ccd cmos camera for telecine machine

[Richardson Leao]

Hi all,

 

I am converting a stenbeck table 16mm to a telecine machine but i am stuck trying to get a camera for it. First i got a sony with 420 tvl resolution and then i remove the ccd from the case and used a single macro lens between the ccd and the film frame to do 'on the fly' scanning. Well, that was almost good but i believe the resolution of the camera is the drawback of my system. I also tried to use a minidv camera (consumer) to do it and that is also not good and i have to have 2 extra close-up filter to make the system compact enough. Anyway, I would like some opinions in regard a camera to buy. I am very tempted in getting a board camera as they are small and they can be mounted in my custom-made xyz system. But are they any good? I found these with 550 tvl:

 

http://www.rfconcepts.co.uk/board-pinhole-sony.htm

 

but i dunno any thing about board cameras. Also, should i go for ccd or cmos?

 

many thanks!

[Phil Rhodes]

I very much doubt you would be happy with the quality of that - it won't be as good as a decent 3-chip miniDV camera.

 

There's two ways to go for homebrew telecine, as I see it - use a DSLR, or a machine vision camera. The latter might be better as you could get a monochrome one and take sequential exposures with RGB lighting.

 

Neither would necessarily be realtime.

 

P

[Paul Bruening]

Are you shooting the screen or the film plane?

[Richardson Leao]

Are you shooting the screen or the film plane?

 

the film plane.

 

I think the machine vision monochrome idea is great. And which software I'd put the channels together?

[Paul Bruening]

This is an interesting approach. IIRC Steenbecks use continuous transport and isolate the image with a rotating prism. Is that what you have? If so, will you use a fast shutter setting on the camera to capture the film plane images?

[Phil Rhodes]

Many machine vision cameras will use the cameralink format, which either goes direct into a PCI(e) slot or uses firewire. In any case, it's likely that at least a demo application capable of capturing single frames exists.

 

Personally I would not try to capture film in motion - I'd look at it more as a non-realtime scanner, in which you'd step a frame, grab it, step a frame, grab it, etc. Some of the best machine vision cameras don't support their full resolution at 24hz anyway.

 

This becomes even more necessary as you begin to consider RGB backlighting and perhaps double-passes for higher dynamic range, unless you're willing to splurge money on several cameras and space them out down the length of the film.

 

With sufficient resolution, you can obviate registration concerns and stabilise to a sprocket hole in as a post process.

 

P

[Richardson Leao]

Many machine vision cameras will use the cameralink format, which either goes direct into a PCI(e) slot or uses firewire. In any case, it's likely that at least a demo application capable of capturing single frames exists.

 

Personally I would not try to capture film in motion - I'd look at it more as a non-realtime scanner, in which you'd step a frame, grab it, step a frame, grab it, etc. Some of the best machine vision cameras don't support their full resolution at 24hz anyway.

 

This becomes even more necessary as you begin to consider RGB backlighting and perhaps double-passes for higher dynamic range, unless you're willing to splurge money on several cameras and space them out down the length of the film.

 

With sufficient resolution, you can obviate registration concerns and stabilise to a sprocket hole in as a post process.

 

P

 

i didn't add a shutter as the spinning prism does the job (in some angles, no light passes straight to the ccd. maybe i'll try until the perfect camera appears. I was more thinking in using the table to cut the film before doing HD, specially if I get a camera that could invert the image (negative). Then I could cut before scanning. But thanks for all the replies!