Build Test Outline


#21

Matthew. Thank you for your response. I was suggesting the cogbelts as a possible option for your design. I actually have a four-plate 16mm Steenbeck editor which I want to modify as a a flash scan machine. Your link looks to be a good source of replacement belts as the existing belts whilst still working are a bit soggy from years of use and grease getting on them.

The optical path from a camera could be a bit tricky for me. The SI2K has a 2/3" sensor capable of 2K apparent resolution. It is actually a bit softer than that by the time an OPLF and debayering has had its way but for late sixties to late eighties 16mm reversal and workprint, it would be adequate.

I found when making AGUS35 adaptors that an approx 24mm image width required approx a 40mm zoom setting on a half-inch sensor camera like an EX1 with a 4+ achromatic dioptre in the path to enable close focus from about 7 inches which includes the bent path through a set of erecting prisms.

With the SI2K’s 2/3" sensor, I will need a longer focal length lens and a dioptre to get a close view on the 16mm film image through the split wheel of the Steenbeck. The original Steenbeck projection lens may be capable of throwing an image which may be resolveable into the camera with a technique similar to an auto-collimator. The lens on the camera then need only be the same or very near to the focal length of the Steenbeck’s projection lens.

The wrinkle in the mix is that the Steenbeck projection lens is very compact and may porthole the image of a more common c-mount lens of the same focal length. The Steenbeck lens is placed such that the path of focus through the shutter prism of about 40mm diameter is correct to the film plane. With the prism shutter out of the path, the placement of this lens in the original mounting will be incorrect and it may no longer satisfactorily project the film image in the same scale.

I’m just going to have to suck it and see once I get a lamp working. The original Steenbeck lamp is not working.

In the Sony 90 degree mirrored projector to 2/3" video telecine arrangement, there is a 16mm C-mount lens, a forward offset of about 5mm to achieve closer focus upon the image out of a 50mm projection lens. I got sort-of okay images from it but with the artifacts of a shutter projector drifting slowly in and out of sync with the camera frame rate.

I may try that lens with the Steenbeck projector lens to see how I get on but don’t hold much hope for the arrangement. The 50mm projection lens has a nice wide front element. The Steenbeck projection lens is tiny by comparison.


#22

This clip might be of academic interest. It was a sort-of autocollimator style technique I used to check that the viewfinder screen and focal plane in a CP16 were coincident using a Sony PD150 1/3" video camera looking into the CP16 film camera gate and viewfinder screen through the film camera’s Angenieux zoom lens.


#23

Cool! I remember the PD-150 and used one on several occasions. It was a solid camera! It seems you’ve been hacking for a while! Thanks for sharing.


#24

For general information. I dismantled the optical block of the Steenbeck. After playing with a few lenses and combinations found that the small projection lens of the Steenbeck will project directly onto the camera sensor of the SI2K without the shutter prism in place between it and the film plane or any interposed optic needing to be in place between the front of lens and the SI2K sensor.

Interestingly I found that at some stage in the past, the optical block had been dropped and there was a crush deformation on one sharp corner. The retaining screw for the projection lens was also missing.

In the Steenbeck projection path, after passing through the prism shutter and the lens toward the reverse projection screen, the image is rotated upright by a prism block and reflected off two surface-coated mirrors. To focus onto this more distant target, a correction optic which can be adjusted has been interposed between the erecting prism/mirror and the screen box.

Trying to capture an aerial image from the optical path beyond this optic did not work.

The sharpness of the image is reported by the camera’s focus assist as poor, at best about 25 points.

This may be related to the frame being optically enlarged for side-to-side image coverage with a top and bottom edge crop in the SI2K Super16mm format frame. With a left and right side edge crop for a full frame top to bottom, the apparent resolution might improve if this is dependent upon the sharpness of the image on the film. The film was a one light workprint struck in 1989.

The projection lens itself has no aperture control so is operated fully wide open at its rated f2.8, likely is not as sharp as it could be with a tighter aperture. There is a “goat’s eye” aperture of sorts painted across the rear element of the lens to mask off the distorted inputs from the shutter prism.

The lens optical elements are permanently assembled into a cell which cannot be dismantled so there is no way to insert a fixed iris of say f4-f5.6 to sharpen the image up.

There is enough adjustment available for the projection lens to be moved relative to the film plane to achieve left-to-right edge framing or top-to-bottom edge framing. There does not appear to be any distortion. Because of my jury rigged test arrangement, edge or corner softness was hard to detect as the film image itself had areas of soft and sharp focus.

For left-to-right edge framing, the lens had to be moved forward in its holder 10mm from its normal fully inserted position. The lens is inscribed “Staeble Proj-Trigon 1:2,8/35”. For left-to-right image framing, the distance from the rear element to film plane is approx 62mm. In its normal position the distance from rear element to film plane is about 52mm.

I will try an inexpensive Fujian C-mount 35mm CCTV lens with adjustable iris sometime in the future. This is unlikely to work with the Steenbeck due to the optical axis being very close to the metal platform of the Steenbeck.

However in another arrangement, this lens may be satisfactory and could be arranged closer to the film plane so that a MFT format camera frame would be fully covered by the projected image. There are MFT adaptor kits vended as a set with this lens, the 35mm Fujian found on eBay.

Cheers.


#25

Furthur to the above, I received the Fujian 35mm CCTV lens. Its performance is adequate but its rearmost structure interferes with the delrin optical block of the 16mm Steenbeck flatbed when the lens is positioned for standard or Super16mm projection coverage onto the S2K 2/3" sensor.

I have since discovered that the Fujian CCTV lenses are also available in 25mm and 50mm focal lengths. I tested for frame coverage with a 50mm Optar Illumina camera lens. There is a lot of barrel structure forward of the front element. The projection surface has to be too close to the front of the lens to be visible.

I tried a Nikon 50mm f1.4 prime lens which has less structure forward of the front element. This enabled more distance between the rear of the lens body and the Steenbeck optical block for a Super16mm projection of image onto the SI2K 2/3" sensor size to be observed. It would seem that using the smaller Fujian 50mm CCTV lens will be a more viable option than the 35mm previously mentioned.

The Fujian 25mm will not be workable with the Steenbeck because the rear element has to be much closer to the film plane and would interfere with the actual transport mechanism of the Steenbeck id used in the normal optical path of the Steenbeck.

For a completely new optical path for scanning built from scratch, where there are no transport components getting in the way, the 25mm may be okay. As it is nearer to being a “normal” lens for a 16mm sized image, I would expect that there may be less distortion than a longer focal length lens in this price range might introduce.

I have ordered a 50mm off eBay. The last one took about a month to arrive.

For curiosity sake, I also tried a 75mm Scheider Kreuznach Tele Xenar lens off an old Bolex. This also yields a Super16mm sized image to the 2/3" sensor area of the SI2K with a longer distance between the rear of the lens and the film plane on the Steenbeck. However I know this lens to be low in the sharpness numbers and flarey from previous tests on the SI2K camera. I would not be inclined to use this vintage lens on a scanner.


#26

Furthur to my previous post, here is a frame from a static piece of film projected through the Fujian 35mm CCTV lens at f16. It may yield sharper in the zone of f4-f8. As tests go it is very primitive. The light source was from a common household torch through a 5 micron grind glass diffuser and there are bright and dull spots.

Everything was jigged up with cardboard stacks, masks and sticky tape. The film is slightly wider side-to-side than the image and cropped top and bottom by the camera sensor. It is therefore over-enlarged for a standard 16mm frame and will display softer.

For a $24-00 lens this is probably about as good as it will get. Edge-to-edge is uniformly sharp. There was a lot of dust on the strip of film. The film is a contact one light workprint from negative, so a little sharpness will have been also lost.

It made 63 points at best on the sharpness numbers but that was with the groundglass hard up against the back of the film and its texture showing through. With the groundglass and light moved out of focus, the sharpness numbers fell away to 51 at best. This may be the sharpness of the film image.

I would have expected the specks of dust to yield more crisply but some may be on the reverse of the film slighly off focus. At f16, the diffraction effect may also be in play. The film grain itself is just apparent. If it is printed grain from the neg it will be that bit softer. If it is grain in the print itself, then optically, things could be better than they are.


#27

This is great research, @agus35monk. It will surely be useful when I get to optics. Thanks for blazing the trail.

It would be great if you had a document/chart/etc that detailed what lenses you’ve tried, their prices, notes on the results, and links to where they can be purchased. It would be a great reference for all.

M


#28

Furthur to my previous posts on lenses, the Chinese vendor from whom I ordered a 50mm lens sent a 35mm lens in error. At the price they charge it did not warrant the cost time and effort of sending it back. I ordered another 50mm lens and this time it came okay.

In a quick jig up with a projection onto a white surface, I found the image size can be adjusted as with the 35mm lens and is of comparable quality. With a longer focal length it is more critical as to focus compared to the 35mm. The 50mm allows the longer distance between the lens and the film plane which I need to enable me to clear the optical block of the Steenbeck.

This will enable me to machine up a tube with a plain end to fit into the empty lens space in the optical block of the Steenbeck and with a C-Mount thread on the other to mount the lens. There is also a longer distance between the front of the lens and the camera sensor plane.

This will allow me to machine up another tube adaptor with C-Mount for the camera and filter thread for the lens. For the compact scanner, the 25mm or 35mm lenses may be the better choice as they can be placed closer to the film plane and the distance between the front of the camera ad the film strip can be shorter.

These lens appear to have no coatings, certainly no colour cast that I could spot. They might flare but so long as the image remains sharp, available enhancements in modern digital image editing software should take care of that.

The material I have been using for smaller diameter mounts and adaptors on other projects has been hollow round flowcast bronze metal stock. It machines cleanly, is more robust than aluminium. It does not bind or pick up destructively like aluminium does if a thread fit is a bit tight. The material is used to make bronze plain bearings.

The vendor in Australia was Bohler Udderholm a German outfit. However the last time I tried to buy from them, they were a bit picky. They are accustomed to larger engineering orders. Now that the mining construction boom has fallen away, they may not be so much under the pump to get their big orders out.

I’m busy this week but hopefully a weekend or two hence, I can jig the lens up with the camera and grab a still frame from it.


#29

Furthur to the above, here is a link to the eBay page which has the Fujian lenses listed. Make sure the lens illustrated is the focal length you want to order. In my instance, whilst the lens was listed as 50mm, the image was of a 35mm lens and of course that was what arrived although it was marked on the delivery slip as a 50mm lens.

http://www.ebay.com/sch/i.html?_from=R40&_trksid=p2050601.m570.l1313.TR0.TRC0.H0.TRS0&_nkw=Fujian+lens&_sacat=625


#30

Furthur to the above, I jigged up the 50mm Fujian lens between the Steenbecks optical block and the SI2K camera with the Fujian lens rigged so that its rear faced the film as a projection lens. Because of the necessary far forward offset to achieve focus, the use of an extension tube from the camera to the lens with the lens in its normal rear to camera arrangement, the exit pupil of the rear of the lens vignettes slightly.

As a projection lens, the 50mm fits to the film plane of the Steenbeck “just”. The C-Mount of the rear of the lens had to be screwed into the hole in the optical block which conveniently is a mere fraction narrower than the C-mount thread diameter. In this state, a clean top-to-bottom frame is achieved to the SI2K sensor. The side-to-side frame is a vignette to the SI2K sensor.

In this state, the sharpness numbers are good at 95 and comparable to the real-world sharpness yield of some of my camera lenses but not the best ones. I think the limitation is the actual film image as the sharpness numbers go down as the film image is enlarged to the sensor, moving the lens closer to the film plane of the Steenbeck and repositioning the camera to fill the frame.

The flange face of the lens prevents it from moving close enough to the Steenbeck film plane to achieve a clean side-to-side fill of the SI2K sensor area. It would only require a small approx 0.5mm deep clearance shoulder to be skimmed from the Steenbeck optical block to allow the lens to move rearwards the required distance.

This is encouraging because it means only one tube adaptor need be made for fitment between the camera and the front of the Fujian lens. The offset distance is approximately 62mm for a side-cropped full top-to-bottom 16mm image in the SI2K Super16mm image frame.

This arrangement is not going to work with the Blackmagic Pocket Cinema Camera because it has a larger sensor area than the SI2K. For this camera to be usable it is likely that either the side cropped iamge be acceptable or some material would need to be machined off the Steenbeck optical block to allow the lens to be placed furthur rearwards.

Neither the SI2K or the BMPCC may be suitable for flash-scanning at realtime frame rates because they are not global shutter sensored cameras.

Because the Fujian lens is slightly wider than the base surface of the optical block, it may be necessary to mill a clearance channel about 3mm deep in the aluminium deck of the Steenbeck transport assembly. It is not a biggie but a difficult pest of a task if it needs doing.

For completely novel scanner designs from scratch, the 35mm lens may be the better choice as it requires less distance between front of lens and camera to enable a M4/3 sensor image coverage.


#31

Furthur to my previous post regarding modifying a Steenbeck ST16 for scanning with a SI2K camera.

I have found that I needed to create an approx 0.7mm deep clearance relief in the delrin optical block of the Steenbeck to enable the wider diameter of the C-Mount Funian 50mm lens flange face to rest about 0.7mm inside the block to establish a full side-to-side sensor coverage of the SI2K Super16mm frame.

It is important to be sure not to cut this relief too far because the hole in which the original Steenbeck projection lens is retained need to be preserved if the machine is to be restored to its original function.

It is likely that for a 2/3" HD single-sensor camera that this relief may not need to be cut. A deeper relief may be needed if anyone decides to use the Blackmagic Pocket Compact Cinema Camera for scanning. This camera has a larger active sensor area for “Super16mm” than the SI2K. I do not know if this camera can be hacked for triggered single frame work.

Cutting the relief into the delrin block is a bit of a mission. The block appears to be both cast and machined. The average home handyman is unlikely to have a four-jawed lathe chuck of sufficient depth to accommodate the shape of the block.

With considerable difficulty and care, I have carved out the necessary relief with a dremel tool with a reverse tapered miniature burr bit. Unless you have the dextoral skills of a fine engraver or dentist, it is not a task I would recommend.

The job could be done with a specialised reaming tool which is something I do not possess.

As an alternative to cutting a 4mm deep clearance for the Fujian lens barrel in the metal deck of the Steenbeck film transport, I have examined using shims of approx 4mm to raise the capstan and optical block.

The feed rollers and plattters would also need to be raised. It is likely the film path may tolerate a slight misalignment. However the onwind to the take-up platter is likely to climb and collapse the roll as also would the rewind plattter.

The little 90degree prism/collimator lens which lights the film would also need to be raised with a shim to re-establish the optical centertline relative to the film frame. The alignment from the lamp should not be affected.

When I get time, I will re-assemble the Steenbeck and test the optical path with a faster camera shutter to see if it is workable. Then if that is successful, comes the task of finding or modifying a trigger circuit for the camera and for the LED flash or interrupt panel.

Any detailed advice or circuits and info relating to a suitable LED will be muchly appreciated.


#32

The sharpness yield by the Fujian 50mm lens into the SI2K comes up to 109. The sharpness yield with the lens reversed as a projection lens from the film into the SI2K was 95. This is consistent with the image softness of a one light contact workprint used for the test and maybe the use of the lens reversed in the optical path.


#33

Furthur to my earlier post, if anyone is modding a Steenbeck ST16 complete flatbed editing table or just the transport, the following dimensions apply for a prototype tube to connect camera to optical block via the Fujian 50mm lens set with its own C-Mount flange about 0.7mm recessed into the delrin optical block.

These dimensions should not be taken as gospel as there may be variation between examples of the Steenbeck ST16 and variations demanded by the camera type used and its sensor size. My camera thus far is the SI2K “mini” head.

The “optical block” I define is the portion of the optical path which includes the upper capstan guide wheel and the holder for the Steenbeck’s projection lens.

For convenience in machining the tube, I have chosen the IMS C-Mount for the SI2K camera.

The C-Mount at the camera end is 4mm from tip to flange.

The 34mm ( 37.3mm ) filter mount depth from front of the Fujian lens is 3mm.

The length of the tube body front of Fujian lens to camera C-Mount flange face is 64mm ( subject to change ). Total length including C-Mount and filter threaded tubes is 71mm, also subject to change.

Thread pitch of both C-Mount and filter mount is 0.7mm.

Recommended outer diameter of tube should not exceed lens barrel diameter which is 44mm due to need for clearance of the metal deck of the Steenbeck either by machining a clearance relief channel or raising the film transport path no more than necessary…

Front of Fujian lens to film plane at capstan centre is 113.4mm. ( subject to change ).

This arrangement requires the alloy deck of the Steenbeck transport to have a clearance relief of 4.5mm or more machined into the surface under the lens or the film transport components to be shimmed higher by 4.5mm. The cog belt drive should accommodate the resultant upwards misalignment as it has a generous depth of working surface.

The recommended metal stock is Bohler Udderholm Flowcast hollow bronze which comes in a convenient internal and outer diameter size requiring little machine work to cut to size. There is a hole in the centre which is more convenient for final machining than having to drill a hole through first.

The material machines cleanly with a good finish. Threads fit up cleanly without picking up and tearing if they are tight, unlike aluminium which is less forgiving.


#34

Footnote to the previous posts. The Fujian lenses do have anti-reflection coatings and seem to yield good contrast.


#35

Thanks for the thorough and thoughtful project updates, @agus35monk!


#36

It has been a good long while since I posted here. I have only recently got the Steenbeck ST1600 running again. The drive belts had turned into jello and it needed servicing and a power resistor cooked off within 30 minutes of testing it. I roughly jigged the SI2K camera up using the previously tested 50mm C-mount lens. As a static frame, the image is “adequate”.

The quality of the light is an issue. The halogen lamps have a distinct yellow hue which is near to impossible to colour-grade out. Best results were by using a fast shutter and as little ND as possible to keep the infra-red pollution to a minimum. This introduced a magenta cast when heavy layers of ND were used.

The projection lamp needs to be replaced by something lower power and benign in colour temperature.

I ran the camera and let the Steenbeck do its thing. I believe speed regulation is by analogue means. The Steenbeck and the camera run at slightly different frame rates.

As the SI2K has a CMOS sensor with rolling shutter issues, it cannot be used directly to the film at high shutter speeds and synchro offset to reposition the passing frame. In forward motion of the film, rolling shutter shrinks the frame vertically by almost half. In reverse motion, the image becomes stretched vertically.

So is is necessary to use the polygon prism system in the optical path as the intermittent movement. It flares - a lot and in its history has been injured by somebody with a wedding ring on the finger. Surprisingly, the image does not seem to be severely degraded as to sharpness compared to imaging direct from film to camera sensor but forget about acceptable contrast.

I am trying to upload a test to Youtube but the service seems to have degraded somewhat lately or maybe my computer is too old. The image upload has been “processing:” for about an hour for a two minute clip.

When I have it, I shall post a link here to the footage. It is rough, with hideous colours and transition artifacts from the prism shutter not qujite in sync with the camera frame rate.


#37

Furthur to the above post, here is the youtube link.

The image has been heavily cropped to mask the worst of the prism artifacts and to reframe an offset image which happened due to the camera moving in its jury rigged support.


#38

I have made some better camera mounts and fixed the Fujian 50mm lens core furthur from the film plane with stacked CS-Mount - C-mount adaptors to compensate optically for the polygon prism. Until I have an alternative lighting system of lower power, artifacts from stopping the lens down and blemishes on the ND filters in the lighting path remain visible.

I have determined that it should be possible to conveniently mount an interruptor disk on the prism drive which although it rotates at a different speed, registers back to a zero point on the film drive sprocket every three rotations of the prism drive. There is a bonus in that the prism drive is damped against backlash by a spring-loaded felt brake pad.

The hall-effect or optical interrupt sensor should conveniently mount to the plastic gear cover and thus avoid having to build new support structure. This arrangement will require an added hub extension through an existing adjustment hole in the plastic cover and a longer through-bolt to secure the prism holder. If the prism is eliminated, then a dummy prism holder would need to be substituted.

For now, with the SI2K, the polygon prism remains necessary because the camera is a rolling shutter type and requires an intermittent film movement to avoid artifacts.

Here is a link to another test with the camera in a better fixture. Old B/W film reproduces better than colour at this point in the progress. Cyclic prism artifacts are apparent because the camera is running wild and not slaved to the Steenbeck’s frame rate. Optical resolution is adequate as dust on the film shows sharply. The fixed soft spots are on the stack of ND lighting gel pieces which remain in near-focus due to the deep depth-of-field of the lens at f16. I am thus encouraged to continue.


#39

Furthur to the above post. I have since tried substituting for the right-angle prism/condensor optic in the deck of the Steenbeck, ND filter gels and a simple white matte reflector canted at 45 degrees towards the film. The lighting of the image is much more controllable and the lens aperture can be returned closer to its sweet spot.

Exposure now falls within the range the camera’s manual white-balance can do its thing and colour rendition is much improved. Any bend in the matte-white reflector, a thin piece of white paper, shows up in the image as a darker area. Interestingly I accidentally discovered that allowing ambient light to fall on the opposite face of the reflector, whilst on its own not enough to register in the camera, confers a apparently beneficial effect on the image not unlike that of pre-fogging motion picture film.

Allowing an unfocused indirect ambient light spill onto the camera sensor appears to confer a similar effect. Small motes of dust on the sensor which are unavoidable in this sort of experimentation also go away with direct spill.

So far, I have also found that counter to my expectations, setting the camera gain to minus 3db, which can provoke clipping in the whites is also apparently beneficial to the image. Adding a 1/8th contrast filter into the optical path and maybe even lighting it may provide a means of saving over-contrasted images. I am only a mug playing with lots of unknown in this adventure so please, do not accept these suggestions and observations without a critical eye of your own.

My editing computer has laid down. It becomes paralysed on the “welcome” message. The OS is Windows 7 Professional. So I cannot get you any motion vision. I may be able to extract a frame frab directly from the camera.


#40

Here is another clip, this time from a slightly improved arrangement with uses reflected light rather than lamp light directly to the sensor through the film and shutter prism.

Focus could have been a bit better, I set focus by moving the entire camera head on its mount which is not yet fixed to the table. Vertical adjustments are available via slots in the mount plates whnich are right-angle pieces.

To enable some moveable friction in the adjustment which is very finicky, a spring washer and two flat washers are used as a sandwich along with wingnuts on two long screws. Once the initial alignment is found, the rear two of four screws are also tightened.

The ideal would be to also have vernier screws but that is more engineering on something which may yet need heavy revision so I am not going there for the time being.

I set focus by using scratches on a piece of film set statically in the transport capstan. In motion the film may well position differently. The scratches may have been on the rear face of the film, not the emulsion side so the emulsion surface may be slightly out of focus.

There is not much practical to be done until I get the camera to slave to the Steenbeck’s frame rate. There is not much point if I am unable to get that function to work.

As you will observe, the image is stable. The Steenbeck’s film transport has a backlash damper which is very helpful. The cyclic softness and image stretch will go away once the camera is correctly slaved to the frame rate with a fast shutter.

Reversal film, colour and B/W seems to yield the best results. Workprint is a bit too contrasty. I have yet to try to invert neg images. That in itself is a high science with all manner of challenges.

The originating film stock was Kodachrome reversal, from memory either 80 ASA or 100 ASA daylight balanced. The originating camera focus was soft. It was relatively easy to grade. Workprint is a lot harder with very crashy colours and contrast. The vision is all trivial family stuff and very boring.