Building a Super 8 Transfer Unit

[Scotness]

Alright so after a false start with my Eumig S802 I've just bought a Eumig 610D from eBay



According to super8data.com it's got the following specs

Eumig Mark 610 D

marketed in 1973

standard 8 mm \ super 8 mm film

format change: auto, by lever

lens: Eumig Vario-Eupronet f: 1.3 \ F: 15-30 mm

lamp: 100 W, 12 V, EFP

reel capacity: 120 m

projection: forward, reverse, still

projection speed: 3, 6, 9, 12, 18 fps

film loading: automatic threading

sound: double-system, socket for synchronized sound with tape-recorder

motor: magnetic motor

power source: ' V AC, 50/60 Hz

weight: 6000 g

dimensions: ?

made in Austria by Eumig

The 3 and 6 frames per seconds are the key things for me.

My aim is to first build a SD transfer unit. It was posted this morning so I should probably get it early next week. From there I guess I've got to look at triggering the capture, the actual lense/camera set up and getting a light source with no hot spots behind it.

I was interested to read on Daan Müller's page that Andreas linked to in his obituary that he was working with a "spot light" as that gave a sharper image than a diffused light source. From what I've been reading most people who have built their own transfer unit have been going for a diffuse light - would anyone like to chime in with their experiences or thoughts

Thanks
Scot

[hirudin]

I have no experience in building transfer units, but in photographic enlarging, using a point light source results in higher local sharpness & accutance, but it also enhances grain, scratches and other defects. Diffuse light sources reduces this effect; you still have the same sharpness (you're still focusing the lens on the negative) but the result appears softer because the image is essentially masked. Alot of photographers will argue that there's not much actual difference, but part of the issue is that when people think they're using point sources, they're actually using light bulbs with frosted surfaces or large filaments, which increases the source radius and thus the diffuse qualities of the light. So if you want a point source, really devote yourself to researching the smallest light source possible to enhance the image; that way you'll get the sharpest image possible, but you may have to do some extra post-processing for scratches etc.

Hope this helps.

-Jesse-

[Scotness]

Thanks for that information Jesse I'll keep it in mind.

Scot

[RCBasher]

With all due respect to the late Daan Müller, his light solution was pretty poor and a compromise due to the constraints of having a continuous motion transfer system (CTS).
It consisted of simple white LED array device and an opal glass diffuser. A CTS system requires exposures of just a few 10’s of microseconds at 25fps or vertical blur will become apparent. If the original image is not too sharp in the first place, one may get away with up to 100us exposure. To put this into perspective, using a sensor with a vertical resolution around 1000 pixels, in a CTS system running at 25fps a point on the film will travel 2.5 pixel lines in 100us.

This meant the Müller scanner needed every bit of light available in order to keep exposures as short as possible. Heavy diffusion would attenuate the light intensity considerably, so this could not be afforded in the design, so what better than to make a case for less diffused light in order to cover for this problem?

Unfortunately the inadequate diffusion also resulted with very uneven frame illumination, being bright in the middle and fading rapidly towards the frame edges.

For a stop motion system, like with a modified projector, we have loads of time for exposure, as much as 30000us if needed, so nice diffusion is not a problem.

For the Müller CTS system, I have designed a replacement lighting system using RGB LEDs which flash at the equivalent of over 10x the power of the original LED system. It allows for heavier diffusion for an even field, but I have had to also use a lens to keep the intensity up, which tends to lessen the diffusion effects a bit. With a wetgate transfer, it is probably an ideal compromise where speed is an issue. It is currently in test (but not in a CTS scanner) with someone well known in this forum.

BTW Scot, there is some information about the Eumig on my website which may help you understand how the speed mechanism works and where to mount a trigger sensor.

Frank

[Scotness]

Thanks for the advice Frank - and your extensive web page! I'll be looking though it alot

Scot

[Scotness]

Okay so the projector has arrived - a bit grubby from storage (cockroach crap etc) but no big deal - I'll give it a clean up tonight. After that I'll start having a go with my macro lense and see how close I can get - that's probably an easy way to start before I have to start looking at electronics etc

[VideoFred]

For the Müller CTS system, I have designed a replacement lighting system using RGB LEDs which flash at the equivalent of over 10x the power of the original LED system. It allows for heavier diffusion for an even field, but I have had to also use a lens to keep the intensity up, which tends to lessen the diffusion effects a bit. With a wetgate transfer, it is probably an ideal compromise where speed is an issue. It is currently in test (but not in a CTS scanner) with someone well known in this forum.

It works perfect, Frank

The spreading of the light over the entire frame is very good. And because of the very short exposure times, all possible mechanical vibration blur is eliminated. The result is a very sharp picture. Eventual film scrathes are more visible, but I have noticed that the correct alignment of the Ledlight tube is important. And of cource a wetgate transfer eliminates this problem.

Fred.

[VideoFred]

The 3 and 6 frames per seconds are the key things for me.

Scot, you should realy consider a machine vision camera with trigger input. Then you will be able to transfer true progressive frame-by-frame at variable speed between 0 and 20fps.

Fred.

[Scotness]

Thanks Fred - I always thought they were a bit out of my price range though -- what kind of prices have you found them for?

Scot

[VideoFred]

The cameras from The Imaging Source are relative cheap. I have used the DFK 31BF03 for a few years, last time I checked the price for a new camera it was around 400 Euro, software included. Of cource you need a lens too. A 50mm C-mount lens with extention tubes schould do it for the 8mm format.

Fred.

[RCBasher]

They are relatively poor performance too! 460 Euro at the moment but with only 8 bit output option and no gamma or LUT before conversion to 8bit make TIS cameras a desperate choice when for not much more you can get something like the IDS UI-3240CP-C. Downside it is USB3.0 so currently limited software options (but comes well supported by IDS supplied s/w). It is CMOS, but has a larger pixel size than the TIS (5.3x5.3 versus 4.65x4.65) giving potentially higher dynamic range, higher pixel count (1280x1024 versus 1024x768), 12bit output, 60fps and global shutter.

Frank

[VideoFred]

You are the expert Frank! No doubt about that.

I still think the TIS camera is a good option for beginners. It comes with easy to install aquiring software. All my filmtransfer clips on Vimeo where made with the TIS. So the end ressult is not that bad. 8)

It is very well possible that the camera you are suggesting is much better. But no one has tested it for now. To many problems with software and drivers can be a nightmare for average computer users.

Fred.

[RCBasher]

You are the expert Frank! No doubt about that.

Ha, would be nice if that were true - I could give up the day job! :lol:

It is a tricky thing to give advice when it comes to someone spending money, so I think it only fair to point out the good and the bad as far as we know it so a better informed choice can be made. I have a collection of emails now from guys who have TIS cameras, are having problems and then asking for my help... :idea: think I'll send them on to you Fred.

There is no doubt you have managed good results with the TIS, but it has limitations and you can't fix those for one or two hundred extra after buying camera. In this thread, Scot doesn't indicate what is "out of price range", maybe the PGR GH2 Firewire b camera at around 1500 is cheap enough - it would certainly be a couple of orders better and good match for 8mm!

But no one has tested it for now. To many problems with software and drivers can be a nightmare for average computer users.

Well actually I have tested their GigE version. Their own software is ok, where their problem is (which I mentioned above) is compatibility with open standard software - because they are not!

Anyway, the most important advice is make sure one buys a camera with the ability to return it if it does not meet your requirements. You usually get a couple of weeks or so, so best to be informed in advance of what to look for and test and have everything else (like good lighting) in place before getting the camera and doing the evaluation.

Frank

[Scotness]

Thanks for the info guys - I know you'll both recoil in horror but my budget doesn't extend to much - I was planning on using a JVC single CCD miniDV camcorder running to the PC via firewire - I already have it of course so don't have to spend anything - and that camera has always had lovely colour rendition on it

Scot

[Scotness]

Hi Frank and Freddie - I'm just after a bit of advice re the lense assembly - this is one part I'm a bit unclear on. I note on your web pages you both use extension tubes and a macro, going straight onto a ccd in a machine vision camera.

I've been experimenting with my camcorder with a macro lense and can almost fill the whole frame of the camera with the Super 8 image - what would I have to do to make it fill it - just use some extension tubes? If I was stick with using this camcorder what would be the ideal way set up to get the image right?

I've also been thinking about an HD webcam instead - assuming I could take the front of it off and then use an image path similar to what you are doing now.

Any advice/guidance on this would be great - thanks

Scot