DIY telecine using DSLR?

Karel Bata · April 3, 2009

I'm looking for a way of making a camera rig that will hold a DSLR and focus on an area the size of the negative image area of 35mm motion pictures (22x16mm) and Super 16 (13x8mm). This will be mounted on a rig that slowly transports the film, and triggers the camera and a flash. The images will be recorded via an HDMI feed to a PC which will be running software originally designed for professional stop-motion animation which will automatically assemble an HD file.

So I'm looking for:
1) a suitable camera rig? It needs to be micro-adjustable.
2) a suitable lens? It needs to be pin sharp, with no vignetting etc..
3) a suitable camera? It doesn't have to be DSLR, but that's what comes to mind, and anyway it would be nice to purchase a D90... :photo

I'm wondering if a rig built for a microscope could be adapted...? :scratch Basically a sort of macro version of a rostrum camera...

Any ideas? :thumb

cmason · April 3, 2009

you may want to check out RedRock Micro, which is primarily geared to video on dslr:

http://www.redrockmicro.com/redrock_dslr.html

for panoramic automation, the Gigapan system looks really cool (awesome pic of Pres Inauguration on the web)

http://www.gigapansystems.com/

edit: re-read your question, and I suppose I mis-understood: are you looking to use a dslr to photograph a negative? Why not just use a video camera and capture stills?

ziggy53 · April 3, 2009

Transferring negative film stock into digital is probably best accomplished with a dedicated film scanner and a film strip adapter. You would further have to adapt the hardware if you want automation but you could probably accomplish that with a custom film gate and custom transport mechanism. None of this will be cheap or easy.

What is the project and what quality level do you expect?
Will you be using older/archived film?

Karel Bata · April 3, 2009

Do you know how much a 2K Northlight scan costs?

Building a scanner may well be cheaper than paying for a scan! It's not just me doing this, there's several cinematographers dotted around the world who are talking about it. One guy in Oz has built a scanner using an adapted Mitchell 35. He then batch processes it all in PS.

I'm thinking that an intermittent movement is unnecessary and exposures can be taken using a continuously moving film (maybe using an adapted old flatbed editor - going cheap these days!) and a flash gun that will freeze the movement. TV telecines in the early days were done using a flash, so it's not anything new. Also software exists for turning an image sequence directly into an HD stream - like Stop Motion Pro.

My concern here is the image capture end of it. Probably a DSLR, a suitable lens, and a mount. I recall a dedicated film slide copier we used to have back at college. Maybe something like that...

I have a 35mm short I could do with transferring cheaply, and I fancy toying around with HDR moving images shot on film.

ziggy53 · April 3, 2009

A dedicated film scanner, like the Nikon Super CoolScan 5000 ED with the SA-30 roll film adapter would be a good platform to adapt. It already has compensation available for different film stocks and Digital ICE technology to handle dust and scratches. If you try to use a conventional camera with negative film stock you will run into:

Negative film Color mask removal.
Limited gamma, which might require a bias light to bring up the shadow values for proper dynamic range.
Dust and scratch removal.

If you are trying to create new "stop motion" movies why not just use the digital camera to do the initial image captures. It's not like it's anything new, ala Tim Burton and "Corpse Bride":

http://www.stopmotionworks.com/articles/cbrdstrpdbare.htm

Karel Bata · April 4, 2009

I'm not actually interested in doing stop motion animation, only in using the dedicated software for what is an identical process as far as the hardware is concerned.

Mucho thanks for the link. Very interesting.

I notice "We originally selected the Nikon D2H ... However, random noise was visible as pixilation in dark areas when the shots were played back as a movie. This pixilation effect was only visible in stop-motion photography, an application the Nikon hadn’t been designed for." I don't understand why that should be - it's just taking a succession of JPEGs isn't it? And do more recent Nikons, the D90 in pardticular, exhibit this prob?

Also "having so many JPEG still frames seems to challenge the system. That demand is unique to stop-motion production. Adding more RAM didn’t help, so to cut down on JPEG overhead, they were converted into QuickTime." And that's just for an animation sequence of a few seconds! Hence the desirability of converting to a HD video file on the fly with dedicated stop-motion software.

Negative film Color mask removal - a photoflash is already at daylight and more blue gel can be added.
Limited gamma - apparently not a problem with the guy from Oz.
Dust and scratch removal - Absolutely. A major worry this, that of removing any dust during the scan to avoid further scratching. For removing existing dust and scratches there are tools in After Effects, though they are tedious to use.

Nikon Super Coolscan 5000 ED?

Expensive. Besides, the image area of a 35mm movie (22x16mm) is substantially less than that of 35mm still film so half of each scan would be thrown away, and 35mm motion picture film is one frame per 4 perfs and at a 90degree orientation relative to stills (sideways

). Anyway, all the mods required would invalidate the guarantee!

You may find one of the threads where we're talking about this interesting Custom DIY telecine

Karel Bata · April 4, 2009

cmason wrote:

Why not just use a video camera and capture stills?

Hmm.. Sometimes a solution stares you in the face. Not sure how that would work in terms of linking up all elements: camera exposure; flash; computer software. Also not sure if the image quality at a macro setting would be good enough, or if other artefacts, like those reported with the Nikon 2DH, would creep in. Worth looking into though.

Still need a rig.

ziggy53 · April 4, 2009

It would be helpful if we knew your application and quality expectations?

LiquidAir · April 4, 2009

Lens:
You want a good quality true macro lens... I have a Canon 100/2.8 macro and it would work fine for this.

Camera:
Look for a body which can capture across USB directly to a computer. Right now you can pick up a used Canon 5D pretty cheaply (that's what I shoot with, so I know it) and it'll give you very nice captures.

Lighting:
If you use a strobe, you can certainly get away with a continuous drive. However, you may have trouble with consistency. Cheaper strobes can vary by half a stop from pop to pop. High end studio rigs (like Profoto) advertise 1/10 stop repeatability, but they are quite expensive. For your purpose the first light I would look at would be an Alien Bee B400. It has a shorter flash duration than most mono lights (which will help stop a continuous drive) and Alien Bees are well regarded for consistency. One way or the other, you will likely need to do some amount of software exposure compensation to avoid flicker in your video. That may be a bit of code you have to write yourself as a Photoshop plugin.

Triggering and drive:
So here's what I'd do: I'd build a continuous drive system using a clock motor with two optical triggers; one for the camera, the other for the strobe. The camera would be set to a longish exposure time (say 1/2s) and the second trigger would fire the strobe would fire while the shutter is open. One way to do it would be with a IR photosensor detecting the sprocket holes and a simple counter circuit to trigger the camera and strobe.

MT Stringer · April 5, 2009

I'm not sure what you are talking about so I thought I would throw out this link for camera mounting plates.
http://overxposed.com/oxminiplate.html

Mike

LiquidAir wrote:

Lens:
You want a good quality true macro lens... I have a Canon 100/2.8 macro and it would work fine for this.

Camera:
Look for a body which can capture across USB directly to a computer. Right now you can pick up a used Canon 5D pretty cheaply (that's what I shoot with, so I know it) and it'll give you very nice captures.

Lighting:
If you use a strobe, you can certainly get away with a continuous drive. However, you may have trouble with consistency. Cheaper strobes can vary by half a stop from pop to pop. High end studio rigs (like Profoto) advertise 1/10 stop repeatability, but they are quite expensive. For your purpose the first light I would look at would be an Alien Bee B400. It has a shorter flash duration than most mono lights (which will help stop a continuous drive) and Alien Bees are well regarded for consistency. One way or the other, you will likely need to do some amount of software exposure compensation to avoid flicker in your video. That may be a bit of code you have to write yourself as a Photoshop plugin.

Triggering and drive:
So here's what I'd do: I'd build a continuous drive system using a clock motor with two optical triggers; one for the camera, the other for the strobe. The camera would be set to a longish exposure time (say 1/2s) and the second trigger would fire the strobe would fire while the shutter is open. One way to do it would be with a IR photosensor detecting the sprocket holes and a simple counter circuit to trigger the camera and strobe.

Karel Bata · April 6, 2009

Broken link MT.

Ziggy, I want to transfer film (the moving kind) to a suitable video file. Maybe HD for Super16, or something higher in resolution for 35mm. At the moment I'm just exploring the possibilities, throwing out ideas and seeing what's possible. I'll probably settle on Super16.

Like here's a wild and crazy idea: how good are enlarger lenses? The kind for making traditional prints from neg. I've got a decent color one gathering dust, and it occured to me that maybe the film could be transported through the head (with an appropriate pressure plate), a flash bolted on top, and a video camera pointed at the print area. Unfortunately the image then goes through two lens systems, but it might just work!

Very nice sites you have LiquidAir! You want a good quality true macro lens... I have a Canon 100/2.8 macro and it would work fine for this.
Very good lens indeed, but if it's only to be used for one application, with one focus distance, aperture, etc. maybe it's overkill...

Camera: Look for a body which can capture across USB directly to a computer. Right now you can pick up a used Canon 5D pretty cheaply (that's what I shoot with, so I know it) and it'll give you very nice captures. Ah, it would be very nice to own that camera! Any idea how long it takes to transfer an entire RAW image over USB? Does the 5D have a HDMI output?

High end studio rigs (like Profoto) advertise 1/10 stop repeatability, but they are quite expensive. That's terrible! The ones I've worked with on Photosonics shoots are much more accurate, but as you say expensive.
One way or the other, you will likely need to do some amount of software exposure compensation to avoid flicker in your video. Oh dear. This may be a major stumbling block. However, as with the lens, the flash would be used for only one application, so I don't need one designed for all the uses a photographer would put it to. I'm sure there must be a cheap way to go. It's the power supply that determines accuracy isn't it? Would a HUGE battery acting as a buffer help fix that?

I'd build a continuous drive system using a clock motor with two optical triggers; one for the camera, the other for the strobe. Why two triggers? Why not just send a trigger to the camera, which would then trigger the strobe irself? Of course with a video camera it would get more complex.

A lot to think about. Thanks guys.

I'll do some tests with the enlarger when I get a few spare hours.

MT Stringer · April 6, 2009

It was working when I posted the link. Their site must be down.
You probably don't need any of that equipment anyway. They make plates and accesories for mounting cameras used in remote mounting situations, etc.

Mike

Karel Bata wrote:

Broken link MT.

Ziggy, I want to transfer film (the moving kind) to a suitable video file. Maybe HD for Super16, or something higher in resolution for 35mm. At the moment I'm just exploring the possibilities, throwing out ideas and seeing what's possible. I'll probably settle on Super16.

Like here's a wild and crazy idea: how good are enlarger lenses? The kind for making traditional prints from neg. I've got a decent color one gathering dust, and it occured to me that maybe the film could be transported through the head (with an appropriate pressure plate), a flash bolted on top, and a video camera pointed at the print area. Unfortunately the image then goes through two lens systems, but it might just work!

Very nice sites you have LiquidAir! You want a good quality true macro lens... I have a Canon 100/2.8 macro and it would work fine for this.
Very good lens indeed, but if it's only to be used for one application, with one focus distance, aperture, etc. maybe it's overkill...

Camera: Look for a body which can capture across USB directly to a computer. Right now you can pick up a used Canon 5D pretty cheaply (that's what I shoot with, so I know it) and it'll give you very nice captures. Ah, it would be very nice to own that camera! Any idea how long it takes to transfer an entire RAW image over USB? Does the 5D have a HDMI output?

High end studio rigs (like Profoto) advertise 1/10 stop repeatability, but they are quite expensive. That's terrible! The ones I've worked with on Photosonics shoots are much more accurate, but as you say expensive.
One way or the other, you will likely need to do some amount of software exposure compensation to avoid flicker in your video. Oh dear. This may be a major stumbling block. However, as with the lens, the flash would be used for only one application, so I don't need one designed for all the uses a photographer would put it to. I'm sure there must be a cheap way to go. It's the power supply that determines accuracy isn't it? Would a HUGE battery acting as a buffer help fix that?

I'd build a continuous drive system using a clock motor with two optical triggers; one for the camera, the other for the strobe. Why two triggers? Why not just send a trigger to the camera, which would then trigger the strobe irself? Of course with a video camera it would get more complex.

A lot to think about. Thanks guys. I'll do some tests with the enlarger when I get a few spare hours.

LiquidAir · April 6, 2009

Karel Bata wrote:

Like here's a wild and crazy idea: how good are enlarger lenses? The kind for making traditional prints from neg. I've got a decent color one gathering dust, and it occured to me that maybe the film could be transported through the head (with an appropriate pressure plate), a flash bolted on top, and a video camera pointed at the print area. Unfortunately the image then goes through two lens systems, but it might just work!

Very nice sites you have LiquidAir! You want a good quality true macro lens... I have a Canon 100/2.8 macro and it would work fine for this.
Very good lens indeed, but if it's only to be used for one application, with one focus distance, aperture, etc. maybe it's overkill...

You can certainly improvise a solution, but make sure you seal the camera body to keep dust out. I took some photos of the moon hand holding the camera in the focal plane of a telescope and in mere minutes my sensor was coated with dust bunnies.

Camera: Look for a body which can capture across USB directly to a computer. Right now you can pick up a used Canon 5D pretty cheaply (that's what I shoot with, so I know it) and it'll give you very nice captures. Ah, it would be very nice to own that camera! Any idea how long it takes to transfer an entire RAW image over USB? Does the 5D have a HDMI output?

I don't know what the continuous shooting rate is for a 5D for over USB. The 5D does not have HDMI out; for that you might want to go with the newly anounced Rebel.

High end studio rigs (like Profoto) advertise 1/10 stop repeatability, but they are quite expensive. That's terrible! The ones I've worked with on Photosonics shoots are much more accurate, but as you say expensive.
One way or the other, you will likely need to do some amount of software exposure compensation to avoid flicker in your video. Oh dear. This may be a major stumbling block. However, as with the lens, the flash would be used for only one application, so I don't need one designed for all the uses a photographer would put it to. I'm sure there must be a cheap way to go. It's the power supply that determines accuracy isn't it? Would a HUGE battery acting as a buffer help fix that?

The strobe is driven direcly by a large capacitor and it is the total charge on the capacitor which determines the power. Photographers usually need a variable power strobe which, of course, you don't need. You can get fixed power slave strobes (like this http://www.bhphotovideo.com/c/product/439970-REG/Morris_11146_AC_Screw_In_Slave_Flash.html). They are cheap, but the cycle time is too slow for your purposes and I know nothing about the repeatability.

I'd build a continuous drive system using a clock motor with two optical triggers; one for the camera, the other for the strobe. Why two triggers? Why not just send a trigger to the camera, which would then trigger the strobe irself? Of course with a video camera it would get more complex.

On a continuous drive system, the frame-to-frame alignment is determined by the timing of the strobe pop. In my experience with high speed photography, I get much better results triggering the strobe directly rather than letting the camera do it. For your case, the difference might not matter.

Nikolai · April 6, 2009

For continuous drive and flahs lighting you can use RAIF (Redundant Array of Inexpensive Flashes:-) approach to eliminate lighting inconsistency. Instead of getting one-two uber expensive high end flashes which you run into the ground by using them at a high speed rate (or you'd have to advance the film verrry slow) you can get several banks of el-cheapo low power low cost strobes and channel several of them into each frame at a very low powere level. Tnen you can build a simple switch which would activate only certain bank of those (simplest case would be a see-saw pattern), so the each bank would fire only ever 2d (or 3d or 4th) frame, thus leaving it time to rest even if you roll the film almost at a relatively high speed.
Even if each flash would deviate in its consistency, several of them in a bank would most likely deliver a fairly good and relaible average.
Just an idea..

Karel Bata · April 8, 2009

Great minds think alike it seems http://www.flashscan8.us/flashSCAN8Overview.htm

So my idea of doing it on the cheap with a continuous movement using a flash is a perfectly workable one.

I'm going to take a closer look at this baby.

zoomiefog · April 15, 2009

I have contemplated a similar project. A machine vision camera would work great for this this. These cameras are designed for industrial and manufacturing applications but I have seen people use them for animation, especially stop motion. They usually have c-mount lens mounts. They are available in many different resolutions and usually connect by firewire. As such, firewire cameras are usually compatible with many capture applications, including stop motion software. Edmund Scientifics Optical department has many lenses available in the c-mount variety, including relay lenses like you would need to focus directly on the film frame. For myself I was envisioning modifying a projector to run very slowly and adding a swith or sensor to the frame advance mechanism in the projector to trigger the software to capture each frame. usually through a keyboard stroke like the spacebar using a modified keyboard. Edmund scientific sells machine vision cameras. but many can be purchased online at other places if you google them. My 2 cents. BTW I would consider the longevity of a DSLR. Using it for that many pictures is going to be hard on the camera especially the mirror mechanism. Also what about using high powered LED's for a light source? Flash is limited in frame rate and life span too.

ziggy53 · April 15, 2009

zoomiefog wrote:

I have contemplated a similar project. A machine vision camera would work great for this this. These cameras are designed for industrial and manufacturing applications but I have seen people use them for animation, especially stop motion. They usually have c-mount lens mounts. They are available in many different resolutions and usually connect by firewire. As such, firewire cameras are usually compatible with many capture applications, including stop motion software. Edmund Scientifics Optical department has many lenses available in the c-mount variety, including relay lenses like you would need to focus directly on the film frame. For myself I was envisioning modifying a projector to run very slowly and adding a swith or sensor to the frame advance mechanism in the projector to trigger the software to capture each frame. usually through a keyboard stroke like the spacebar using a modified keyboard. Edmund scientific sells machine vision cameras. but many can be purchased online at other places if you google them. My 2 cents. BTW I would consider the longevity of a DSLR. Using it for that many pictures is going to be hard on the camera especially the mirror mechanism. Also what about using high powered LED's for a light source? Flash is limited in frame rate and life span too.

Zoomiefog, welcome to the Digital Grin.

My former employer used machine vision in a pick-n-place automated packaging machine application. Unfortunately the video cameras that they typically use in machine vision do not generally have enough resolution or color fidelity to qualify for most film to computer digitization projects. Likewise LCD lighting tends to be poor at color rendition due to a rather extreme number of gaps in the output spectrum. (LCD lighting is not a continuous spectrum light source.)

You are correct that flash lighting is subject to a finite number of flashes and there can be a variation in output illumination over the life of the flash tube as well.

jogle · April 15, 2009

This is a really cool idea. I work in Visual Effects for film and am familliar with what you get from high end film scanners. HD for 16mm is overkill and going higher on 35mm just leaves you resolving the film grain with higher detail. All our work goes out to film at 2048 pixels wide (just above 1920 pixels you get from HD) That said, the scanners capture 12 bit colour per pixel so you'd best use sraw on the 5DMkII and down res that to HD.

The idea to use a flash is a good one, If you can get an old projector or film editing table, then you've basically got all the bits you need for the film gate, geared film transport, reels and all those helpfull bits. replace the bulb with your flash and attach a precise stepper motor to your film transport, you can get some very precise motors for not a lot of money. I'd aim for somewhere between 1 and 2 fps so you don't saturate your data link or cook your flash. Uber expensive film scanners only get ~4 fps anyway.

a 100mm lens will be fine even on the smaller frame cameras, extension tubes will help you get more then 1:1 magnification Remeber the film goes top to bottom in a cinema camera and left to right in a SLR. so your 5D sensor is closer to 70mm cinema frame size.

For the camera, I'd go with something a generation or two old. say a 30D or something second hand and cheap. If you're taking photos you're going to wear out the shutter and mirror in no time. The rated acuations
on a 30D are 100,000 clicks. That's only 70 min of film!

LiquidAir · April 18, 2009

jogle wrote:

For the camera, I'd go with something a generation or two old. say a 30D or something second hand and cheap. If you're taking photos you're going to wear out the shutter and mirror in no time. The rated acuations
on a 30D are 100,000 clicks. That's only 70 min of film!

If you are using a strobe you don't actually need a shutter... I wonder if you can get a cheap DSLR an disable the shutter.

jogle · April 19, 2009

I've seen images from a 1Ds that had a shutter blade that had come loose, it was a big black section diagonally across the image. So I'd imagine you could remove the blades carefully and it'd work.

I'm sure there is some error code for "shutter don't work" so you'd have to be careful to not let the camera know you've disabled the shutter. Maybe it'd be worth testing once you've actually broken the shutter on one

ziggy53 · April 19, 2009

I don't know that you would necessarily need either a mirror box or shutter. Perhaps this would be a good application for the new Micro Four Thirds camera body and a macro lens? Given sufficient illumination noise would not be an issue and I should think there would be plenty of resolution.

Nikolai · April 19, 2009

ziggy53 wrote:

I don't know that you would necessarily need either a mirror box or shutter. Perhaps this would be a good application for the new Micro Four Thirds camera body and a macro lens? Given sufficient illumination noise would not be an issue and I should think there would be plenty of resolution.

I honenstly think this is a project for a mirror-less camera, like a good P&S, or maybe rangefinder...

dokworm · December 15, 2009

Nikolai wrote:

I honenstly think this is a project for a mirror-less camera, like a good P&S, or maybe rangefinder...

I think a microscope camera would be your best bet, especially a cooled one to keep digital noise to a minimum. Something like this: http://www.tsdt.net/Html/ProductView.asp?ID=161&SortID=
Also, if you want and equivalent of a wet-gate system, then do two passes with the film, one using an infrared light for illumination. The infrared will pass straight through the film and give you a 'mask' of the physical scratches and marks on the film. It is a trivial programming task to then use that to remove only the actual physical defects on the film and not the fine details.

ziggy53 · December 15, 2009

dokworm wrote:

I think a microscope camera would be your best bet, especially a cooled one to keep digital noise to a minimum. Something like this: http://www.tsdt.net/Html/ProductView.asp?ID=161&SortID=
Also, if you want and equivalent of a wet-gate system, then do two passes with the film, one using an infrared light for illumination. The infrared will pass straight through the film and give you a 'mask' of the physical scratches and marks on the film. It is a trivial programming task to then use that to remove only the actual physical defects on the film and not the fine details.

Dokworm, welcome to the Digital Grin.

Thanks for the recommendation. Is that camera you linked lacking an IR cutout filter?

DIY telecine using DSLR?

Comments