Building a frame accurate home transfer system

[Konton]

As I'm starting again to work on multiple fun telecine transfer projects, I thought I'd post this article from SUPER8FILMAKER.

Right now I've got a Bolex 18-5L I've ripped the lens, blades, and light out of to turn into a frame accurate transfer system. Before I do any more mods, I'm going to test this magnet and reed switch option they have suggested in the magazine for sending out a signal whenever a frame is in place.

I choose this projector cause it looks pretty cool with an all metal body and a 5fps option for the transfer. Should I be able to get all the mechanical pieces in place, the plan is to place a Raspberry Pi http://www.raspberrypi.org/quick-start-guide with a camera module inside where the light used to be. This will probably be a slow build, but I add updates here is anyone wants them.

[gianni1]

Can you upload some close up photos of your projector?

I'm upgrading an old GAF 2688Z projector by putting in a 2 Watt LED bulb. Some time ago, I had help from the kind users of this forum to upgrade the GAF projector to frame by frame telecine device. I fixed a micro switch next to the film advance claw, then connected the switch to through a pair of PP3 (9V) battery clips to a spliced USB computer mouse which drives a stop motion app to capture the firewire video signal from my DV camcorder. The projector does 6 frames per second.

Gianni 8)

Flickr Picture Set: Upgrading an old GAF 2688Z projector

[Konton]

Thanks for the input. My first plan was to put a magnet on the main shaft here and setup a reed switch close by:



Basically what this article suggested:



I can see that it might be easier to add a micro switch by the advance claw, perhaps in the space right above the claw goes up and down. How would you mount it?



I'd add a raspberry pi with a camera board in the space where the bulb and shutter was:



And an LED with a diffuser will go where the lens used to be:



Thanks again! My understand is your GAF is probably way better for what I'm trying to do. I know Roger uses those. I just thought a metal Bolex projector with a computer built in would look pretty cool sitting on my desk transferring film.

[JeremyC]

Just wondering if anyone has had experience with using a BW camera and taking RGB shots with it of each 8mm film frame in the projector and then combining them? I want to experiment with that to get over debayering loss of resolution but am not sure how much residual vibration there is in the film gate after each frame is advanced for different makes of projectors at different speeds e.g. I was envisioning a system runnning at under 10 frames a second with a camera triggered off at three times that and I was thinking of experimenting with an SD camera first.

Any comments, any experience?

[gianni1]

Bolex is arty, the GAF is just a plastic box. Stick with your design, beauty in life is good.

Using a B&W camera and three combined exposures per frame reminds me of High Dynamic Range processing with digital SLR's.

Also there's the challenge to drive three captures per frame, with Red, Green, and Blue filter or light per capture.

Gianni 8)

[JeremyC]

Bolex is arty, the GAF is just a plastic box. Stick with your design, beauty in life is good.

Using a B&W camera and three combined exposures per frame reminds me of High Dynamic Range processing with digital SLR's.

Also there's the challenge to drive three captures per frame, with Red, Green, and Blue filter or light per capture.

Gianni 8)

Its not too difficult to put together an RGB Led sequential flash system which at the same time outputs a pulse to trigger a camera. My concern is any residual vibration in the film frame while the frame is being catured sequentially.

[Simon Lucas]

As I'm starting again to work on multiple fun telecine transfer projects, I thought I'd post this article from SUPER8FILMAKER.

Right now I've got a Bolex 18-5L I've ripped the lens, blades, and light out of to turn into a frame accurate transfer system. Before I do any more mods, I'm going to test this magnet and reed switch option they have suggested in the magazine for sending out a signal whenever a frame is in place.

I choose this projector cause it looks pretty cool with an all metal body and a 5fps option for the transfer. Should I be able to get all the mechanical pieces in place, the plan is to place a Raspberry Pi http://www.raspberrypi.org/quick-start-guide with a camera module inside where the light used to be. This will probably be a slow build, but I add updates here is anyone wants them.

Konton,

many thanks for posting that article. I'm interested in developing an optical printer and that really helps.

I also have the same Bolex 18-5L. I admit I bought it for it's looks and it's been sitting in it's for years. I'll be interested to see how you get on with your project. Please keep us updated.

[RCBasher]

Just wondering if anyone has had experience with using a BW camera and taking RGB shots with it of each 8mm film frame in the projector and then combining them? I want to experiment with that to get over debayering loss of resolution but am not sure how much residual vibration there is in the film gate after each frame is advanced for different makes of projectors at different speeds e.g. I was envisioning a system runnning at under 10 frames a second with a camera triggered off at three times that and I was thinking of experimenting with an SD camera first.

Any comments, any experience?

My LED system has this ability (one trigger input causes three camera triggers in sequence with R, G & B LED flashes) but to be honest I wouldn't bother. I thought it would be a good idea and mentioned it on my website, the M&Ms image was taken that way, but there are a few difficulties in getting a full system working. Frame registration is the first of course, keeping in mind that unless the RGB frames are aligned pixel perfect then you will lose more than you will gain over using a Bayer sensor. The second problem is knowing which frame is which, not initially obvious as a problem but there is no buried information in each frame to say which colour LED it was exposed with. If the system is run very slowly then it would perhaps be possible for the PC host software to keeps tabs on the captures, but you would get well bored with the slow frame rate before very long. Running at 24fps would be a tad more difficult. I would recommend to use a Bayer camera with +25% pixel count over the proposed mono sensor and a very good de-Bayer routine. The results will be pretty much indistinguishable from the sequential R>G>B method for all practical purposes and significantly faster and simpler to use.

Frank

[JeremyC]

I would recommend to use a Bayer camera with +25% pixel count

A larger no of pixel counts is an interesting idea, would this also depend on the pixel size?

[Andreas Wideroe]

Why don't you just ditch the Bayer cameras and go for a trilinear linescan camera instead? It will give you much higher resolution and much better speed.

/Andreas

[RCBasher]

Why don't you just ditch the Bayer cameras and go for a trilinear linescan camera instead? It will give you much higher resolution and much better speed.

/Andreas

Line scanners leave a lot to be desired for film scanning, especially with amateur hardware. They usually result in vertical weave as producing over a thousand evenly spaced mechanical movements per frame takes some doing.

[RCBasher]

I would recommend to use a Bayer camera with +25% pixel count

A larger no of pixel counts is an interesting idea, would this also depend on the pixel size?

As a rule of thumb, the higher the pixel count the smaller the pixel size. The solution of course is a larger sensor area but the cost goes up. Larger pixel sizes are to be desired for higher dynamic range, the ICX285 sensor (2/3", 6.54um) sets a pretty good benchmark and in a good camera makes any benefits from HDR a moot point, assuming optimum lighting of course. Sony have now moved to a newer technology and on paper promise similar (if not slightly better) dynamic range from the same size of sensor but around 4.5um, effectively giving a nice increase in pixel count "for free" as the sensors seem to be similarly priced. Hope to get to test one some time in July

[Andreas Wideroe]

Just for reference; the pixel size in a Spirit is about 10 um.

[Konton]

Wow. My thread sure got hijacked. Sorry I personally couldn't be of help. I know there are a few camera out there that use 3 CMOS sensors rather than a bayer filter. I don't know if it's worth all the trouble.

I ended up going to a reed switch rather than the micro switch. The microswitch took up too much space and didn't really allow much tweaking. But I have broken neodymium magnets lying around and they are super strong! So I just let it stick on the main shaft where I can move it as necessary until I'm ready to epoxy down.



You can see the green reed switch 1 cm above the broken magnet. I ran 12 volts through to a LED to test out the result. This was the result.

http://www.youtube.com/watch?v=VYj7c_FQLQ4

I still need to mount the reed switch, but seems to work okay. I don't get the camera till August (damn backorders), so not much will happen until then.