Custom DIY Telecine

[John Brawley]

As I understand it, Spirit does mathematics on the image data itself to refine the vertical position of the frame - as a line-array device, the film is in continuous motion, which would otherwise make it far too easy to end up with vertical stability issues.

 

Of course, this may just be a case of waiting until the sensor sees the peak white of a sprocket hole goes by as a timing device, but I think it's a bit more than that.

 

P

 

 

You may be right, but as far as I know, the only system that actively images the sprockets for image stabilisation is the Vialta. The spirit does analyse the picture for image stability but this is fed in to the transport system.

 

This is going to be the biggest drawback to using one of these DIY rigs. That and the fact most stills cameras are only good for 50000-100000 cycles. (just over 30 hours @ 24fps)

 

jb

[Keith Walters]

This is more idle musing, but I wonder how it would work if you had a continuous film movement that ran the film past a Xenon flashtube or even a bank of LEDs that was triggered by the sprocket holes. Cheap vidicon-based telecines used to do this in the early days of B&W TV, flashing the tube surface during the vertical blanking period.

There was no intermittent movement because the small amount of film movement during the Xenon flash was so small it was not noticeable.

 

If the film was run slowly enough for a digital still camera to be able to capture and store the individual images, the vertical smear would probably be too small to worry about there as well.

 

Basically, a trigger pulse would be derived from every second (or fourth or whatever) sprocket hole. This would first trigger the camera, which would be set for a relatively long exposure. While the "shutter" is still open, the flash would fire, producing the actual image, which would then be salted away by the camera, and the camera readied for the next shot. Obviously the frame rate would be determined by how fast the camera can take pictures, but the slower the better.

 

By the way, an excellent source of ultra-flat backlighting panels is the "Digital Picture Frames" that are currently flooding the market. Even the cheapest ones have surprisingly sophisticated construction. Usually there is an ultra-clear acrylic panel with a row of holes down one side where 4 or 5 white LEDs are inserted so they shine crossways through the panel. A large number of small holes are drilled partway through the back of the panel and the LED light hitting these makes them appear as hundreds of tiny white lights. Several sheets of some sort of Fresnel-like diffuser film are then laid over the top, and the result is in an incredibly flat illumination, far better than is usually warranted by the dodgy LCD panels often used:-)

[Phil Rhodes]

My, my isn't all this fantasising fun?

 

You wouldn't want to try the constant motion trick with white LEDs, since they're prone to a fair degree of afterglow from the phosphor downconverter, but RGB LEDs should work if you can achieve the same intensity over a short enough time. The idea occurred to me a few years ago when I was considering trying for a little box you could mount - or even handhold - under film on a rewind bench, and have it flash every four sprockets (or every multiple of four sprockets, depending on the achievable flash rate).

 

If our suppositions are correct, it means that vertical registration is limited only the triggering mechanism. You could try for a purely optical gap-detecting approach, though I'd prefer a reflectance sensor. Most off-the-shelf devices designed to do this work in the infra-red and the film is designed to be transparent in that area, as well as variable transparency of intra-sprocket film depending on the stock in use, gate coverage in the camera, and the intricacies of processing.

 

This approach is very nice because you are not reliant on a mechanical contrivance, which might not be exactly the same every four frames, for vertical registration. Assuming you can achieve suitable tolerances of noise, trigger consistency and general stability of this optical detection system, which I think you could, the problem would be furry edge that can sometimes collect on sprocket holes, and the minute perturbations they would introduce. Worth a try, though.

 

You'd still need a properly machined gate to maintain horizontal stabilisation, and clamp the film appropriately. I wonder if a projector gate would be good enough for this.

 

As to shutter wear - yes, you'd want a DSLR modified to get the shutter and mirror out of the way, and control brightness with flash intensity.

 

P

[Hal Smith]

It would make much more sense to start with a flat bed editor as a transport. Flatbeds are going for pennies, there have been a few offered for free here on the Forum and I bought a 35mm KEM telecine dirt cheap off eBay last year.

 

There's a patent that covered a flatbed transfer system. Ed DiGiulio at Cinema Products was the inventor so it's a pretty sound idea from an engineering point of view. Have a look at: http://www.google.com/patents?id=XdsyAAAAE...=result#PPA1,M1

[Phil Rhodes]

Yeah, that'd probably relieve a lot of the sheer engineering involved with film handling.

 

I don't know how modifiable the gate is on one of those, though, and you'd still have the registration issue to deal with.

 

P

[Hal Smith]

Yeah, that'd probably relieve a lot of the sheer engineering involved with film handling.

 

I don't know how modifiable the gate is on one of those, though, and you'd still have the registration issue to deal with.

 

P

 

On my KEM the gate size is a function of the camera's lens focal length. The optical "gate" itself is larger than Super-35 in that if you want to you can see all the way out to the sprocket holes.

 

Registration would be controlled by the method that you used to sense film position. The KEM has a tach sensor somewhere down in its mechanical guts that sends a pulse per frame to the speed control circuitry. Just how position accurate that sensor is I don't know...but being German engineered I suspect it's pretty good.

 

Like many American engineers I both love, and hate, the German approach to engineering. Ever work on a Mercedes? Yuck! Or Arriflex electronics? Double Yuck!!!

 

I have a friend who works as the worldwide support manager for a US electronics manufacturer owned by a large German company. He has stories to tell...including the fact that the software engineers who designed and wrote their equipment's operating systems now have to get to get approval from Herr Doctor Engineers in Germany for any and all changes to, and new, software. Said German Dr. Ing's have abolutely no experience with their product line, their software, don't understand what it does, don't know the customer's needs...but they get to decide whether or not revisions and improvements are needed.

[Phil Rhodes]

A physical sprocket engagement may be perferable over an optical sensor because - experience shows - trying to sense sprocket edges is very subject to being tricked by fluff and roughly punched holes, at least on a microscopic scale. The sprocket itself has an averaging effect when the tooth hits the edge of the hole and compresses any fuzz. Yes, I'm entirely serious; the first line-array telecine by Marconi initially had this problem due to a lack of drag on its tacho sprocket.

 

Also, if you have - say - a 16-tooth sprocket, you're only using the same piece of mechanics once every four frames, which would give you similar stability issues to a projector with a Maltese Cross style movement - possibly not good enough.

 

I wonder, though. You could prove the stability fairly easily.

 

P

[Keith Walters]

It would make much more sense to start with a flat bed editor as a transport. Flatbeds are going for pennies, there have been a few offered for free here on the Forum and I bought a 35mm KEM telecine dirt cheap off eBay last year.

Yes, as long as you are single and have a garage the size of an aircraft hangar :lol:

I know, I and many others have wept tears of blood at the thought of such technology winding up down at the dump (along with Analog TV studio equipment that probably cost more than I will ever earn in my entire lifetime), but in the cold light of day, you know you are unlikley to ever do anything with it.

[Keith Walters]

A physical sprocket engagement may be perferable over an optical sensor because - experience shows - trying to sense sprocket edges is very subject to being tricked by fluff and roughly punched holes, at least on a microscopic scale. The sprocket itself has an averaging effect when the tooth hits the edge of the hole and compresses any fuzz. Yes, I'm entirely serious; the first line-array telecine by Marconi initially had this problem due to a lack of drag on its tacho sprocket.

 

Also, if you have - say - a 16-tooth sprocket, you're only using the same piece of mechanics once every four frames, which would give you similar stability issues to a projector with a Maltese Cross style movement - possibly not good enough.

 

I wonder, though. You could prove the stability fairly easily.

 

P

My hare-brained scheme is to make the sprocket assembly from scratch, by first gluing a 200mm or so strip of film to a similar-sized strip of 1mm thick copper, painting every part of it with bituminous paint except for where the copper is exposed through the sprocket holes, and then electroplating more copper through the sprocket holes to form a series of copper "mesas".

 

After that, the newly-plated copper would be etched in an ammonium persulphate solution (as used for etching printed circuit boards) to smooth the bumps off a bit.

 

After that, the film would be removed and all the bitumen removed cleaned, the bumpy surface tinned with lead-free solder, and then baked at red heat for a short while which will turn the electroplated copper into much harder bronze.

 

The excess solder would then be dissolved in hydrochloric acid leaving a precision replica of the holes in a strip of film.

 

If two of these were then fitted to 60mm diameter (or so) machined drums they should allow the manufacture of an ultra-precision transport assembly.

 

As for accurate location of the sprocket holes, possibly the best solution is to have more than one sensor, and use some sort of fuzzy logic system to make a "best guess". I wonder if some sort of capacitive pickup would be more practical.

[Phil Rhodes]

Now that's utterly hairbrained.

 

I wouldn't worry about the sprocket - a chunk of machined metal from people who've been doing it for years would likely have stability problems. I'd personally far rather try and deal with it in electronics or software.

 

P

[Mark Williams]

This got me thinking so I dusted down my old eumig 8mm projector to see if I could telecine some super8mm Kodak 200T I had done professionally and apart from the strobing I think my results are better. Does anyone know a way to get rid of the strobing effect? I used a Sony EX-1 at 18fps with the projector set the same. I colour corrected in after effects although because of the strobing it seemed a hit and miss affair. I got one bit right only to find the rest had changed colour temperature on me. Unless I'm doing something else wrong.

 

http://www.vimeo.com/3620427

 

Sorry about the music.

Edited March 13, 2009 by Mark Williams

[Phil Rhodes]

Apparently, it's a private video.

 

P

[Mark Williams]

Sorry Phil It must have defaulted to that. Should be okay now.

 

Mark

[Paul Bruening]

Hey Mark,

 

I see a couple of artifacts:

 

You've got banding. I'm not quite sure what that is. Does your projector have a shutter fan located between the gate and light source? If so, it may be contributing to flicker. You'll want to determine if it allows for two projections of the same frame. If so, it is there to divide the normal frame rate into two. The idea is that a 24 fps image, for example, is divided into 2 yielding 48 images per second and reduces the perception of flicker. Since you're using the unit as a telecine, you might remove it and put all its little parts in a box so that it won't interfere with your telecine goals.

 

You've got double framing. Every once in a while a ghosted image came up in your video. That could point to the recording of two frames at once. I can't answer for how your video camera does this. But it points to a sync issue.

 

Are you using a really slow shutter on your video side? If so, it may be giving you some benefits with overt flicker but adding that banding and ghosting problems. Just guessing, though.

 

Generally, it looks like you're on to something useful and cost effective. Please, keep us up on how the project proceeds.

[Mark Williams]

Hi Paul

 

I think the problem has something to do with it being a three bladed shutter on the eumig projector and yes it does have a fan. I was pleased by this though compared to the professional telecine which had previously put me off 8mm neg for good. I wonder if the huge amount of grain seen on my pro telecine transfer might be because of sharpening. Or some other reason.

 

I can see the possibilty for good 8mm neg telecine now .

 

I'll be trying to telecine some vision three 16mm the same way soon using an elf projector.

 

Mark

[Paul Bruening]

Hey Mark,

 

What's the latest news on the project?

[Mark Williams]

Hey Mark,

 

What's the latest news on the project?

 

Hi Paul

 

The latest is I'm waiting for a split reel for my elf. Also someone has said they will telecine it professionally for free. Although that may not happen. When I get some footage I'll post it here!

 

Mark

[Stephen Hargreaves]

Just some updates on my experiences with the DIY telecine:

 

I had some successful results using an optical printer rig originally used for film projection to bolex, and this rig has pin registration, so shake in frame wasn't a problem. I found that the amount of filtration that worked for the majority of stocks tested to eliminate the orange masking was 2 1/4 Full CTB conversion. It still needed some pulling around with curves and the channel mixer in photoshop though. I shot raw with a digital rebel XT, 100 ISO at 1/400, and the results seemed to be best around there, I took the lens off of the camera, and used the lens with a bellows that was used for the bolex camera rig on the optical printer, thus, basically projected the image onto the sensor.

 

I am going to have to do a ton of color correction to get all the colors where I'd like them to be, but with all the time in the world, and absolutely no budget, this seems attainable.

 

Does anyone have any suggestions, about shooting specs, IE white balance, ISO or shutter speed? Any ideas would be useful.

 

I plan to post my results soon.

[Topher Ryan]

Just some updates on my experiences with the DIY telecine:

 

I had some successful results using an optical printer rig originally used for film projection to bolex, and this rig has pin registration, so shake in frame wasn't a problem. I found that the amount of filtration that worked for the majority of stocks tested to eliminate the orange masking was 2 1/4 Full CTB conversion. It still needed some pulling around with curves and the channel mixer in photoshop though. I shot raw with a digital rebel XT, 100 ISO at 1/400, and the results seemed to be best around there, I took the lens off of the camera, and used the lens with a bellows that was used for the bolex camera rig on the optical printer, thus, basically projected the image onto the sensor.

 

I am going to have to do a ton of color correction to get all the colors where I'd like them to be, but with all the time in the world, and absolutely no budget, this seems attainable.

 

Does anyone have any suggestions, about shooting specs, IE white balance, ISO or shutter speed? Any ideas would be useful.

 

I plan to post my results soon.

 

I'd love to see these results. I'm 100% with you that this sort of scanning is attainable, and moreso by the day with the advances on the digital and post-processing side of things. The film transport bit has been there for decades, the consumer sensor market is pretty much there now, someone just needs to marry the two in a low cost solution. HD scans are out of reach for most low-budget shooters and SD one-light telecine leaves MUCH to be desired.

 

 

I started a thread in the Bolex subsection asking about which camera to use for a scanning rig, but it quickly turned into me just rambling on about about my bolex rig brainstorms. I don't know if you guys venture over there much so I'll link it here in hopes that I can get some feedback or criticism...

 

http://www.cinematography.com/index.php?showtopic=37666

 

Thanks

[Karel Bata]

I've been thinking about this for a while now, and am delighted to find this thread!

 

You realise that what you're trying to do, assembling a moving image from a series of stills, is essentially stop frame animation? And the software is already out there, like Stop Motion Pro. You go straight to HD with no messing with PhotoShop. Other software and advice here http://www.stopmotionworks.com/stopmosoftwr.htm

 

Registration probs? Use something like Red Giant's Magic Bullet Steady in AE - which is designed to fix probs with bad sprocket holes, and then 'fix it in post' :o .

 

I agree with Keith that using a flash is the way to go. It's already at daylight so will need minimal filtering. You'll need a big pro one so the bulb doesn't burn out after several thousand flashes. If you connect the flash to the camera's hot shoe it will trigger it for you -no need for any more electronics! You just have to find a way to trigger the camera every time a frame is advanced. To me, done this way, it doesn't look too complicated.

 

Ideally we should have a wet gate, but that's maybe asking too much...

 

My big worries are with dust and scratches. :huh: So do we need a seperate thread for DIY ultrasonic cleaning? :( Seriously - how do you clean the film of any crap it's picked up without scratching it further?

 

 

BTW a good source of LED panels is Luminous Film And you can use these for making your own lightweight softights. ;)

[Bert Smith]

This is a very encouraging thread. I cant imagine that the sensors within the Arriscan and Spirit machines have the tens of millions (hundreds, perhaps?) of dollars of R&D that something like Canon's 5D MKII sensor has. But adapting it for use photographing individual frames of film seems like a difficult task.

 

Even the cheapest digital point-and-shoot cameras now have 12+ megapixels. Please someone, come up with a good solution! Anything under $5,000 and decently user-friendly (big problem there, I assume) would surely have plenty of buyers. Even a tiny project that we had telecine'd at Fotokem was half of that amount I just mentioned. It would pay for itself within just TWO short projects!

 

So I for one, would be first in line to purchase a sub-$5,000 film scanning system.....IF IF IF...it yielded results decently close to what we've gotten at Fotokem in the past. Or at least "close enough" to fool art directors and agencies. The prospect of potentially eliminating all future film scanning costs with a single large purchase is overwhelmingly exciting. With film scanning costs eliminated, why not shoot 35mm all the time? Why not begin those personal and self-promotion projects that have been on hold because of cost? And good lord, think of a 2-perf 35mm system of this nature! No scanning costs, and double the footage from the same amount of film! Shooting on glorious film for less then the rentals costs of renting the RED? Can such a world really exist???

 

However, for the non-technically-minded like myself, I can only sit and wait for brilliant individuals to come up with a good system, and then hopefully hire them to build the same thing for me. I cant imagine I am completely alone there.

 

Please someone/s....make this a reality!

[Phil Rhodes]

Can such a world really exist???

I doubt it.

[Justin Cary]

This post has been dead for a while but I thought I'd post a Reversal scan and a Negative scan we did with an old GAF projector and a 5DMKII. We're basically doing what the post originated from. We're just telling the camera to take a still frame of every frame and stitching the images together in FCP. It's very basic right now but we are working on purchasing a camera that is mirrorless and shutterless to replace the MKII. The results for Super 8 are quite good and we'd love to nail down the negative conversion but for now we're just looking for a really great 2K scan of Super 8 reversal. I will post a negative scan as well but it's pretty bad as we were using an LED that didn't cover the mirror. It's exciting to see how people are pushing this format. I purchased a Canon 514XL-s in 2007 for .99 on eBay... You can't pick a 24fps Super 8 camera up for much under $100 these days :) Anyway, here's what we've got going so far and we are dangerously close to purchasing the camera to give us 2K images at 15fps!

 

Reversal:

 

Negative:

[Will Montgomery]

Great work. Dust is pretty harsh but images on the reversal look very good. Fun project.

[Justin Cary]

Great work. Dust is pretty harsh but images on the reversal look very good. Fun project.

 

Yes the dust is pretty awful especially on the negative stock. I haven't taken a great deal of care of either of these reels as they've been the test reels for this machine. I do have 5 or 6 rolls that have been processed and prepped but never opened. I hope to throw them on the telecine this weekend and have much better results. I'll post the results next week.