Having Trouble with DIY Telecine (8mm) System

[badener]

Hello I'm back again,

in the meantime I got an El Nikkor 50mm lens and the Lifecam Studio.
The webcam is modified (lens and autofocus removed), but not finished yet, but I could do some tests.
Here are two pictures, to show my work.

Now I'm looking for capture software (found video capturix for test).
Which one do you prefer and what is your favourite resolution for capturing?

Best regards,

Siegfried

[Leshka71]

Hi Siegfried,
Welcome back I prefer to use CineCap. It is really good and it's building AVI file on fly.

Regards,

Alex.

[xrkund]

I hate to resurrect an old thread, but I've finally gotten around to transferring my film, and I'm running into a few issues, mostly optical.

Allow me to describe my setup.


I've got a 1960 Bell & Howell projector (at work right now, can't remember the exact model) but it looks identical to Leshka's, except it does not have variable speed control. The bearings in the original motor were shot, and I'll be replacing it with a DC motor with variable speed to control my framerate.


I've removed the original lamp and plan on using an LED, with I'll dim with a pulse width modulator (identical to the motor controller). I'm also planning on using opal diffusion when I get it all figured out.


I say "planning" a lot because I'm attempting to mock it all up before finalizing my parts list. Currently the projector setup is no motor with an LED flashlight and wax paper diffusion, but that's beside the point.


As for camera and capture, I'd like to avoid the optical sensor, frame-by-frame route, as I believe its a little too involved for me. I'll be using a Canon DSLR in video mode for capture. The variable speed DC motor will allow me to synchronize it with the framerate of the camera. I plan on using the Magic Lantern firmware to more easily adjust camera framerate, shutter speed, and ISO.


I've got a 50mm Canon FD mount lens with manual aperture and somewhere around 42mm of extension tubes. The projector lens is 15mm.


I've gotten very, very close to a usable image with the 50mm lens and all of the extension tubes placed about 1.5" away from the projector lens. It seems that I need to be about .25in closer to the lens for best focus, but it is physically impossible to get it closer, considering the projector lens is flush against the body of the projector. It also seems that even if I was able to get the lens closer, the curvature of the projector lens would still be cropping some of the image (on an unmodified gate).


I've contemplated the condenser lens route, but don't really know what to buy, how big, flat on one side, curved on both sides, etc.


Has anybody used a DSLR on here for a transfer? I've read elsewhere that people have and have seen good results, but never see a detailed enough description of the lens setup. I believe one problem I have with image size is that the sensor on a DSLR is significantly larger than a webcam, so it requires more magnification.


If anyone has any tips or suggestions, I sure would appreciate it.

[wadiigo]

xrkund...check out the attached link. It may or may not help but its real interesting to watch.

https://vimeo.com/20950590

wadiigo

[xrkund]

Thanks for the reply! I have seen this one. Is a very impressive tender. There are some things I'm not clear on in his setup.

Was the projector lens removed? Was the disassembled 35mm lens used in conjunction with another lens on the camera, or was that it? In using an APS-C sensor, not full frame, how is that going to affect getting the entire image?


I'll need to experiment some more, but my best results were with the projector lens, the 50mm, and all of my extension tubes.

[wadiigo]

I don't know how many still follow this thread but for those that do, you might find this interesting.

http://kinograph.cc/

[xrkund]

I've been following this as well. They just updated their website with a little more information, timeline, etc. They do have some sample footage, which looks really, really good, however, it is still going to cost $1500+ to build one.

[basil3w]

Does anyone know if Avi Synth or CiniCap will run on a Windows 8.1 machine?
Thanks,
basil3w

[VanBerg]

Has anybody used a DSLR on here for a transfer? I've read elsewhere that people have and have seen good results, but never see a detailed enough description of the lens setup. I believe one problem I have with image size is that the sensor on a DSLR is significantly larger than a webcam, so it requires more magnification.

Hallo,

I was on this forum, on this thread some years ago when I started to digitize my Super 8 and 8mm. I then stopped because my camera was only SD and the result was not good enough for me, especially on the dynamic range. I wanted to wait till I had an HD camera.
In a family with young children dad's hobby comes last but finally I took up where I had stopped.

I am now going to transfer my films using a Canon EOS 700D so and the first results seem fine.

In short:
I use a Eumig 610D with LED lighting. I have the shutters removed and film at 6fps.
The projector lens has been replaced by an enlarger lens: a Schneider Componar-S 4.5/80. The long focus gives me enough room for all the equipment. The diafragma in the lens lets me control the light.
The lens of my Canon is removed (to protect the mirror I put a macro-extension ring on the camera on which I fixed a clear filter)
I film at 24fps at a resolution of 1920x1080, at iso 200 and shutterspeed 1/50, with a picture style as flat as possible to get the most out of the dynamic range: Marvels Cine.

Because I removed the shutterblades the resulting film contains duplicate frames and frames on which the projector advances the film. In Avisynth I made a function that isolates the frames I need.
After that I can make all the colour and dynamic range corrections I want.

The next few weeks I will be going on with this. If you would like to know more, just let me know.

regards,
Frans

[Khaver]

I'd be interested in seeing your avisynth script. I do something similar to you but uses a 2 step process where in the first step an avisynth script compares the current and previous frame an writes out a text file that shows the difference between the frames. I then use a utility that uses the text file to isolate the desired frames and it writes out a new text file that the Remapframes.dll avisynth plugin can use. If you've found a way to do all this in one step that would be fantastic.

[VanBerg]

It's not a script, it's a .dll I made. Since I started this a few years ago I have only tested it against the version of Avisynth I use, which is version 2.58.

I made this because I am better at software than at hardware. Microswitches and alike are really not for me, so I found it easier just to continuously film and get the still frames later.

This is how I use it:

LoadPlugin("path\8mmtools.dll")
Workclip = FilmUnduplicate (Workclip, rate = TelecineRate, threshold = TelecineThreshold)
# Remove surplus of frames
# rate = projectorspeed (in my case 6fps)


In order to find the right frame and not the one advancing the film it is necessary for this function to find multiple frames of the still image. Which is why the fps-rate of the projector may not be more than 1/3 of the camera fps-rate. In that case there are three frames per still image and the function can find one of the two that don't include the moving of the film.
I have now tried them on two 3-minute reels and have found no errors yet.

In the .dll there are two more functions:
-SpotRemove, which removes spots and dirt
-FilmLevelScale, which is way of adjusting the levels a bit more flexible.
But I have to say: it has been a few years since I looked at these two, so I have to look up how to use them...

[Khaver]

VanBerg, thanks for uploading this. I will try it with some test clips I have that were also taken from a capture of video projected using a projector without a shutter.

How do you determine the threshold to add to the parameters? I've found using my script that getting the threshold correct so my utility can find the right frames to use when the exposure of the film varies from underexposed to normal to overexposed at different places can be difficult. How well does your plugin work when the exposure varies? And what happens when the film has several seconds of all black film or all white film?

Again, thanks for providing your plugin.

[VanBerg]

Well, it still is a work in progress. I hope it can be of use to you.

When filming without shutter there is a very big difference between still frames and the one that contains the moving of the film. That is what my program looks for. Frames that are very different to the previous and the next one are considered to be of no use, it's the ones that have no great difference that we are looking for: they must be the ones containg the still image. So all my program does is taking one of the frames without difference inbetween the very different ones. (sorry, english is not my mother tongue) When the program cannot make up his mind, that means there are no differences whatsoever, I let it take a random frame out of the set.

At the moment the films I have tested are normal exposed up to a little underexposed. But it may well be that my program has to be changed when I encounter problems with films. I found that white strokes of film do not pose a problem and I expect black strokes won't be a problem too, but I have had none of them yet.

If you find the program of use but have problems with some specific films, you are welcome to let me know so that I can try and solve it.

[Khaver]

I've created a program that analyzes the text file generated by Avisynth's internal Compare function when fed a video file from a camcorder that captures the screen from a projector with it's shutter removed and running around 6 fps. It first runs the whole video file (or a selected region) using a modified version of avs2avi or ffmpeg to generate the Compare text file. It then analyzes the text file and picks out the best frame (the one that's has the least differences from adjacent frames) and recreates a new Avisynth script on the fly using StainlessS's FrameSel.dll plugin, that then gets run by the modified avs2avi utility or ffmpeg to be encoded to a new, progressive, flicker free, video file or image sequence. You can add any Avisynth function or plugin to pre and/or post process the video before analyzing and encoding. You can also have it just generate the text and Avisynth scripts files so you can make your final video using an encoder of your choice that accepts avs scripts.

Though I'm not the originator of this method of a "poor-mans-telecine", I've been working on this for a number of years and finally got it to a point where it works pretty much flawlessly when fed fairly normal exposed film captures (it has settings to compensate for under or over exposed movies). It does not require any threshold setting so it's basically load-and-run.

This will be a low cost commercial software program ($30-$40 USD) but a trial is available now for download at cinextractor.blogspot.com

Though this post can probably be construed as a shameless plug for commercial software, it does tie into this forum thread and if you go back through it, you'll see that I have discussed this method in the past. If you want, you can cut up your video captures into 1 minute or less segments and use the trial version forever without spending a dime.

[Khaver]

I've decided to release a freeware command line version of my Cine X-Tractor utility. It's for extracting progressive frames from captures of films projected using a projector with the shutter disabled or removed, a.k.a the John Meyer method. Below is a printout of the readme file.

Cine X-Tractor (Command Line) 1.0.0 by khaver (c)2014
      khaver(at)netzero(dot)net

Note: XTractorCL.exe is based  on the source  code of AVSMeter 1.7.9.  AVSMeter
      is a freeware  command line utility  for benchmarking  Avisynth  scripts.
      Since AVSMeter is freeware,  Cine X-Tractor Command Line  utility is also
      being  released  as  freeware  with  the  following  restrictions.   Cine
      X-Tractor Command Line  (XTractorCL.exe)  can be  used  to  process video
      captures of projections  of your own personal films,  and any other films
      to  which  you  own  the rights.  You  may  also use  it to process video
      projections  of a 3rd party's  personal films  or films to  which the 3rd
      party  owns  the  rights,  either  pro bono  or  for profit.  YOU MAY NOT
      include  the XTractorCL.exe  executable  file as  part of  any commercial
      software package.  YOU MAY NOT charge anyone or any entity for the use of
      the Cine X-Tractor Command Line utility.
      
      For information on AVSMeter, see the following:
      http://forum.doom9.org/showpost.php?p=1585008&postcount=1

Cine X-Tractor (XTractorCL.exe)  is a command line utility  that inputs a video
file  capture  of 8mm,  Super 8  or 16mm  that  was  projected  at a frames per
second  rate  of at  least 1/3rd the  capture  rate using  a projector with the
shutter  removed or  disabled.  This  technique  is  based  on the work of John
Meyer who regularly posts in the doom9.org forums.  

See his post: http://forum.doom9.org/showpost.php?p=1531961&postcount=1

Software Requirements: 
      The Cine X-Tractor executable file, XTractorCL.exe (included)
      StainlessS's FrameSel AVISYNTH plugin (included)
      The open source video and audio frame-server, AVISYNTH (included)
      Any video encoder capable of importing Avisynth .avs scripts

Optional Software:
      Additional Avisynth plugins for pre and/or post processing video
      Additional Avisynth plugins for decoding various video formats
      Video player to preview your captured and converted clips
      
Included Files:
      XTractorCL.exe - The Cine X-Tractor command line executable file
      FrameSel.dll - StainlessS's Frame Selector AVISYNTH plugin
      Avisynth_258.exe - 32bit AVISYNTH installation file
      default.avt - An example AVISYNTH template file

Since  this is a command line  utility it does not  required any  installation.
Just  place all the files  in a folder that's easily  accessed from the command
prompt.  You will  need to  have the  32bit version of  Avisynth  installed. If
you don't  already  have  Avisynth  installed,  run the  included  installation
executable (Avisynth_258.exe)  as administrator  and complete the installation.
Once this is done, you are ready to use Cine X-Tractor.

First, a little background on how this works.

Avisynth,  the open  source  video and audio  frame-server,  has the ability to
measure  the  differences  between  adjacent  frames  in a video  file.  If the
difference in brightness, color, movement, etc. is great, a large metric number
is generated  for that frame.  If the  difference is slight,  the metric number
generated  is small.  Keeping track of  these metric  numbers for each frame is
the key.  As the  video file  frames  are sent  through Avisynth,  these metric
numbers, paired with the corresponding frame numbers,  can be written to a text
file  that  can be later  analyzed.  Cine X-Tractor  first creates  an Avisynth
script that includes the built in function, "Compare", which it then loads into
Avisynth  and internally plays it at the fastest frame rate possible,  with the
"Compare" function  outputting  a  metrics text  file  based on the differences
between  adjacent  frames.  It then loads this metrics text file,  analyzes it,
and outputs a new Avisynth script, that when loaded into a video encoder,  will
will produce a video containing only the "clean" captured frames.

So,  how does this  help us  in our quest  for true  progressive,  flicker free
captures  of  our 8mm,  Super8,  and  16mm  films?  First,  you need  a working
projector  that can  play your films  at a  slow speed.  If you don't currently
have  a projector,  garage  sales, flea markets  or charity  stores  are a good
source.  Once  you have  one that can  play the  films at  a slower than normal
speed,  you'll need to open it up and  remove or disable the shutter*.  This is
crucial  because  we want the  camera to capture  the frames  as  they're being
pulled  down in the projector  gate  without the  interference  of the shutter. 
This way your  video camera will  capture multiple,  flicker  free frames while
the film  is at rest  in the projector  gate,  and a few blurry,  moving frames
while it is  being pulled down  to the next  frame.  Those  frames  that are at
rest will generate  low metric numbers by Avisynth.  The blurry,  moving frames
will have high metric  numbers because each frame  will be quite different from
the adjacent frames.  By analyzing the metrics text file generated by Avisynth,
Cine X-Tractor  is able to choose the best  frame from the group of frames that
have low metric numbers,  and send them, one by one, to the video encoder.  The
result... clean, progressive video captures of your movies!

*Warning: do not disable the film pulldown mechanism.

For the best  results it  is recommended  to use a matte white  poster board on
which to project your movies,  with the video  camera as close to the projector
as possible  to minimize  perspective  distortion  of the captured image (there
are  Avisynth  plugins  that  can  correct  perspective  distortion).   A  good
location  is directly  under the  projector lens.  Locate  the poster  board so
the projected  image is sharp  and moderately  bright.  Too close and the image
will be  overly bright  with a  hot spot,  and  result in a  highly  contrasted
image with  the video  camera unable  to capture the  darker portions.  Too far
and  the image  will not have  enough  contrast.  Experiment  before  you begin
your  main  capture  session.   You  can  use  one  of  those  rear  projection
contraptions  sold as a  DIY film to video  conversion tool,  but these tend to
have grainy  screens with  a hot spot in  the  center.  The  poster  board in a
completely  dark room  will give you an  excellent image.  There are other more
complicated  projector/camera  setups that have the potential of providing even
sharper  and more  vibrant  colors.  It's  just a matter  of how much  time and
energy you want  to put into it.  An additional measure to ensure good captures
is to  use  as  many  manual  controls  on  your  video  camera  as  it allows,
especially focus and turning off auto-exposure.

Once  you've  captured  some  film  projections  and  transferred  them to your
computer, you can begin using Cine X-Tractor.

From  a  command  prompt,  enter  the  folder  where  you  installed  the  Cine
X-Tractor executable (XTractorCL.exe).  Type  "xtractorcl" to  view the various
parameters and options.

Usage: XTractorCL (videofile|@filelist) [switches]
       videofile = path to single video file
       @filelist = path to text file containing list of video files
Switches:
-h [0|1]      -h or -h 0, display short help.  -h 1, display full help
-f float      Output fps (18)
-r start,end  Subframe range (avisynth trim sytax) (all frames)
-c float      Cadence (-1 = calculate) (calculate)
-n int        Stats interval (0 = calculate) (10)
-s 0|1        Avisynth source filter=Directshow|template source (0)
-t file.avt   Template file
-p  0|1|2|3   Processing=none|pre|post|pre&post (0)
-e -1|0|1|2   Exposure=under|normal|over|auto (0)
-d            Delete temporary files
-l            Create log (CXT-date_time.log)
-x textfile   A text file containing a single command line that
              will be run for each generated output .avs file.
                Use < as the placeholder for the generated .avs file.
                Use > as the placeholder for the output file.
              Output file will be the name of the video file without
              the extension. Use quotes where appropriate.
              Example: "C:\Apps\x264.exe" --crf 24 -o "D:\encodes\>.mp4" "<"

Cine X-Tractor needs  one or  more video files  to process.  For a single file,
type the full path to the video file, using quotes if the path contains spaces.
If you want to  batch process  several video  files at  once, use @ immediately
followed by the  path to a text  file that contains the path to each video file
on separate  lines.  If the  path  to the  text file  has spaces, enclose it in
quotes.  The paths  of the  video  files  listed  in the text  file do not need
quotes.  The various  switches and parameters  available  to single  video clip
processing are also available, on a per clip basis in the batch list, by adding
additional values  after the  path to the  video  file  separated  by "|".  Any
parameters not included  will use the values  added to the  command line or the
default values.  These values must be added in the following order:

videofile|fps|start|end|exposure|source|cadence|processing|templatefile

See the command line switch definitions below for possible values.

-f    (fps) A floating point number  representing the actual  frame rate of the
      film.  This is usually 16 for 8mm films, 18 for Super 8, and 24 for 16mm.
      This will be the frame rate of your final  encoded video.  The default is
      18.

-r    (start,end) Integers that indicate to process  beginning at frame "start"
      and end at frame "end".  Entering 0 (zero) for the  end frame  means  "to
      the end of the clip".  The default is all frames.

-c    (cadence) A floating  point number  representing  the ratio  of the frame
      rate  of your  capture  device (camcorder), to  the  frame  rate  of your
      projector.  If your camcorder  captures  at  29.97fps and your  projector
      projects at 6fps, the cadence  value would  be 4.995.  This value is used
      to select the correct number of frames during sections of complete  black
      or other  solid  colors  in your film.  It's  also  used  to  verify  the
      accuracy of the  analysis  of the metric  file.  If  the total  number of
      frames of the output  file is not close  to the total number of frames of
      the source  clip  divided  by  the  cadence  number  than  Cine X-Tractor
      couldn't process the video accurately  enough and no  final output script
      will be  written.  This  usually  happens  with  underexposed  or  highly
      overexposed  films.  See the  exposure  switch  below.  The default is to
      have Cine X-Tractor  calculate the cadence  from the  metric file itself.
      This is done  in a  matter  of seconds  so it's usually  best to  use the
      default.  A value of -1 will  also  tell  Cine X-Tractor to calculate the
      cadence.

-n    (interval) An integer that changes the stats update interval.  Leave this
      at the default, which is 10, unless you use heavy pre-processing Avisynth
      functions that slow down  processing speed.  Then use a value of 0 (zero)
      to have Cine X-Tractor calculate the interval.  This takes a few seconds,
      but will provide more accurate real-time stats updating.

-s    (source) Either 0 (zero) or 1 (one).  The default,  0, tells  Avisynth to
      use your installed DirectShow  video decoders to decode your video clips.
      A value of 1 tells  Cine X-Tractor  to use  the Avisynth  source function
      found in the provided template file (see -t below).

-t    (templatefile) A path to a  template  file.  The template  file is a text
      file containing additional  Avisynth functions  that can be  added to the
      Avisynth analysis script and the final output Avisynth script.  If the -s
      switch  or the (source)  value in the  batch list is 1 (one), instead  of
      using a DirectShow decoder, Cine X-Tractor will  search the template file
      and look for a line containing "#__source__#".  If it finds one,  it will
      replace the default  DirectShow source  function in the  analysis  script
      with everything on the line after #__source__#.  This must be an Avisynth
      video source function, either an internal one, such as "AviSource" or one
      used by an installed Avisynth  input source  plugin.  If #__source__#  is
      not found, Cine X-Tractor will exit with an error or continue to the next
      file in a batch list.  If there are  more than one lines  pre-pended with
      #__source__#, the last one will be used.

      If the -p switch or  (processing) value is 1  (pre-processing)  or 3 (pre
      and  post  processing), Cine X-Tractor  will  search  the  template  file
      looking for lines pre-pended  with #__pre__#.  All lines  pre-pended with
      this will be added to the end of the analysis script.  This allows you to
      add functions that  may help in a more  accurate  analysis  of the video,
      such as cropping away parts of the captured frame irrelevant to analyzing
      the  film, like  the  film gate.  Another  possibility  would be  if your
      projector/camera setup allows you to capture the sprocket hole along with
      the image.  You could crop away the image leaving just the sprocket hole,
      and the adjacent frame comparison would clearly differentiate between the
      sprocket hole at  rest and  while the  film is  being pulled  down in the
      gate.  This would allow accurate  progressive frame  extraction no matter
      what the exposure of the film.

      If the -p switch or (processing) value is 2 (post processing) or 3,  Cine
      X-Tractor will search the template file looking for lines pre-pended with
      #__post__#.  All lines  pre-pended with this will be  added to the end of
      the final output Avisynth script.  Use these to enhance the output script
      with functions to crop,  resize, flip, de-noise, etc.,  your final video.

      There are many  Avisynth functions,  both internal and plugins,  that you
      can use to improve  the appearance of your films.  Take the time to learn
      the Avisynth scripting language.

      If the template file indicated does not exist,  no pre or post processing
      functions will be added to the analysis or final output Avisynth scripts.

-p    (processing) An integer value of 0, 1, 2 or 3.  A value of 0 (zero),  the
      default, means  no pre or post processing.  A value of 1 (one) means only
      add the pre-processing  functions  in the template  file to the  analysis
      script.  A value of 2 means only add the post processing functions in the
      template file to the  final output  script.  A value of 3 means  use both
      the pre and post processing functions in the template file.  If you use a
      value of 1,  2 or 3, you must also include the -t or (templatefile) path.
      If you don't,  or the  template  file  does not  exist,  no  pre or  post
      processing  functions  will  be added  to the  analysis or  final  output
      Avisynth scripts.

-e    (exposure) An integer value  representing the exposure of your films.  If
      the entirety of  the captured  clip is  underexposed,  use a value of -1.
      What this will do is  brighten and add  more contrast  to the video clip.
      This will help in the analysis of the video.  For normally exposed  film,
      the default value of 0 (zero) should be used.  For overexposed film,  use
      a value of 1.  This  will invert the video (negate it) and then tweak the
      brightness and  contrast.  For  film  that  contains  scenes  of  various
      exposures, a value of 2 will tell Cine X-Tractor to add a function to the
      analysis script that  automatically adjusts  the brightness of each frame
      to a normal level.  This setting  works extremely  well with all types of
      footage, but it does slow down the processing rate nearly by half.

-d    Tells Cine X-Tractor  to delete all the temporary files after processing.

-l    Create a log file in the format CXT-date_time.log.

-x    (textfile) Path to a text file containing a single command line that will
      be run for each generated final Avisynth output script.
          Use < as a placeholder for the path to the Avisynth output script.
          Use > as a placeholder for the output file name.
      This output file name will be the name  of your source video file without
      the path or extension.  Use quotes where appropriate.
          Example: "C:\Apps\x264.exe" --crf 24 -o "D:\encodes\>.mp4" "<"
      In this example, if the source video file is "D:\captures\Super 8-01.mov",
      the command line will be expaned to:
      "C:\Apps\x264.exe" --crf 24 -o "D:\encodes\Super 8-01.mp4" "D:\captures\
      Super 8-01.avs"

[Khaver]

Here's the link to the Cine X-Tractor website where it can also be downloaded from. http://cinextractor.blogspot.com/

[VanBerg]

I got a request about the .dll I posted last year. It seemed I had accidentally posted the version with debug-information instead of the normal release. Strangely enough this only gave problems in Windows 8.1...

So here I post the version without debug-information.

[ghostanno]

Hi VanBerg,

I have a WorkPrinter XP and have been experimenting with different setups, Frame grabbing with a Canon DV Camcorder with okay result.
But also trying to setup a this with a HD camcorder but CineCap and HD is a no GO and there are not many options for WorkPrinter and HD
setups. I did try out the Cine X-Tractor from Khaver on a DV cam recording and did get this to work.

But I also came a cross your 8mmTools.dll what I did not figure out yet is how the AviSynth script is suppose to be

This is how I use it:

LoadPlugin("path\8mmtools.dll")
Workclip = FilmUnduplicate (Workclip, rate = TelecineRate, threshold = TelecineThreshold)
# Remove surplus of frames
# rate = projectorspeed (in my case 6fps)


That did not compute

Do you have complete working example of how you use your dll file a AviSynth example file?

It would be great testing your dll file with a few minutes of HD recording.

Regards
GM