Cheap 35mm Scanner

Rolfe Klement · February 3, 2007

Why not make a cheap film scanner that contains a CMOS chip

Something like a Nikoin coolscan but for longer runs of footage (1000ft)

It doesn't need to run at anywhere near realtime- but you could load it and let it run over night - it would write to the huge cheap firewire drives available now

The big difference would be the fact that once the film was transferred to harddrive or RAID then modern PC's or MAcs could crunch and edit - as opposed to Da Vinci etc doing it all realtime

You would have an option to scan at variable resolutions (up to 10Mb - since there are cheap 10mb chips out there) - write to various formats (DPX etc)

Firewire 800 or USB 2 connecter on the back

I am sure someone could even mutilate a Nikon cool scan to do it now - basically the drive mechanism and auto file writing need a bit of work

It would mean anyone could originate on Film Stock but go digital easy

what do you think? Has anyone done this before?

thanks

Rolfe

Stephen Williams · February 3, 2007

Why not make a cheap film scanner that contains a CMOS chip

Something like a Nikoin coolscan but for longer runs of footage (1000ft)

It doesn't need to run at anywhere near realtime- but you could load it and let it run over night - it would write to the huge cheap firewire drives available now

The big difference would be the fact that once the film was transferred to harddrive or RAID then modern PC's or MAcs could crunch and edit - as opposed to Da Vinci etc doing it all realtime

You would have an option to scan at variable resolutions (up to 10Mb - since there are cheap 10mb chips out there) - write to various formats (DPX etc)

Firewire 800 or USB 2 connecter on the back

I am sure someone could even mutilate a Nikon cool scan to do it now - basically the drive mechanism and auto file writing need a bit of work

It would mean anyone could originate on Film Stock but go digital easy

what do you think? Has anyone done this before?

thanks

Rolfe

Hi Rolfe,

I know someone doing this with s8 & 16mm, using a DSLR.

Stephen

Alex Haspel · February 3, 2007

i suppose the missing registration is an issue not easy to deal with.

Troy Warr · February 3, 2007

People have been talking about this topic in a recent reduser.net thread.

They mention one particular setup that uses Super 16mm, and I don't see why you couldn't build a similar device for 35mm. Still, I don't think that people who can afford to shoot and develop 35mm film would have use for a "cheap" scanner. Unless the results are comparable to a professional transfer (which I doubt they could be) to maximize image quality, why shoot 35mm in the first place? It does seem like an interesting option as an alternative to Super 16mm or Super-8 telecine, though, and after a handful of transfer sessions, it should pretty much pay for itself (assuming that you're able to perform the work of a colorist).

Dennis Kisilyov · February 3, 2007

Dude, get a M42 slide duplicator setup (cheap on ebay) a M42 to Canon EOS adapter, a 50mm prime lens.

Grab a cheap 35mm projector with a pause button and a decent light.

I would not suggest using negatives.

P.S. Last time I checked Sprit Datacine did not have a reg pin either, (I may be very wrong about this.)

Matthew Buick · February 3, 2007

Wouldn't that just be ''videoing'' the film? If so, not good.

Nate Downes · February 3, 2007

i suppose the missing registration is an issue not easy to deal with.

Actually if you used a linear scanner it would be super-easy. Using the intermittant, yes, it would be a bit of an issue.

The nikon scanner is a linear scanner, not an intermittant. And a linear scanner would be cheaper to produce, but be dramatically slower. We're talking 8-10 hours to scan roughly 10 minutes of footage.

Dennis Kisilyov · February 4, 2007

Wouldn't that just be ''videoing'' the film? If so, not good.

Nono, pause button. You'll grab stills of each frame with a DSLR - I don't think they make slide duplication units for new school digital cams. But the setup someone else posted for 7k is a slide duplicator bellows setup.

I think for 35mm you'd need a 35mm lens not a 50mm prime though.

Light has to pass through the negative always, and there is always a lens involved, even in Nikon Cool Scan.

Edited February 4, 2007 by Dennis Kisilyov

Matt Pacini · February 9, 2007

Nono, pause button..."

I've been working on a 16mm/S16 scanner for several years now.

There are MANY problems to overcome that make this infinitely more complicated than what's being talked about here (and on most forums I hear when this idea comes up), but the flatbed or Nikon scanner idea just simply doesn't work.

To address this one thing, the "pause button" idea doesn't wash - it's gotta be automated, or you will spend literally 60-100 hours a week for 4-12 months digitizing a film if you were to actually do it this way.

Do the math. I did.

Figure 7-15 seconds a frame, then you have to save them out, have them sequencially numbered, etc .,etc. Multiply that by 24fps, by 60 (per minute) by 60 (per hour).

See what I mean? The whole "hey, just use a scanner!" idea gets pretty silly when you factor the time/labor in.

What I'm battling with now, is the software controller aspect of it; I've got all the hardware mostly built, but the control, sensing with feedback loops, autosaving & converting of files, etc., is the real bitch of this endeavor.

MP

Dennis Kisilyov · February 10, 2007

I've been working on a 16mm/S16 scanner for several years now.
There are MANY problems to overcome that make this infinitely more complicated than what's being talked about here (and on most forums I hear when this idea comes up), but the flatbed or Nikon scanner idea just simply doesn't work.

To address this one thing, the "pause button" idea doesn't wash - it's gotta be automated, or you will spend literally 60-100 hours a week for 4-12 months digitizing a film if you were to actually do it this way.

Do the math. I did.

Figure 7-15 seconds a frame, then you have to save them out, have them sequencially numbered, etc .,etc. Multiply that by 24fps, by 60 (per minute) by 60 (per hour).

See what I mean? The whole "hey, just use a scanner!" idea gets pretty silly when you factor the time/labor in.

What I'm battling with now, is the software controller aspect of it; I've got all the hardware mostly built, but the control, sensing with feedback loops, autosaving & converting of files, etc., is the real bitch of this endeavor.

MP

A pause button or a stepping projector would allow for a parralel port relay to be attached (Drivers developed for Linux Coffee), If I were actually doing this, I'd do it on Linux, which is open enough to allow to mod Gimp and drivers for multi frame acquisition.

Converting a set of still tiffs at 24fps into a single QT file is rather easy.

And Gimp has been used in the motion picture industry for a while to do frame by fame polishing in sets.

My biggest fear is flicker. Since the F-stop on a DSLR is auto controlled typically the lens will stop-down at somewhat different places. Also an inferior light source can - (a. burn the film) and (b. provide inconsistent exposure frame by frame.)

The DSLR would run of the PICT bridge protocol, this way each frame would get dumped right into the machine, in lieu of using a flash card storage.

Depending on the DSLR elected the pict bridge can also trip the shutter.

So 1 frame step on the film via relay, 1 exposure on DSLR, Store Frame on HD - Advance frame on projector.

That program loop can run as a simple script and will go through a 100ft day roll faster than manual loading/cutting and scanning with Cool scan or the like.

P.S. Matt if you need any programming help with your project, let me know. I may be able to...

Edited February 10, 2007 by Dennis Kisilyov