Homemade 35mm Adapter

[Bob Hart]

That line of investigation has pretty much proven a dead-end although I think someone with a very small camcorder has claimed to have succeeded with it. You may find by searching this site for "roof prism", discussions where this option is mentioned.

[Jonathon Wilson]

Yeah - I thought about that as I took my Canon AE-1 apart when I was first starting. The problem is the viewfinder is tiny.... the prism ends up with an exit aperture which is down in the 10mm across range and it's very hard to get the camcorder focused on it.

I mount my camera upside down and use a $20 external b/w CRT monitor, mounted on the left side, like a viewfinder. It's a 5" diagonal and the CRT is much better for focusing than LCD. It works great. Should have pics and things in a couple days.

[Thomas Smet]

how do cameras work to go onto the 35 mm film then? I thought the viewfinder method might be just a little to small. Does the image go through another prism to output to film? I would take apart my camera to take a look at it but my wife would kill me. Also I thought maybe of not using the actual parts inside a camera but creating a larger version based on using the design as a reference. If you flip the camera upside down isn't the image still horizontally flipped? Maybe if we forget the prisim the mirrors could at least fix this problem. Then if we flip the camera or external monitor upside down then everything will be correct.

[Jonathon Wilson]

Film cameras record their images onto film upside down. Then when you get the pictures back, you turn 'em right side up :)

The viewfinder (at least in a 'reflex' viewfinder) is an 'alternate' path that the light takes whenever the shutter button is not depressed.

In my AE-1, the light comes in through the lens - inverted at this point and hits a mirror that can flip up out of the way. It bounces from the mirror up into a roof prism and its exit aperture is right next to a powerful mangifying lens that you look into to see the image.

Once you press the shutter down, the mirror flips up out of the way and the (inverted) light goes straight back to the film, without anything else in the way, recording inverted.

So - the viewfinder is not actually the same path of light as that which goes to the film. In the Canon AE-1, for example, there is a fresnel in there to help reduce hotspot - but that also means the visible rings of the fresnel in the viewfinder image -- at least when you magnify it enough. Recording the viewfinder image directly is not a good solution primarily for this reason.... yucky fresnel rings in most.

[Jonathon Wilson]

Here's a good picture of the inside of a camera like mine -

http://entertainment.howstuffworks.com/camera7.htm

[Thomas Smet]

I was taking a look at the images from that site and may have thought of a way for it to work without a prism. Instead I can use mirrors in the place of the prism. If they are at the correct angle they should bounce properly. The design would almost be the same as how the viewfinder works on a camera. Instead however large mirrors would be used to keep the image the same size as it would be with the ground glass method. We may still need the rotating ground glass however. Instead it would rotate on its side to take the place of the translucent layer of glass in a film camera that the light hits before the prism. The mirrors however should at least fix the upside down image. I'm not sure about the backwards image however. Basically we would just be making a slightly larger version of what is in a viewfinder for a photo camera. It isn't actually all that complicated to use mirros to filp images. It is basic high school geometry. I use to do lab experiments in school like this. We once took everybody in the class and gave them all mirros and placed them at different points and angles in the room. We then bounced an image between 30 people and projected it onto the wall. I do realize however it is much harder to get this stuff working inside something the size of a lens then it is to do it in something the size of a classroom. I might be able to come up with a diagram of the design if somebody else wants to build it.

Another thought I had was to take a second 35mm lens and rip it apart to take one of the convex/concave lens out and use that before the image hits the ground glass. So basically if a camera lens usually has 3 lens inside of it we would create a lens with four lens instead which might flip the image back to normal. I am not sure at this point if this would introduce any more distortion to the image because the lens would have been optimized for the way it was designed. The reason I thought this may work is because when light passes through the first lens it is flipped. Then it goes throught a second convex/concave lens and flipped back to normal. Finally the light hits a third lens and once again gets flipped. The forth lens should hopefully flip it back again. Note I don't know the exact number of convex/concave lens inside right now. I just threw three out there to test the theory.

[Bob Hart]

I had a go at replicating the porroprism (90degree) prism and roof prism arrangements with simple mirrors between the taking lens and the subject.

Mirrors are viable but in that position the assemblies, (especially the roof prism arrangment) have to be too large to be useful in a practical sense.

The mirrors also must be surface coated otherwise you end up with all these beautifully creative but otherwise useless parallel lines writ all over your image.

Between the taking lens and the GG is not an option for either due to the field of view/focal length combination we are dealing with.

Between the GG and the camcorder is an option with 90 degree prisms. Surface coated mirrors should also work. Any roof prism or replicating mirror path which would yield sufficient field of view for the prosumer camcorders many people are using would have to be be too large to be a practical proposition.

This is not to say it is impossible. It is simply my opinion relatively unqualified on the topic. I did not do any more than superficially examine the possibilities with mirrors and decided not to go any furthur.

[Thomas Smet]

I found a right angle prism that is 50mm x 35mm x 25mm. That should be big enough to give us a larger image for the video camera. I also found some nice 50mm x 50mm optics mirrors that could be used. These objects are a little large right now but they could be cut down if needed. I figured size was going to be a restriction. I am almost positive this can be done will simple math and science. Yes the adapter might be pretty large but it would work. It would at least be smaller than a rig to flip the camera upside down. If I can make it to not have to use the rotating ground glass then it would be smaller than the current boxes people are making. I think I might still need the gg though to enlarge the image. I thought of getting a magnify mirror to enlarge but it might add to much distortion. I haven't done anything like this before but it should work "if" I can make everything small enough with the needed precision. Good thing my dad owns his own machine shop where he has tools to cut things down to the micrometer. The problem right now is that I understand the math and science of the optics but not the rest of it.

[Bob Hart]

Thomas.

You might have a look at www.dvinfo.net/media/hart and anything in the files there which refers to "erector" or "prism erector" which illustrates where I have been heading in the matter. None of these diagrams should be built to as they are wrong. The hypotenuse face of 52mm referred to is incorrect. It should have been 56mm.

I have a working erector version of the Agus35 using prisms but am waiting for a 7+ achromatic dioptre. Currently there is far too much chromatic smear and muddyness in the colours for it to be of any use.

In my tests to work up this version, I found that the right-angle mirror arrays in ninety degree opposition, needed to be as if they were 40mm x 40mm x 56mm prisms to permit sufficient coverage of the 24mm x 18mm 4:3 academy motion picture frame. This arrangement would just fit within the focal length permitted by a 7+ stack of Hoya close up lenses.

The 56mm hypotenuse face gives you two half faces of 28mm which is just enough to provide 24mm less 2mm to allow for disk circumference and safe clearance of the front prism/mirror edge.

Allowance also has to be made for a face contact for mounting which takes the form of a plane face with a hole in it for the image area. Anything approaching correct alignment is otherwise impossible.

My prisms were 40mm wide across all faces as this was the version economically available to me. You might get away with 30mm wide across all faces but this would require a custom build thus more expense.

25mm wide across all faces might be a bit tight. You will be left no headroom for poor build quality of the adaptor and this would put it out of reach of many kitchen-table home builders which is one objective of this exercise.

I am not an optical engineer or any sort of engineer for that matter. Hopefully these comments may be of some help in your researches. Please keep us posted on your progress.

[Steven Fokkinga]

Hi all,

Does anyone know if and where you can buy this: http://www.lilliput.cn/207.htm
seperately? I've seen tons of the 7" versions around on ebay and other retailers but i can't find the 5" versions... They are X/Y flipable so great for an agus35.
Also, the resolution of these screens is 960 x 234 = 224,640 pixels so comparable with the viewfinder of for example a vx2100 (211k pixels), this could be enough for focusing (imho it's possible to focus on a vx2100-lcd at normal conditions). Or am i wrong?

Thanks a lot.

[Bob Hart]

The good news is that an erecting Agus35 is achieveable. There seems to be no rainbows when the +7 achromatic diopter is used. So far the rough tests have been in low light and there is much work to be done yet.

The bad news is that it is a much trickier precision build than the more or less 5mm (1/4") build error tolerance of the non-erecting Agus35.

The prism path will not tolerate bad alignment and some form of precision adjustment is mandatory.

The timber (craftwood) mounting method is pretty much of a dead-end for long term stability.

Cold-casting nolathane in a precision mold may be an alternative. The other alternative is having a precision prism mount machined = expense.

The SLR lens mount, camcorder mount, prism path and disk, must be maintained in a closer alignment along three centerlines, all of which have to be maintained in close alignment with each other.

Provided the home-handyperson can work to the skill level required, the erecting version can still be an economic alternative but if it is to be entirely shopped out, then for cost-effectiveness, the professional products either as a purchase or rental investment are probably a better option

[James Hurd]

Does anyone have a good idea of what kind of DCX/PCX I should use to kill the blue high-lights around the edges of the glass? I'm using the spinner. Everything seems to be working great except for the high-lights around the edges... Will post some pics soon!

Thanks for any help!!

[Bob Hart]

James.

Brett Erskine seems to have the most clues on this subject.

Here's a list of questions which may help pin it down a little which hopefully some others with the knowledge will respond to.

What sort of close-up lens are you using between the disk and the camcorder? (Simple singe-piece close-up or macro add-on lens) or more complex achromatic diopter ( an add-on lens made of two or more pieces of glass ).

What camcorder are you using?

What size image are you framing with your video camcorder on the projected disk image? (24mm x 18mm academy motion picture 35mm film frame or the almost 2x larger still-camera 35mm film frame.

What aperture (f stop or t setting) are you using on your lens on front of the Agus?

The achromatic diopter matched to the camcorder or the 2" (50mm) telescope eyepiece -(unfortunately only 44mm effective diameter) seem to take care of the rainbows.

From my understanding of posts which refer to a PCX, this may be a separate condenser lens intended to minimise or eliminate the center hotspot problem (the dark corners). My understanding is that a PCX will not eliminate chromatic errors already existing at a previous stage in the image path.

[James Hurd]

Thanks for the response Bob,

1. The macro is a Hoya +15 diopter I believe. (Could be a 10+. It's the largest they make for a 72mm thread).
2. DVX100A
3. Haven't measured that yet. (Interesting... 24x18)
4. It's a canon EOS. Focus and Zoom.

I'm using a DCX 50mm Diameter with a Focal Length of 250mm.

DVX-DIOPTER-DCX-GG-Lens


Interesting point on the frame size. Maybe I'm not in far enough..

Some Pics:
http://www.sunrushmusic.com/1.jpg
http://www.sunrushmusic.com/2.jpg
http://www.sunrushmusic.com/mini35_cap1.jpg
http://www.sunrushmusic.com/mini35_rack2.wmv (Shaky video. Using a different lens and hadn't got everything fastened down yet...Encoding maybe wrong too.)

[Bob Hart]

James.

That's a serious looking piece of hardware there. did you machine it up yourself or buy it in?

From looking at the pics, the cam to groundglass distance seems about right for 10+. I can't say for sure but your close-up lens seems to be a single element, not an achromatic. If it is no thicker front to back than about 1/4 to 3/8 of an inch, it is likely to be only a single element.
You can expect to get rainbows on the outside and some softness in the corners depending upon how much area of the groundglass image you are taking in.

There seems to be softness in the lower left corner which corresponds to the right upper corner of your cam's CCD array. I can't tell whether it is your Canon EOS lens center axis being parallel with but not being centered upon the center axis of your camcorder's optical path, or a slightly skewed groundglass throwing the SLR or the camcorder focus off in the left of imager.

It seems to me that the camcorder is framing the projected image about 2mm (1/10th of an inch approx) to right and slightly higher than center but the subject could also be soft in that area which would be correct. If your cam and Canon EOS lens center axes are correctly centered, you may have a very slight bend in the optical path to the right and downwards.

There is another factor which may be in play. The subject may be moving from left to right, which would soften the closer objects in the image more. Due to the superior image of the progressive scan, interlace artifacts are not there to tell me if this is the case.

There is another factor which occurs in the Sony DSR PD150/VX2000 and DSR PD170/VX2100 camcorder families. I understand the GL1 may have a similar issue but I don't know for sure.

Whilst it does not degrade the camcorder image in normal operation, the center axis of the optical path of the camcorders does not co-incide with the image center of the 3 x CCD array but is offset to the right which sets the image frame off to the left as viewed.

You will see this offset in some night-vision news-footage where in difficult lighting circumstances, the intensifier display has been deliberately vignetted by zooming back to enhance apparent resolution.

Because of this effect with the PD150, I found that setting up the centers by viewing through the cam itself did not yield as clear an image as setting up the centers and alignment by mechanical methods and measurement. This is not to say the Panasonic optical path is not centered to its 3 x CCD array, but it would be worth testing for if you want to get into that last 15% of sharpness, before you start trying other remedies.

That area of softness to the left in your .jpg image is consistent with the softness I get with the PD150 on the left margin if I am not zoomed in quite right enough to the groundglass in the AGUS35 or the intensifier display in the NV.

It would be interesting to set up a test with a center mark aligned mechanically with the center axis of the optical path and then shoot a test to see where the center point falls in the aquired image. For this test, you would only use the cam and the close-up lens to film the test card, not the rest of the Agus appliance.

Don't take all this too seriously as it is almost all speculation and theory. I am also neither engineer nor industry professional.

Due to my woeful landline speed I did not attempt to download the .wmv file.