Film-Tech Forum ARCHIVE: The Kinograph

This is topic The Kinograph in forum Film Handlers' Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=011478

Posted by Brad Miller (Member # 2) on 06-18-2013, 02:48 AM:

Kinograph, a DIY Digitizing Tool, May Be The Savior Of Film

Posted by Jim Henk (Member # 3628) on 06-18-2013, 03:02 AM:

Wow. I hope this is as good an idea as it sounds.

Did I read right (not bloody likely) that it could not only produce a cleaned-up digital file, but potentially produce (or make it directly possible to produce) a cleaned-up positive print without having to have insanely expensive internegatives made, etc?

Soundtracks - I wonder how many formats are possible...

Posted by Monte L Fullmer (Member # 2797) on 06-18-2013, 03:49 AM:

(I made a post on this site in FB..) Bet any film projector that was removed from a location for the conversion of digital, would be a good candidate to be incorporated into this Kinograph process.

Bet one could use the optics in reverse: place the camera in front of the lens (using a long throw prime from an anamorphic setup to magnify the image for the camera), place a light source (13w CFL spirial lamp with 5100K natural light temperature) that can produce an even field of light in a small area, and do the capturing in this fashion.

In fact, soundheads should be still included with these projectors to aid in the sound transfer in the Kinograph process.

Soundtracks: If one had a Century JJ penthouse, Simplex magnetic penthouse, or a Motiograph magnetic penthouse, one can go farther with capturing the magnetic tracks along with mono optical.

Optical stereo and DTS=transfer sound from their respective processors.

-Monte

Posted by Manny Knowles (Member # 1171) on 06-18-2013, 04:35 AM:

Monte,

Not only would that be more expensive to make, but it wouldn't really be a new invention. I doubt such a contraption would count as a thesis project, which is what Kinograph is.

Also, not sure if that approach would be in keeping with the "open source" vibe this guy is going for.

Finally, it would limit the user to the gauge of the projector.

Posted by Leo Enticknap (Member # 534) on 06-18-2013, 04:46 AM:

I like the basic concept: advance the film intermittently, but in a fundamentally gentler way than a projector would, and then use a digital SLR to capture each frame as a discrete image. I can see only one problem with his system:

quote: From the article
...the roller uses its strategically-placed bumps to trigger the switch at the start of each frame.

So in effect it uses a crude form of pin registration. This would, I'm guessing, impose a significant limit on how much shrinkage the system would be able to handle.

Professional restoration software can iron out minor variations in the position of the frame borders in order to remove vertical and horizontal weave (e.g. Diamant or Da Vinci Revival), but the cost of that would be significant (depending on whether the licence allows for-profit use or not, potentially into five figures). AFAIK more affordable software, e.g. Adobe After Effects or FCP, can't without bespoke plugins.

So it's probably not going to give you quite the result of a Kinetta or an MWA Vario, but there again at a hundredth of the price expecting it to would be a tall order! And as each frame has to be photographed with an individual shutter movement of the D-SLR, I'm guessing that the scanning speed will be something like 3-10 seconds per frame, depending on how long the shutter cycle takes and how long the camera takes to write each frame to the memory card (and in what resolution and sort of compression). As far as I know, there aren't any D-SLRs that can write DPX files, meaning that unless you hooked it up via HDMI to a computer that could, the output couldn't be used in a professional restoration workflow as yet. But it sound like it would produce a vastly superior result to a Tobin or a Moviestuff machine (which can't do 35mm anyway) for a similar sort of price range and scanning speed, which is certainly a major achievement. It may well bring near 2K transfers within reach of smaller archives and museums that will never realistically be able to afford something in MWA or Kinetta territory.

quote: Jim Henk
Did I read right (not bloody likely) that it could not only produce a cleaned-up digital file, but potentially produce (or make it directly possible to produce) a cleaned-up positive print without having to have insanely expensive internegatives made, etc?

Probably not, sadly. In a professional digital restoration workflow, the end output will be a folder full of DPX files - one for each frame. If an output to film is needed, a CRT or laser film recorder (e.g. Arrilaser or FlashRecord) will then read the DPX files and burn them, frame by frame, to the output film roll.

This process is so slow (typically 20-50 seconds per frame), that it would simply not be a cost-effective use of a machine costing six figures to burn release prints directly. The usual workflow is to burn an IN, and then continuous contact print the release prints from that. In a low-budget restoration, that IN serves as the preservation master: in a higher-budgeted one, a fine-grain pos will be burnt as a master, and then an IN and prints made photochemically from that.

Five years or so ago it was being speculated that laser film recorders would eventually speed up to the point at which for a minority interest archive title, it would become viable to burn single prints. But film is disappearing so quickly now that I doubt if it'll happen.

Posted by Manny Knowles (Member # 1171) on 06-18-2013, 05:28 AM:

Hmm...It's been years, but IIRC After Effects has built-in image stabilization. The last time I played with it was around 2001, but I can't imagine why they'd DROP a feature.

Also, if you view the Thesis Presentation, Matthew Epler (the inventor of Kinograph) explains that the point of this is not to provide a cheap method of executing a full-on restoration. Rather, the intent was to have a way to see/hear what was on a reel of film, and then determine if it was something that warranted grant-writing for a proper preservation/restoration effort.

Demo of Image Stabilization in Adobe After Effects

Posted by Leo Enticknap (Member # 534) on 06-18-2013, 08:14 AM:

quote: Manny Knowles
Hmm...It's been years, but IIRC After Effects has built-in image stabilization. The last time I played with it was around 2001, but I can't imagine why they'd DROP a feature.

In the video you linked, he's stabilising a very wobbly camera in a single video file, not a series of individual frames. In a slightly shrunk film capture, the wobble is going to be so much more subtle ... I wonder if that filter would do anything. About 3-4 years ago I spoke to someone who had done some extensive playing with AE to see if she could stabilise a standard def PAL telecine transfer of a seriously shrivelled 16mm home movie: she told me that she wasn't able to achieve very much, but when a friend of hers, who had access to Da Vinci, put it through that, it came out rock solid.

I have access to Adobe CS6 and do some occasional consultancy for a company that makes scanners in the $100-200k range. When I get a moment, I'll ask them for an (uncorrected) transfer of a few seconds of some seriously shrunk stuff, and see if I can do anything by recreating the steps in that video. Should make for an interesting science experiment! However, I'm not in any way an expert with After Effects - in fact, apart from PAL/NTSC conversions, I've barely used it. So it may be that it's not the software that fundamentally lacks the ability to stabilise a shrunk transfer, but me that fundamentally lacks the ability to instruct it accordingly. Still, we'll see...

quote: Manny Knowles
Also, if you view the Thesis Presentation, Matthew Epler (the inventor of Kinograph) explains that the point of this is not to provide a cheap method of executing a full-on restoration. Rather, the intent was to have a way to see/hear what was on a reel of film, and then determine if it was something that warranted grant-writing for a proper preservation/restoration effort.

Point taken, and I wasn't trying to belittle his achievement in any way. Quite the opposite - in fact, I can't see any theoretical reason why you couldn't achieve at least a Bluray-quality result with a rig like this. With one of the new, mega hi-def D-SLRs (e.g. the Nikon D3200, with a 24mp sensor), you might even be able to aim for 4K. The quality of lens in the camera would be a major factor - good optics and a tiny aperture (the kit lens that comes with a typical entry level D-SLR would not get you very far, probably) - as would the file formats it's capable of saving and hardware/software capabilities of the computer you then use to edit and render. The real difference between something like this and what the big archives and post houses use would essentially be in the amount of time needed to transfer and process a given volume of footage, would be my guess.

Posted by Manny Knowles (Member # 1171) on 06-18-2013, 01:24 PM:

quote: Leo Enticknap
In the video you linked, he's stabilising a very wobbly camera in a single video file, not a series of individual frames. In a slightly shrunk film capture, the wobble is going to be so much more subtle ... I wonder if that filter would do anything.

The technique allows for defining the reference pixels by way of a line segment. Which, for the purposes of our discussion here, should be aligned with something like a frame line or, perhaps even better, the top or bottom EDGE of the frame, or CORNERS of the frame, or opposite sprocket holes. Something like that.

I toyed around with After Effects some time around 2001. At that point, I was just curious what the tools were designed to do and wasn't evaluating the end result with a super-critical eye or anything.

quote: Leo Enticknap
The real difference between something like this and what the big archives and post houses use would essentially be in the amount of time needed to transfer and process a given volume of footage, would be my guess.

There's that old saying:

Good, fast, cheap.
Choose any TWO. [Big Grin]

^^That right there is the single-most useful bit of knowledge I got out of my undergrad degree.

Posted by Jack Theakston (Member # 4484) on 06-18-2013, 01:57 PM:

quote: Manny Knowles
The technique allows for defining the reference pixels by way of a line segment. Which, for the purposes of our discussion here, should be aligned with something like a frame line or, perhaps even better, the top or bottom EDGE of the frame, or CORNERS of the frame, or opposite sprocket holes. Something like that.

The problem with using any frame line is that frame lines also jump around depending on the matte in the cameras that were used and hard mattes.

I agree that using the sprockets would probably be the best alignment system, and might even adjust for things like cupping in the "gate," since you're looking for vertical alignment as well as horizontal.

Posted by Manny Knowles (Member # 1171) on 06-18-2013, 04:04 PM:

Well, I was kinda figuring the workflow would involve a "bulk" pass and then fine tuning shot-by-shot for the ones that need it.

Also, one could try isolating two points in the frame that don't move. That won't work with any shots that have moving camera, but it's something else that could be tried, if it would make sense in a given situation (i.e., a static shot).

Posted by Jack Theakston (Member # 4484) on 06-18-2013, 06:45 PM:

True, although you might want to do that on several points to fix any geometric distortion.

Posted by Manny Knowles (Member # 1171) on 06-18-2013, 08:37 PM:

I wonder if that's possible in one pass. Could be -- it's been over a decade since I last checked any of this out.

As far as shrinkage, and the tolerance of the rollers...

I recall from a brief internship with David Shepard (in the 1990s) that he had a set of special sprocket rollers for different % of shrinkage. Once the % of shrinkage was determined, he installed the appropriate rollers before lacing up the source material in the printer. In the case of Kinograph, one could design/make (i.e. 3D print) separate rollers for different degrees of shrinkage to work with the Kinograph. This would be a neat accessory/option for Mr. Epler to offer his customers. The little "bump" detector (his virtual intermittent) would still work just fine.

quote: Leo Enticknap
So in effect it uses a crude form of pin registration.

I've been giving this some thought -- focusing on the corner of a sprocket hole could provide a good form of (virtual) pin-registration -- right?

quote:
I have access to Adobe CS6 and do some occasional consultancy for a company that makes scanners in the $100-200k range. When I get a moment, I'll ask them for an (uncorrected) transfer of a few seconds of some seriously shrunk stuff, and see if I can do anything by recreating the steps in that video. Should make for an interesting science experiment!

Leo! Do it! Tell us what happens! Or, better still, make a video and SHOW us what happens!

Posted by Mark Gulbrandsen (Member # 72) on 07-01-2013, 05:28 PM:

For a tiny extra investment it would be pretty easy to soup up the quality of this gizmo by using an old projector head or even a pin registered B&H movement. That would provide better registration right from the start at almost no cost.