Custom DIY Telecine

[Stephen Hargreaves]

Hi,

 

I'm working on a film in which we're shooting s16, and then we are planning to edit digitally. The only problem is we don't have money for an HD telecine, and I've investigated other routes, and I've come to the conclusion of doing it myself. The budget is micro, almost none, so I'm looking for the cheapest, yet highest quality result. Time is really no matter, I can dedicate as much as needed.

 

I have access to an optical printer, and what I was planning on was using the projector, but then instead of capturing the image using another film camera, using my Canon Digital Rebel XT SLR instead. I have heard of this (and scene some results), and the idea was that you could take off the lens and simply project the image directly onto the camera's image plane (in this case, the CMOS sensor).

 

I have a few questions regarding this:

 

1. Can I use the negative and then invert it in Photoshop later? I have heard from one of my instructors that this wouldn't work that well, because it would leave an orange looking mask or wash to the resulting image. But could that be corrected? Or would I have to opt for a work print?

 

2. What would be the best workflow for converting these stills into workable video footage for use in Avid or FCP. Would it be best to merely bring in all the stills at a 1 frame duration, then compile and export the sequence as a clip? And if so, what would be the format I would go with? Let's say in FCP I would want to work with something high res, like Apple Pro Res, good idea or not?

 

 

Thanks for you help!

Stephen

[Kristian Schumacher]

Hi Stephen,

 

I am certainly no expert on this, but I have done a little bit of fiddling around with it. I rigged an old 16mm camera to trigger my Nikon DSLR to take a photo every time the shutter is open. Then backlit the film from inside the camera and take a photo of the film itself from the outside of the gate with a macro lens. Seems quite doable although I never carried through with doing a lot of footage.

Simply reversing the image in PS gives a horrible result - very blue and difficult to bring back to a normal colour balance. You will need to spend some time getting those colour correction settings right for it to be decent. Or you could shoot reversal film

Please let me know how you get on with this. I may still give it another go one day - but there are just too many fun DIY 16mm projects to do ;-) I have converted a few cameras to s16 and built my own underwater housings so far.

 

Best of luck,

 

Kristian

[Phil Rhodes]

This has been much pondered before.

 

You will need a decent macro lens for the DSLR, one that's capable of shooting as flat a field of focus as possible. Focus is microscopically critical on targets of this size, and the easiest way to do it may be to put the camera on an X/Y vernier table and wind it back and forth with the focus fixed, although that will change the image size somewhat.

 

Use blue filters (either between the lens and the film, or just gel the backlight CTB) and correct for the orange mask optically, as accurately as possible, to minimise the amount of noise-provoking digital manipulation you have to do.

 

Consider shooting more than one exposure of each frame, and using HDR software to combine them, for greater dynamic range.

 

Make sure you're shooting the emulsion side of the film, not the back (although in a camera, with super-16, you may have no other choice than to get this right).

 

You will wear out the shutter mech on your camera quite quickly. If you're doing a lot of this, consider having a DSLR modified to keep the shutter open and mirror up, and flash an LED backlight at it.

 

I've always wanted to try this - all I've really done is shot some frames of 35mm neg, but I suspect it can be done.

 

P

[Freya Black]

Doing this with neg is really hard.

Apparently some cameras have a built in negative function that will convert into a positive image in the camera before reaching the computer and this can help but is still difficult.

 

You are better shooting reversal for this. Of course 16mm ektachrome is expensive, so may take away the advantage.

 

Maybe it would be better to do this on S8, or alternatively if you could shoot 16mm B&W reversal and find cheap processing that could also work.

 

Hope that helps!

[Paul Bruening]

It's not only doable but the results are cheaper than buying the scans. However, there are some unavoidable costs of getting the rig up and running with dependable results. If you're just doing a short, then the start-up costs aren't worth it. If you'll be doing multiple feature lengths then the savings per scan more than offset the start-up costs. Keep in mind that while scan sizes can be large and Canon RAW is very efficient to store in, the scans will be Bayer patterned. I never minded that. My scans look great anyway. I've got the whole process refined to deliver scans at $0.015 each. I can scan a whole 2-perf movie for around $800.00.

 

As far as converting the negs to digital pos: set a macro in Pshop with these commands- invert, auto levels, then specific image adjustments appropriate to your particular image requirements. Pshop will convert, preprocess and store it all in your preferred file type while you sleep.

 

It's doable. Complicated in design and execution but doable.

[Phil Rhodes]

To do neg well, you need a lot of plus blue optical filtering, or it becomes a complete nightmare. Commercial scanners use illumination that's very heavy in the blue. Not sure how easy it would be to do super-8; you're practically into microscopy there (in fact, a microscope with a camera mount might be a workable approach!)

 

My preferred approach would probably be a monochrome machine vision camera and RGB backlighting.

 

Not at all sure about auto levels - that'd tend to produce flicker, I'd have thought, as it does if you leave auto white balance on a DSLR during timelapse.

 

How do you work out 15 cents a scan?

 

P

[Stephen Hargreaves]

Great,

Thanks for all the information so far. I'm going to play around with the scanner tomorrow, and see what results I can get, and I will take the advice of shooting with a heavy blue filter, see how that turns out. Any ideas about the post workflow for combining the multiple frame files into single video clips?

 

Thanks for everything,

I'll keep my process posted!

 

Steve

[Phil Rhodes]

Any ideas about the post workflow for combining the multiple frame files into single video clips?

 

If you want to maintain high bit depth, you may have no choice but to use Photoshop or the free tool ImageMagick to batch process the camera raw files into something like a 16-bit TIFF, which After Effects will read, then make high bit depth Quicktime or AVI movies from there. That's potentially a lot of lengthy, sizeable rendering and data wrangling, but it will give you results limited only by the sensor and optics in the camera. The same workflow could give you DPX sequences with less messing about, which you could take into a Quicktime reference movie and cut via FCP and Color.

 

If you aren't so bothered about high bit depth, perhaps rationalising that you'd have otherwise used HDCAM, there are more options. You could shoot JPEGs in camera, or do something similar to the above and get to one of the 8-bit formats supported by the free tools mencoder (the encoder counterpart to mplayer) or ffmpeg, which are capable of montaging stills sequences. They only support images up to 2048x2048, and whether more detail than that is relevant is up to you, but you might consider that you wanted enough oversampling to average out the JPEG noise. Either way, get down to an 8 bit 2K image and you can fire it directly into ffmpeg and create either AVI or Quicktime. ffmpeg is also highly flexible in terms of compressed output, and could reasonably be pressed into service to create your offlines as well. It can create various things that FCP will read, as well as Avid's DNxHD codec.

 

P

[Stephen Hargreaves]

Great,

Thanks for the post workflow information Phil!

 

Steve

[Paul Bruening]

I don't mean to sound like a spoil-sport. But, if you're originating on 16mm an HD telecine might be good enough for your needs and still be cheap enough. This thing of DIY scanning is great in the conception but, sometimes, a bit hellish in the actualization.

[Phil Rhodes]

Seconded. I've shot a few frames and thought "yeah, that'd be OK if I did X, Y and Z."

 

I haven't shot ten thousand usable frames on such a system.

 

P

[Kristian Schumacher]

OK, my interest has been sparked again :-) Thanks guys for your input on this issue for the original poster. I just pulled the old camera out and took a snap with a light filtered through a blue plastic lid I found in the kitchen. reversed colours in Gimp. Very quick and rough, but what was last time really hopeless looking now seems completely doable. I will attach my snap if this site will let me <_<

 

Thanks again,

 

Kristian

 

P.S. Picture upload failed again. I have read several other people struggle. What am I doing wrong?

[Freya Black]

To do neg well, you need a lot of plus blue optical filtering, or it becomes a complete nightmare. Commercial scanners use illumination that's very heavy in the blue. Not sure how easy it would be to do super-8; you're practically into microscopy there (in fact, a microscope with a camera mount might be a workable approach!)

 

My preferred approach would probably be a monochrome machine vision camera and RGB backlighting.

 

Not at all sure about auto levels - that'd tend to produce flicker, I'd have thought, as it does if you leave auto white balance on a DSLR during timelapse.

 

How do you work out 15 cents a scan?

 

P

 

Had a friend who did this with S8 neg and got okay results but it was very difficult and he was preety good at colour correction and all the techie stuff. I think they were experimenting with blue heavy L.E.D.'s at one point possibly with some filtering too. Wish I could remember more. I know that even they found it difficult and they had a lot of experience with DIY telecine setups.

 

S8 reversal is much more doable and this is how the Sniper works and possibly the flashscan (don't quote me on the latter tho). You are right as well phil that such S8 setups tend to involve macro tubes and the right lenses and stuff. Theres a lot of people made DIY S8 setups of this nature tho.

 

I'm sorry I'm being somewhat vague about all this, it's been a few years since I was involved with doing film things and already I find my memory is going kinda hazy.

[Topher Ryan]

What would be best for transporting the film in one of these DIY setups? Projector, cine cam, steenbeck, moviola, etc? I have access to all of these and a DSLR that never gets used, so I'd love to see this further discussed or any resources linked.

 

I've heard shutter mechanism death listed as a main hurdle, but honestly, you could probably replace the shutter mechanism every few thousand feet and still come out ahead, money wise. (approx. $100-150 USD last I checked)

 

Has anyone had experience with mirror lock-up mods in their DSLR's? Any models particular good for this?

[Phil Rhodes]

I presume you could probably take a DSLR to an independent camera maintenance shop and have it modified to do this. The way I'd do it would be to lock the mirror up and the iris in place (which can cause flicker on timelapse) and control the exposure by flashing the backlight.

 

But really, it doesn't take long before you are better off doing this with a monochrome machine vision camera, and RGB backlighting.

 

P

[Paul Bruening]

What would be best for transporting the film in one of these DIY setups? Projector, cine cam, steenbeck, moviola, etc? I have access to all of these and a DSLR that never gets used, so I'd love to see this further discussed or any resources linked.

 

I've heard shutter mechanism death listed as a main hurdle, but honestly, you could probably replace the shutter mechanism every few thousand feet and still come out ahead, money wise. (approx. $100-150 USD last I checked)

 

Has anyone had experience with mirror lock-up mods in their DSLR's? Any models particular good for this?

 

Bruce "G'Day" McNaughton did the mods to my Mitchell to make it do the transport part. He's familiar with all the hurdles of a rigged-rig. He supplied the machining, motor and computer controller software.

 

http://www.arandafilm.com.au/

 

Best of luck to you. You'll need it.

[Phil Rhodes]

This is an idle musing, but I have wondered if it'd be possible to obviate a pin-registered movement by imaging sprockets and stabilising in post. It'd cost you a fractional amount of sharpness, but then with a decent DSLR you're oversampling anyway, for HD. Spirit does something akin to this. It'd be a big postprocessing job, but you'd only have to locate the frame very roughly in the right place, possibly using some sort of reflective infra-red sensor to count sprockets.

 

P

[Topher Ryan]

As far as film transport, how about just using a camera with time lapse motor? Then sync that up to a laptop controlled DSLR storing files to hard drive.

 

Couldn't this be done without any fancy mods or computer syncing? I work in 16mm so I'm thinking a bolex (ideally w/ 400' mag) with Tobin time lapse motor. As for control of the DSLR, most come with time lapse software for remote firing/storing on your laptop or external drive.

 

I just looked at the innards of the Bolex, and maybe there is another camera out there that would have more room for getting a light behind a modified pressure plate, but the bolex is the cheapest camera that came to mind with readily available time lapse motors.

[Phil Rhodes]

Yes, that's probably a reasonable approach.

 

You can defer some of the complexity into software or you can do it in hardware; really just depends what you're more comfortable doing.

 

P

[Alex Lindblom]

Dan Hudgins has a great web site on this topic Dancad3D.

 

The layout of the page may not be the best, but the info is great.

[Phil Rhodes]

He's doing nine (crikey) exposures per frame, three each for red, green and blue.

 

I would like to think it ought to be possible to get reasonable results without doing that, but I hesitate to argue with someone who's been there.

 

P

[Topher Ryan]

Bruce "G'Day" McNaughton did the mods to my Mitchell to make it do the transport part. He's familiar with all the hurdles of a rigged-rig. He supplied the machining, motor and computer controller software.

 

http://www.arandafilm.com.au/

 

Best of luck to you. You'll need it.

 

 

Can you offer any advice on the backlighting and pressure plate modification? I would be content to just get a workable system going for reversal at first, to avoid the trouble of negative coloring discussed above.

 

LED's have been mentioned here and to my knowledge will be small (good), and cool on the film (good). Would LED be the route to explore? What lighting technology to the top telecine machines use?

[John Brawley]

Spirit does something akin to this.

P

 

 

I think you mean the Sony Vialta ???

 

JB

[Phil Rhodes]

I think you mean the Sony Vialta ???

 

JB

 

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

[Paul Bruening]

Hello Topher,

 

We did all this in depth back three or four years ago on this forum. Some of the old timers might recall how we worked it out as a community. As a result of the community mind, I built my scan/telecine rig. It was the boiled down wisdom of everyone involved.

 

Dominic was resolutely against letting software register the images. He felt that there would be too much impact on the sharpness of the image. So, I went with pin registration using a Mitchell NC that was already geared to the 2-perf. Bruce did the mods to the gate and pressure plate so that I could get telecine from my XL2, scans from my Kodak DCSProC with Canon's MPE 5:1 variable Macro lens. The gate was cut along the edge so I could always include Kodak's edge numbers for reference on the telecine for accurate accounting of scans from the telecine workprint. Bruce also provided a motor that could do both continuous running and gutsy pulls on single frames. His guy, Paul Moorecroft, supplied a nice little GUI to run both the Mitchell and the Kodak in concert off of my capture computer.

 

I don't really get what all the fuss about capturing negs is all about. Either blue the backlight or do it all in Pshop.

 

What's even better now are the DSLRs. At the time, the Kodak was the only chip big enough to do 4K. Now, there's the Canon XSi for cheap. There's also the D5MII. If I change my head's motor speed to 30 fps, I can get real time HD telecines and drop them into a 24 fps timeline.

 

I haven't tested that MPE lens for smallest size. I may check to see if it can cover a S8 frame as well.

 

I had originally asked Bruce to put an LED backlight in the head. But soon, I found out that LEDs don't produce full spectrum light. While that might be considered a small point, I wanted the truest light possible, so we went with halogen sources. Of course, they're hot so you have to mount them externally. You have to bounce them around to get them to the backlight reflector. That is a real project all on its own since managing the angles and keeping the light even across the backlight surface is a challenge. But, I wanted pure light and I got it, headaches along with it.