Super-8 scanner based on Dominique Galland's software

1

A random internet find, utilising @dgalland’s software

Some interesting mechanical solutions…

2 Likes
2

Yes, I exchanged a lot of mails with Patrick Sibenaler when he developed his Telecine, he helped me a lot to improve my software. He doesn’t use a modified projector but a standalone kinograph type setup by only recovering the frame feed mechanism from the projector. He thus obtains a precise positioning of the frame. It is an interesting solution but still with a lot of work in 3D printing.

3

Thanks for sharing! I’ve added it to the existing machines list, which anyone can edit at any time. Great find.

4

Thanks Matthew,
I also added a link to my project. This project uses a modified projector and a Raspberry camera, the software is in Python.
Yart project

1 Like
5

Your documentation looks fantastic. I really appreciate the time and effort it takes to document code so I appreciate you sharing it on Github. Do you have any video of it working? We (the forums) would love to see it!

6

I just made a small video of the yart project in action

2 Likes
7

I just put on github a document explaining the design of the yart application. If the hardware is inspired by the ideas of the original rpi-film-capture project of Joe Herman the software is completely rewritten with improvements on the following points

  • A simpler and equally efficient multithreaded design.
  • A better object design for the camera and the motor
  • Higher level methods for network exchanges
  • And many other things !
    All this simplifies considerably this application
    I think this will be of interest to all Python developers.
    My apologies if the automatic translation from French is not optimal.
    https://github.com/dgalland/yart/blob/master/Design_en.md
1 Like
8

My project is now stabilized and I have done all my scans. I’m just doing some more measurements for feedback and conclusions.

  • The HQ camera needs to be complemented by a good lens, in my case a Componon S-50. I had started with a 35mm CCTV of good quality but there was some blur in the corners. Probably it is the same with the official 16mm?
  • For a good result an HDR capture with at least 3 exposures is necessary. Because of the wait between each change the capture rate is greatly reduced.
  • The effective framerate for a jpeg capture is disappointing. With the best python programming I measure about:
    Mode 3 full res 4056x3040 3,3fps and in HDR 7 s/frame
    Mode 2 Binned 2029x1520 13fps and in HDR 1,5s/frame
    Mode 1 1920x1080 19fps and in HDR 1,2s/frame

The consequence is that mode 3 full-res is not really usable with HDR,
The image quality in mode 1 or 2 is still satisfactory, I did all my captures directly in 1500*1080 with a good result but after comparison it would have been very slightly better in full resolution. I have a slight feeling of dissatisfaction not to have been able to exploit the HQ really in HQ!

I have not tried other software solutions and it must be said in this regard that the Raspberry camera software based on MMAL firmware and consequently Picamera Python is stable but without prospects of evolution. Everything is evolving for these cameras and others towards the new libcamera library which is still quite experimental and in any case completely different for the development.

3 Likes
9

Thank you for the excellent documentation and for sharing your progress with us. It’s a great project and I’m glad to see it continuing!