those video led lights, such as:
http://cgi.ebay.com/Sima-Universal-LED-Video-Light-/230595257101?pt=LH_DefaultDomain_0...item35b08c630d
I added a diffuser in front of it. This particular product gives a "warm" light so I keep the white balance at auto. The white led array I used before gave me "cold" images which did not like too much. Here is the picture of the setup.
Regards,
Stan

[sjelavic]

Hi, this is something I have been working on for a couple of years. It's a passable realtime telecine directly captured to the Canon 5Dmk2 sensor. (work in progress, as they all are)

Projector: Eumig Mark 501
Camera: Canon 5DMKII



See video description for method.
http://www.vimeo.com/20950590

Results: Videos I transfered using this method:
http://www.vimeo.com/20900718
http://www.vimeo.com/20871186
http://www.vimeo.com/groups/84271/videos/21086292
.
Cheers, James

Hi James,
Beautiful videos... I really like the quality of your transfer. You used good optics and a professional camera. You have a similar setup that I used a while back. I also removed one of the projector lenses and left the 20mm lens in between the projector and the camera. The camera lens was very close to the projector, way beyond the minimum focusing distance. That is why the optics are adjusted such that the camera sees a "virtual" image to the left of the projector. This worked for me pretty good but the problem was with some optical distortion due to the fact that the lenses are very close to the film. Some areas are in extreme sharp focus and some a bit blurred. I saw that a bit in some of your scenes but not sure if that was optical distortion or just film age. This is why I switched to macro lens becuse in my cse the distortion was more pronounced.
Overall great job - congratulations,

Stan

PS. Here is my older setup for virtual image cpture

[sjelavic]

Hi James,
I think I missunderstood what you said. I thought that you kept the camera lens on, but looks like you removed it and all that you have is a single lens between the film and the camera sensor, similar to the professional unit like this one in the youtube video:
http://www.youtube.com/watch?v=P6CnbYpmmnc&feature=related
The only issue with the setup that I can think of could be the lens and its working range. I will get back to you on this with more details.
Stan

[jbd5010]

Nice job getting your work on Gizmodo, James! http://gizmodo.com/#!5782892/how-to-transfer-8mm-footage-to-a-canon-dslr-in-real-time

[Khaver]

Here's a utility for extracting a film capture using a camcorder and a projector with the shutter removed or a one blade shutter and running at a slow speed (4-10fps). Camcorder should be 50i/60i with a shutter speed around 1/100 sec.

It's a compiled AutoIT program. It uses avisynth and some plugins, ffdshow and ffmpeg. Docs included.

[sjelavic]

Thanks for sharing the utility. This is very interesting. My program works in a very similar fashion. There are some differences in the way the missed frames are handled - see the enclosed document. The document also contains the link where to get the progam and the source code.

[Khaver]

Thank you too, sjelavic. I'm still fine tuning my program, like instead of choosing the middle frame of the ones with the low metric, it will choose the frame with the lowest metric. I've also fixed a few bugs. I will post the new program here and the source autoit script soon.

[sjelavic]

I am using the middle frame also. If the frame is missed, there is a running frame count that forces a frame selection. The frame selected is offeset by one frame from the last known good frame. This helps with dark scenes where the threshold range is very narrow. I was also thinking of implementing auto threshold selection based on the running averages but never had a chance to get around to it

[Khaver]

My utility runs through the entire video in an analysis pass that uses the GetDups avisynth plugin to write out the metric for each frame in a text file, then it reads through this file and finds the cadence of the good and bad frames. It does this, like yours, so when there's a dark or white section it will still pick out a frame so the timing of the film won't change. It then writes a new text file with the mapped good frames that the ReMapFrames plugin uses to create the new video. Thanks for the suggestion about auto threshold selection. I might look into that for my utility.

[sjelavic]

Thanks Kaver, your approach is faster I believe because it does not include the decomp steps where the film is broken down into individual tiffs and then filtered and then reassembled back into an avi file. I will give it a try.

[JeffCostello]

Hi all,
I'm a college student working on my senior thesis which I'll be shooting on Super 8. I've been building my own telecine machine over the past few weeks using an old Chinon 6100 projector. Not the best quality item but it seems to be working decently. I've managed to slow the motor (still the original built-in motor) down to 4 fps, I've removed the shutter, and replaced the bulb with a 3W 700mA LumiLED with a small piece of opal glass between it and the film gate. I've ordered a lens (a 5" double convex) and I'm planning on recording the floating frame from that using an HD Panasonic, capturing through capturemate and a mouse trigger.
I'm having trouble designing my sync-mouse setup. I'm fairly new to circuitry and don't know exactly how to proceed. My plan is to use a TCST 1000 optical sensor which will run to the mouse, activating the switch and hitting capture in the computer (as discussed at length in much earlier posts). The sensor will be activated by a circular piece of PCB with a single notch cut out which will be attached to the rotator shaft which drives the pulldown. It should activate once with each new frame. I'd like to run the power to the sensor from the USB (which is rated at 5V on the mouse), as an earlier post suggested, but can't find any specs on the mouse I have (which is an old Dell rollerwheel). I'm terrified at the idea of frying my computer by bad connections (if that is indeed possible). Any advice on how to proceed?? I have to order most of my components offline as I don't have any shops other than radioshack around, and time is of the essence.


Any help is greatly appreciated.

Also,
is there any way to use the optical sensors on the mouse? There are two (one for y axis and one for x axis movement). Seems like there might be a shortcut there.

[Khaver]

Here's a new version of FilmExtractor. A few additions and some bug fixes. I also included the full autoit3 script for your use.

FilmExtractor v1.1 - Used for extracting progressive frames from a camcorder capture of film projected at a slow speed using a projector without a shutter (aka the poor mans telecine).

[boobootrasher]

Hi, Booboo back again. I am very happy with the mechanics of my new set up. The Bell & Howell projector has preformed very well with the tests that I managed to do over the Bank holiday weekend. The only issues that I have now are with the optics. Again, I am using an old 10 x zoom camera with an appropriate condenser lense.

I am so happy with Bell & Howell now that I would like to offer my Eumig 610 D for sale as used at the beginning of this thread. Let me know if anyone on here is interested.
I will post tests soon on my youtube.

[N3XUS Transfer Systems]

Just wanted to throw out our results. We've been working on a frame by frame capture system for about a year now. We've been working on this slowly as non of us have adequate time for our hobbies Lemme know if you have any questions. We're really wanting to get this thing buttoned up and manufactured to sell for around $4K - $6K US. I believe if us film nuts had a way to transfer film ourselves we'd be shooting a lot more of it!


http://vimeo.com/23290192

Justin

[videobill27]

HJi...While I am no expert, I think that you might be able to remove the little fan at the end of the motor, extend the shaft outside of the projector body and connect another variable speed motor.
...Bill

[N3XUS Transfer Systems]

Bill,

We have actually considered that. Unfortunately with our current set up the projector isn't pin registered. We will more than likely be engineering a system that looks more like a Spirit. Thanks for the suggestion!

Justin

[videobill27]

HJi...While I am no expert, I think that you might be able to remove the little fan at the end of the motor, extend the shaft outside of the projector body and connect another variable speed motor.
...Bill

[videobill27]

HJi...While I am no expert, I think that you might be able to remove the little fan at the end of the motor, extend the shaft outside of the projector body and connect another variable speed motor.
...Bill

[videobill27]

Hi...While I am no expert, It looks like you could remove the small fan at the end of the motor shaft, extend the shaft to the rear of the projector and attach another motor. It could be a variable speed DC motor or a synchronous AC motor. If you use the Synchronous AC motor you would need a couple of pulleys and a toothed belt to get the correct speed. Remember to disconnect the power to the built-in motor!
...Bill

[boffee]

I have the Eumig 610D with the Hall effect device detecting a magnet fitted to the inhibitor arm. I have connected the output to Frank's Cine2Digits circuit shown on http://www.cine2digits.co.uk/mdc1.htm
I have connected a led across the mouse switch which is flashing 3 times a second when I run the projector at that speed, so seems I am detecting the frames. I have removed the shutter blades from the projector. As Cinecap is not available, I am using Stopmotion Station to capture the frames, however I cannot adjust the potentiometer on the input of the circuit enough to eliminate all the pull down frame blur. I have changed the capacitor to 220nf across pins 1 &2 of the IC. I seem to be getting a blurred frame every so often, it is difficult to tell how often, maybe every 8 frames or so. I was wondering if anyone has a better circuit or have any suggestions.