Bob,

There are lots of issues doing film transfer and we've both dealt with a lot of them. For me, getting to the edge of the film is not too important, although I've appreciated the enlarged film gate that Roger created on his Workprinters. In my home-brew version of his machine, I'm not going to attempt that.

As for gate jitter, there are some special features that Gunar Thalin added to his Deshaker program specifically for a user that wanted to eliminate this problem with film. It works extremely well, and is what I use. I can send you the settings, if you wish.

The optics are solved by the aerial lens that comes with the Workprinter. I am simply going to use that, with a different mount, with my home-brew 16mm telecine. Using that lens, I was able to capture even the small size 8mm frames, using the 10x zoom on my video camera. With the Nikon D70 and my 70-210 mm lens, I have a zoom that effective is a 300mm zoom (because of the multiplier factor when focused on the imaging element in the D70 which is smaller than the 24x36 size of 35mm film). It would easily be able to zoom into the image created by this aerial lens. If you are interest in sources of supply for this lens (which projects the image into space, thus eliminating the need for a screen of any type), I can dig that up for you. It is the single most expensive element of the film transfer (far more than you pay for a used film projector).

As to your main point about using a digital camera, I'd do that in a heartbeat if my Nikon D70 could lock up its mirror and take pictures. However, Nikon saw fit to remove that feature from its "lower end" DSLR cameras. I don't really want to take 10,000 pics and wear out the innards of my nice camera, and without being able to lock up the mirror, there is no way to avoid a lot of wear. I do note, however, that this is how most of the stop motion people are doing their HD projects.

As to my own modifications, the projector has not yet arrived (purchased on eBay for $9.99 a week ago). However, when it does come, I plan to design and build a solid state motor speed controller so I can get any speed; I'll replace the bulb, like Roger did with the Workprinter, although he used a small incandescent which required him to build a condenser assembly in order to get sufficient diffusion. I am not quite certain what I'll do here, but the easiest is to use a really low-wattage halogen and then wire a diode in series, which effectively drops its light output in half. That's the part of the design I haven't quite figured out. Since I already have the modified mouse from the Workprinter, I'll simply input my trigger to that. Roger used a microswitch that was triggered by a cam he attached to the shutter gear. Since I am an electrical, not a mechanical, engineer, my initial idea is to just use an LED sensor, paint a reflective dot on one of the gears, and have the sensor switch a relay on and off. Since I have all these parts in various bins here, it shouldn't be too hard.

Having said that, it took me four hours just to re-build a battery pack for my old Sunpak video light (can't buy the packs anymore). Good news is that I was able to upgrade from 1,000 mAh NiCd to 1,800 mAh NiMh so I get almost twice the run time. But that's another story ...

Not sure I understand why the FX1 won't work. Is it the cropping of the 4 x 3 to 16 x 9? How long does the camera run when the frame is stopped on the projector? Or does the camera run continuously?

JJK

You might want to consider a computer controlled timing circuit in place of a microswitch or LED sensor. The Led's are better than the Microswitch but they still are not that good, especially when you are dealing with .0001" increments.

JJK

The FX1 won't work because there is no software for capturing single frames from HD cameras. This has to do with the way most HD cameras connect to the computer. Most of them are designed to transfer the HDV stream which is exactly the same data rate as the "traditional" SD DV video stream. However, each SD DV frame is a single discrete entity, whereas HDV frames need to be constructed from information transferred over the period of many, many frames.

The solution would be to have a camera that transferred the uncompressed HD signal, or some sort of DV equivalent where each frame was compressed independent of the other frames. I think some of the really high-end cameras may have this, and I have heard that one of the Panasonic HD or HDV cameras has some sort of similar capability. However, I've spent a lot of time perusing Stop Motion forums, like this one: stopmotionanimation.com and smarter people than me have not figured out how to capture single frames from HDV cameras -- certainly not from the Sony FX1.

When I get my projector project finished, I will proceed as I discussed in my earlier post, where i will capture to tape, and then use a script to dissect the frames and extract only those that I need. Actually, I will probably use AVISynth which lets me actually decide which frames to keep by looking at the content of each frame, something I wish Vegas provided.

In fact, the more I think about it, I may be able to eliminate almost all the work of making the Workprinter clone ... all I need to do is slow down the projector ... and then let the script do all of the work. Hmmm ... this could actually make flicker-free film transfer extremely easy ...

Yeah, the more I think about this, the problem is very similar to some of the crazy inverse telecine scripts I've written in AVISynth where the cadence of the pulldown changes, not only at scene breaks, but sometimes within the film itself. The worst was when I had someone give me a videotape they bought for their father: "Men of the Fighting Lady," a 1954 film about life on an aircraft carrier. She wanted me to transfer it to DVD, which I figured was legal, since she owned the tape and was going to give both to her dad. I finally figured out that the film had probably first been transferred to PAL, and then that transfer had been re-transferred over to NTSC. Because of the 24 to 25 to 23.976 to 29.97 conversions, trying to recover the original 24 fps progressive frames (which you HAVE to do if you want really great DVD encoding), required an amazingly complex script that looked at several groups of frames before deciding which frames could be discarded.

The only problem here will be what to do when the projector loses sync (from bad sprockets).

But, I digress ...

John,
I'd be wary of using an incandescent lamp with a diode in series.
Two problems, firslty the filament runs much cooler, read orange and secondly it's now running at 30Hz (in the US) or 25Hz (down here) the flicker can be quite noticable. A Luxeon 5W LED is a much better bet, you dim the thing by cahnging the current and the color temperature stays the same. You can even switch it on and off in uSecs if you want to.

For the camera, I've got a Sony DSC F828 that I can fire the shutter with via its LANC port and I'm looking at a macro lens that'll go on the front of it for not too many dollars. The aerial focus concept I'm not too certain about, seems to like a LOT of glass. I can by purpose built CCD blocks also and lens assemblies to match, used by the astronomy buffs so they're pretty good, argon cooled etc. What I'm really trying to do is photograph the film in the gate.

I've even thought about repurposing a Nikon slide scanner, they've got some excellent dirt and scratch removal technology based around using IR and visible light and yield around 3K line res off 35mm, really all that needs doing is making up a feeder and film guide. Of course it'll be a very slow scan, at best 1fps but that's no slower than some 35mm scanners working at 8K res.

One thing I noticed scanning a few 1000 slides is the huge advances in film stock. The old slides are so grainy that 4000dpi is a joke yet some I did as 16 bit targas from a recent shoot are just staggering, no grain at even 4Kdpi. I can now see why once 70mm was attractive and why it died off.

Bob.

Bob,

Very useful information. Thanks!

I was aware of the color shift you get by dimming a lamp. I figured I could handle that with the initial color balance, plus additional color correction in post. Every film transfer I've done (or seen done by others) requires color correction in post.

The flicker, however, I hadn't thought of. I don't want to deal with that. I'll check out the LED lamp. That sounds ideal, if the color spectrum is close to some "standard." My only concern would be that it might have some horrible color cast, like fluorescent bulbs, and I'd never be able to get the color to look right.

Photographing the film in the gate is an interesting idea. You are correct that it would required a special lens. Also, any curvature of the film might be a problem. I know that many lenses for Kodak carousels were corrected for curved film. For instance, thirty years ago when I purchased a special Navitar lens for my carousel, it was offered in a version for glass mount slides (no curve) or cardboard mounted slides (where the film bulges slightly). I don't know how flat the film lies in a movie projector gate, but the depth of field with macro lenses is sometimes a problem. The flip side is that there will still be quite a bit of light so perhaps you can solve the problem by stopping way down.

As to film scanners (or flatbed) my advice is: Don't go there. I have a Nikon Coolscan 4000 which has a Firewire interface. I have scanned over 30,000 slides and negatives in the past twelve months using this scanner and have tried every trick in the book to reduce scanning time. The fastest you can do, and only if you scan at resolutions below 1,000 dpi and only if you turn off autofocus and autoexposure for each scan, is about 25 seconds per scan. No way you can process movie film at that speed.

As for flatbeds, there are all sorts of kludges people have built to do this. Some have even constructed jigs that mount on the flatbed to hold the film in exact position, and then they have designed elaborate software to cut and register each frame of film. Bottom line: the approach is slow, and the registration issues seldom work out.

I can tell you from experience that the Workprinter approach works amazingly well, although I have one of Roger's earliest units, and his mechanical modifications were crude and involved rubber belts that wear out after only a few hours use (I now have a lifetime supply after I discussed this with Roger). The cam adjustment is also crude and unnecessary, since the same adjustment can be made, trivially, in software. In fact, I emailed Andi, the author of Scenalyzer, eighteen months ago, and he added a delay adjustment (in the registry file) to SCLives stop motion feature so you can delay when the picture is captured relative to when the mouse button is activated by the Workprinter. Just as with my idea of using a script to pick out the good frames from a continuous capture, you can sometimes save a great deal of money in hardware by being a little clever in how you design the software.

Finally, on the old film stock, a lot depends on the stock. My 30,000 scans included stock going back to the late 1800s. That was pretty grainy. However, in the early 1900s, some photos were shot on the forerunner of Pan-X, and this is some of the most amazingly fine-grained stuff I've ever dealt with. The early Kodachromes from the early 1940s were pretty low res, although Kodachrome supposedly doesn't have "grains" as such. This was true until they made their first major changes in the early 1960s. Most of the other color still photo stocks were pretty grainy, especially the high speed films like high-speed Ektachrome. However, once Kodak introduced Ektar, things changed rapidly, and with scans of pictures made with some of those stocks, even 4,000 dpi was not enough (i.e., when I zoom in, I see pixelization before I see individual grains).

I used to repair 16mm projectors. The film gate has a spring loaded edge guide that puts constant pressure on the edge of the film keeping it in place horizontally. The pressure gate mounted on the lens assembly puts pressure on the film surface to keep it flat up against the backplate. This is where all of the scratch lines on the emulsion come from when a piece of dirt gets jammed into the gate. If you lighten up on the spring pressure the claw won't seat properly in the holes and the picture will jump around. You would also need a huge fan to sink away the heat from the bulb. It would be a nightmare on cooling if you slowed the projector down to 6 frames or even 1 frame unless you had an auxillary fan I don't know what the standard projection lenses were cut for but we used to put a curve into our scope screens by stringing a pencil from the projector and drawing the curve on the floor. Also the claw mechanism must be absolutely perfect and adjusted perfectly to get rock steady projection. The old Bell & Howells were capable of this performance. There were some projectors floating around that did not use claws but rotating faceted mirrors or prisms. Don't know how they worked though.

JJK

Bob Kiger (videography.org and videographyblog.com) is working on intervalometers for HDV, you might want to chat w/him.

I've done some stop motion work but limit my projects to studio work where I simply run a live feed to the computer and use (cough) Primier's stop motion function to snap off stills. I have not used Primiere in a while but I suspect you could send live HD video to it without running a tape.

However, I was never happy with the quality of the images especially when I wanted to go into photopaint for some touch up or "cheating" animation. The project "Ethics Infomercial" found on my website is an example of that workflow:

http://www.wakelydam.com/Video/index.htm

But I'm thinking for my next stop motion project I might do away with the video camera altogether and opt to shoot it with my Nikon D70. I have a better selection of lenses and a huge file to work with. Plus I'll have much better low light capability. At a medium resolution and a 1 gig CF card you could dump 500 frames easy. Then let Vega combine them into still image sequence via project media.

I wish Vegas would add the stop motion feature and some onion skin options...