So who here does two-shot method stereo photography?

[Grim Tuesday]

Focal length, aperture and LR distance doesn't matter too much so long as you're close. A friend of mine took this idea to the limit, and tried to simulate what it might be like to be a giant, looking at a cityscape in the same way we look at a box of legos and took two pictures, 200M apart from an airplane. His final 3D method was red/blue rather than cross-eyed but if you have any red blue glasses around you can see it really works: http://illusionoftheyear.com/2013/05/through-the-eyes-of-giants/

I've messed around with the cross eyed technique on digital and found success with my phone camera, which is something like an 18mm f2.5 equivalent. And I didn't measure the distance - all I did was move my hand a few centimeters between shots. This is all to say that the process in our brain that creates the perception of stereo 3-d-ness is quite amiable to things that don't exactly match our eyes.

 

[DWThomas]

Reading this thread I had a (dangerous) thought -- for closeup work with TLRs there was a device sometimes called a "paramender" which was a little rack-and-pinion fitting placed under the camera so after composing with the viewing lens, you just turn a knob and it brings the taking lens up to where the viewing lens was to avoid parallax problems. Never had one in my hand -- wonder if they would work horizontally with an L-bracket? It could be quicker than the average macro slide contraption.

(Never mind ...)

The examples above from Billy Axeman are impressive (even my aging eyes locked in on them!)

 

[RLangham]

I was tempted to do this. Is there a simple way to calculate correct focal length, aperture and L-R distance?

I only have rules of thumb, but: Use a normal or slightly wide lens (45-50 mm on 35mm full frame), stop down some for depth of field, make sure both shots are the same settings and light doesn't change between, and the side movement distance should be close to human interocular distance -- reasonable to use any value between about 60 and 70 mm. You can and probably should use a wider separation for things like stereo landscapes (which will look flat at eye separation unless you have prominent foreground objects within about fifty feet of the tripod).

So is there an equation where I could calculate an effective separation for a longer lens?

 

[RLangham]

Reading this thread I had a (dangerous) thought -- for closeup work with TLRs there was a device sometimes called a "paramender" which was a little rack-and-pinion fitting placed under the camera so after composing with the viewing lens, you just turn a knob and it brings the taking lens up to where the viewing lens was to avoid parallax problems. Never had one in my hand -- wonder if they would work horizontally with an L-bracket? It could be quicker than the average macro slide contraption.

(Never mind ...)

The examples above from Billy Axeman are impressive (even my aging eyes locked in on them!)

I wonder however whether the separation would be enough. I think the distance between lens axes on a rolleiflex, for example, is less than two and a half inches.

 

[Donald Qualls]

So is there an equation where I could calculate an effective separation for a longer lens?

You could probably make a good case for just multiplying the spacing by the lens magnification factor -- so if you're using a 100 mm (on 35mm full frame), take your images five inches apart instead of two and a half. I haven't actually tried this with other than normal lenses, though, so I'd suggest shooting some tests or, if you've got a subject you can't easily return to, bracket your spacing. Shot one, step over a bit, shoot another, step over some more and shoot a third, then step over yet more for a fourth. In that set you ought to find a pair that feels comfortable to view and looks right.

 

[RLangham]

You could probably make a good case for just multiplying the spacing by the lens magnification factor -- so if you're using a 100 mm (on 35mm full frame), take your images five inches apart instead of two and a half. I haven't actually tried this with other than normal lenses, though, so I'd suggest shooting some tests or, if you've got a subject you can't easily return to, bracket your spacing. Shot one, step over a bit, shoot another, step over some more and shoot a third, then step over yet more for a fourth. In that set you ought to find a pair that feels comfortable to view and looks right.

By that logic I reduce for wides, huh? Counterintuitive in a way. I thought somehow that it would he the other way around.

 

[MattKing]

I wonder however whether the separation would be enough. I think the distance between lens axes on a rolleiflex, for example, is less than two and a half inches.

The paramender raises the camera the 50mm distance between the optical axes of the viewing and taking lenses - so it depends on how big your head is.

 

[DWThomas]

I wonder however whether the separation would be enough. I think the distance between lens axes on a rolleiflex, for example, is less than two and a half inches.

Yeah, a quick check of my Yashica 124g looks like about 1.75". But I fear the paramender, since it traveled vertically, might also rely on gravity and be a little tedious operating horizontally. It might not be too bad a job to make a sliding platform -- zip, click -- zap, click -- but whether it's worth the trouble, and what the other possible alternatives are, dunno. I acquired a low end macro slide sort of device for other reasons a couple years back that, on reflection, was perhaps a bit more low end than I intended.

 

[Donald Qualls]

By that logic I reduce for wides, huh? Counterintuitive in a way. I thought somehow that it would he the other way around.

Yeah, thinking about it, it probably has more to do with distance to the primary subject (and how much depth effect you want) than with the lens used. Whether you're using a 28mm or 135mm, if your main subject is fifty feet away (where things normally start to look flat with the naked eye), you'll want more separation than if it's five feet. Probably not a linear relationship, either.

 

[RLangham]

The paramender raises the camera the 50mm distance between the optical axes of the viewing and taking lenses - so it depends on how big your head is.

And on the average viewer's eye separation, which is about 62mm. One wants to get depth perception that everyone can appreciate comfortably.

 

[RLangham]

Yeah, thinking about it, it probably has more to do with distance to the primary subject (and how much depth effect you want) than with the lens used. Whether you're using a 28mm or 135mm, if your main subject is fifty feet away (where things normally start to look flat with the naked eye), you'll want more separation than if it's five feet. Probably not a linear relationship either.

It sure isn't a problem we're used to. That non-linear variable we're neglecting is determined by the fact that the optical paths of the human eyes converge at the point of focus in a person with healthy vision (unless they're consciously making them converge closer or farther to view stereo images, of course). A really perfect stereo camera would have to pivot both of the optical systems, including their film planes to match the focus distance. This is of course impractical. However, on a system with a moving mono camera, it could be emulated either by adjusting the separation or by actually converging the light paths of both shots... which would be painstaking but easier than designing a converging stereo camera!

The way I'd go about it would be to have two separate small 35mm cameras mounted in portrait orientation on some kind of tripod-mounted jig that one could set the distance on and have it adjusted to converge correctly--probably adjustable for focal length and or hyperstereo.

Barring that rather specialized piece of non-existent hardware, when I want stereo I'm going to use one camera, keep the light paths parallel and determine the separation by guessing, or by keeping it fixed at roughly 62mm/2.5".

 

[gdavis]

Yes, the Sputnik is one of the ones I was thinking of when I said most dedicate stereo cameras were a little sub-par... But then again, I've neither dealt with Sputnik or the Lubitel it derives from...

I had pictured them having really bad lenses, which is weird because I've only had one Russian lens that wasn't very satisfactory. They're all Industars on your Sputnik, right? Which is to say, a Soviet adaptation of the Tessar design... you could probably get some fantastic bokeh from a camera like that, with the shallow depth and out-of-focus performance of those lenses. I wonder what Bokeh does in stereo... I like Industars, at any rate, though I've only had the 35mm lenses and a dedicated enlarging lens that I used as a barrel lens on my Speed Graphic. They're good lenses based on a good design heritage, one from back when speed was generally unattainable so they focused on image quality. Industar-55 pancake lens for z39 or m42 is well worth it--you've seen the samples from it in this thread.

Do you find light leaks on your Sputnik? Looking at it it's one thing I would worry about. And do you find it flares badly?

I've read the lenses are Industars, but I can't see anything on the camera itself that would confirm (labeled as T-22 F-7.5 cm f:4.5). I can see why you might expect they would be cheap plastic junk but they do seem quite nice. As for bokeh, based on what I've read deep depth of field is preferable for stereo photography, but I haven't experimented with it enough to have formed my own opinion. I would guess that areas being out of focus may affect the perception of depth in those areas since there aren't any distinct features for your eyes to converge on. I do have some background blur in my test shots though and it looks OK.

First thing I did after receiving it was flock in the inside and add some light sealing foam around the edges. There's a little chip in the bottom edge that I think is still letting in light, the edge of the film is obviously fogged. So I still need to work on that but the image away from the edge came out nice and clear.

Closer you are to the subject, the narrower you want the lens separation (for a normal depth). You can increase or reduce from that to create a surreal depth effect. If I remember what I've read correctly, wider spacing makes things look smaller, like you're a giant.

The problem with two cameras angled to converge is that you get parallax distortion. A little bit can be tolerated but too much and the brain can no longer fuse the images. I wonder what shift lenses would do. That would allow you to converge the view without introducing the distortion.

 

[RLangham]

I've read the lenses are Industars, but I can't see anything on the camera itself that would confirm (labeled as T-22 F-7.5 cm f:4.5). I can see why you might expect they would be cheap plastic junk but they do seem quite nice. As for bokeh, based on what I've read deep depth of field is preferable for stereo photography, but I haven't experimented with it enough to have formed my own opinion. I would guess that areas being out of focus may affect the perception of depth in those areas since there aren't any distinct features for your eyes to converge on. I do have some background blur in my test shots though and it looks OK.

First thing I did after receiving it was flock in the inside and add some light sealing foam around the edges. There's a little chip in the bottom edge that I think is still letting in light, the edge of the film is obviously fogged. So I still need to work on that but the image away from the edge came out nice and clear.

Closer you are to the subject, the narrower you want the lens separation (for a normal depth). You can increase or reduce from that to create a surreal depth effect. If I remember what I've read correctly, wider spacing makes things look smaller, like you're a giant.

The problem with two cameras angled to converge is that you get parallax distortion. A little bit can be tolerated but too much and the brain can no longer fuse the images. I wonder what shift lenses would do. That would allow you to converge the view without introducing the distortion.

I mean very slightly converged, no more than the human eyes, though.

 

[iandvaag]

What a great thread! Nice to see some stereo discussion here.

I do medium format 3d (MF3D) as my primary photographic medium. I shoot with a Sputnik, a TL-120 (Chinese camera from 2006) and occasionally a Hasselblad for macro shots. I'm working on fixing up some Agfa Isolettes for doing hyperstereos.

The technique you are describing is often called "cha-cha" by its practitioners. It's a a great place to start since only one camera is needed. It can however sometimes make mounting (aligning) the images more difficult, since there are many degrees of freedom that the camera can move between shots apart from the desired lateral translation.There can also be "retinal rivalry" effects, where the scene changes between exposures (commonly seen in leaves of trees due to wind, for example). But it can certainly produce some great results.

the Sputnik is one of the ones I was thinking of when I said most dedicate stereo cameras were a little sub-par... But then again, I've neither dealt with Sputnik or the Lubitel it derives from...

Sure, the Sputnik is not the best built camera, nor does it have the sharpest lenses, but it can produce some truly outstanding results that will knock your socks off. And there are some advantages: it is lightweight and inexpensive. Use a tripod, compose carefully, meter accurately and view the slides in a backlit handheld viewer... you will feel like you are standing back in the very location where the image was made!

They're all Industars on your Sputnik, right? Which is to say, a Soviet adaptation of the Tessar design...

The Sputnik has 3-element anastigmat lenses, not a 4-element Tessar design. They are not critically sharp towards the edges, but they perform well enough at f/16 - f/22 which is where you're likely to do most of your shooting in stereo. I have seen slides shot with a Sputnik that cannot be discerned in a handheld viewer from shots taken with a Hasselblad. A well-tuned Sputnik is a very capable camera.

I wonder what Bokeh does in stereo...

As gdavis says, generally, most people prefer to have everything sharp in stereo images. The great benefit of a stereo image is the viewing experience -- your eyes can dart back and forth among the image planes and investigate all sorts of interesting details. Including out of focus areas is usually less interesting, because you want to be able to resolve the detail, not obscure it. Not to say that you can't break the rule, but most people prefer to avoid out of focus areas.

Do you find light leaks on your Sputnik?

I would say that most Sputniks have at least minor light leaks (probably right when they came out of the factory), but it's easy enough to put some felt in the light trap around the doors which usually solves the problem. While you're at it, I would recommend "flocking" the inside chambers of the camera (coating the shiny, reflective bakelite with black felt). This will improve contrast and reduce flaring artefacts.

So is there an equation where I could calculate an effective separation for a longer lens?

My friend Mike Davis has an excellent spreadsheet calculator on his website. Download the "Stereo Base in Millimeters v6.0" file.
The fields in yellow are what you need to enter. It assumes you are viewing the images in a viewer, not free-viewing.

Generally, a good rule of thumb is that the maximum acceptable on-film deviation (MAOFD) is equal to the viewer focal length divided by 30. This is just an empirical constant, and different people appreciate different amounts of depth. But you can certainly have too much! In the spreadsheet the "%MAOFD" box allows you to adjust the amount of depth to your preference. Leaving that value at 100 is a good place to start.

The on-film deviation (OFD) is what is recorded onto the film, and along with the determines the amount of depth in the image. The on-film deviation can be measured by measuring a stereo pair with a ruler:
1) measure the distance between two homologous points on an object that was furthest from the camera
2) measure the distance between two homologous points on the object that was closest to the camera
3) take the difference between 1) and 2)

Mike's spreadsheet calculates the on-film deviation in advance. To use the spreadsheet, fill in the values in yellow, and print it off. When you are out shooting, measure (tape measure, laser rangefinder, or just estimate) the distance to the nearest object in the frame as well as the farthest object. If you include infinity in your scene, as one often does, it becomes simpler. All you need is to measure the near-distance, find the appropriate row, and use the value in the last column. The values in the table are the camera separation (aka stereo base) in millimeters that will produce the on-film deviation that is specified in the top rightmost box..

A really perfect stereo camera would have to pivot both of the optical systems, including their film planes to match the focus distance.

I think many stereo buffs would disagree with you on this. Think of it this way: you want to present to each eye an image of the world as it really is. When looking at a stereo image in the viewer, the eye will still pivot and converge on homologous points. If you "toe-in" the cameras while shooting, you will get twice the convergence. One exception to this is macro images where this may be advantageous due to the very short stereo base.

Anyway, I hope that wasn't too much information. All the best on your stereo journey! One final recommendation: if you get the stereo bug, join a stereo club or participate in a traveling folio. I manage a medium format 3D folio featuring the work of about a dozen photographers. We each put in 4 slides and ship the box in a loop. When the box comes back around, you simply remove your 4 slides and submit 4 new ones. It's delightful to see the captivating images of others and fantastic to receive feedback on your own images. I'd highly recommend it!

Finally, if gdavis, RLangham or anyone else wants some stereo slide mounts for medium format, let me know. I cut my own cardboard mounts and sell them at cost. If you've just picked up a Sputnik or are trying cha-chas with a "mono" medium format camera, send me a PM with your address and I would be happy to mail you a few stereo mounts for free.

Ian Andvaag
Saskatoon, SK

 

[Donald Qualls]

You know, I've got a matching pair of Brownie Hawkeye cameras -- 6x6 on 620 (at least one will accept a trimmed 120 supply spool, so no respooling, and they're identical in every way I can detect, same sub-model even). They don't have a tripod mount, but the nearly cubical form factor begs for a foam-lined wood or metal bracket. The one on the left would be annoying to wind the film, and as built they have no provision for cable release (nor any exposure adjustments, though they're hyperfocal at about f/16). Tempting to jerry-rig a mount and try a pair of rolls and see if I like what I get.

 

[RalphLambrecht]

Pentax K5-II, Pentax-A 50mm f/2.8 Macro
Horizontal shift 5 cm.

https://wrb.home.xs4all.nl/Photos_2019_2/Photos_CrossEyedStereo_01/Stereo_01.htm
https://wrb.home.xs4all.nl/Photos_2019_2/Photos_CrossEyedStereo_02/Stereo_02.htm

Works for me we can clearly see the stereo effect.

 

[MattKing]

The one on the left would be annoying to wind the film

One right-side up, the other upside down?

 

[RLangham]

You know, I've got a matching pair of Brownie Hawkeye cameras -- 6x6 on 620 (at least one will accept a trimmed 120 supply spool, so no respooling, and they're identical in every way I can detect, same sub-model even). They don't have a tripod mount, but the nearly cubical form factor begs for a foam-lined wood or metal bracket. The one on the left would be annoying to wind the film, and as built they have no provision for cable release (nor any exposure adjustments, though they're hyperfocal at about f/16). Tempting to jerry-rig a mount and try a pair of rolls and see if I like what I get.

Yeah, that would work. Honestly two roughly identical SLR's mounted base to base on a special bracket, with a double cable release is one of the classic stereo systems, but the problem there is that unless they were compact SLR's (Pentax M series, Olympus OM, Zenit 1 or C) they would be a little hyperstereo.

 

[Donald Qualls]

One right-side up, the other upside down?

Yeah, or one on its right side and the other on its left. Problem with side mounting is, like most square format box cameras, there's only the one viewfinder on each camera. And with one upside down, it'd be a PITA to release both shutter simultaneously.

These are not impossible to service; I might open one up and see if how difficult it would be to add a cable release socket. They also have a B mode, so that'd be a sensible addition anyway.

 

[RLangham]

Yeah, or one on its right side and the other on its left. Problem with side mounting is, like most square format box cameras, there's only the one viewfinder on each camera. And with one upside down, it'd be a PITA to release both shutter simultaneously.

These are not impossible to service; I might open one up and see if how difficult it would be to add a cable release socket. They also have a B mode, so that'd be a sensible addition anyway.

People do add cable release sockets and even hot shoes to these. Me, I've never used one in any capacity where I'd need either, and I wouldn't want to modify mine. I have the dynamo flash where you spin a knob and it'll fire either an M bulb or a #5, and I take so few flash photos in general and so very few with the hawkeye and the bullseye that I've only gone through about one pack of Press 25's in the last four years.

As for bulb exposure I'm not sure that I've ever done one with a hawkeye. I did with my old Delta Stereo. It turned out alright.

 

[Donald Qualls]

One of my favorite shots ever with a Hawkeye was done with B -- shooting up the trunk of the "Tree" exhibit in the foyer of what's now the Western Forestry and Conservation Association (now inside the grounds of what's now the Oregon Zoo, then across the parking lot from Oregon Museum of Science and Industry), a 1 second exposure steadied against a handrail. But no, without a tripod socket, the B mode is of somewhat limited utility.

It might be possible to add a cable socket and surely is possible to add a tripod mount without any permanent mods to the camera, more or less the same idea as the "clamp" tripod and cable adapter for Minolta 16 and 16II cameras.

 

[RLangham]

One of my favorite shots ever with a Hawkeye was done with B -- shooting up the trunk of the "Tree" exhibit in the foyer of what's now the Western Forestry and Conservation Association (now inside the grounds of what's now the Oregon Zoo, then across the parking lot from Oregon Museum of Science and Industry), a 1 second exposure steadied against a handrail. But no, without a tripod socket, the B mode is of somewhat limited utility.

It might be possible to add a cable socket and surely is possible to add a tripod mount without any permanent mods to the camera, more or less the same idea as the "clamp" tripod and cable adapter for Minolta 16 and 16II cameras.

Hmm... yes, I can picture a little rig that would do both, in fact. Sort of two L-shaped pieces of wood, with felt pads on the inside of the vertical pieces and the top of the horizontal pieces. They'd screw together to form a U-shape, into which the camera would fit very snugly. One of them would have the tripod bushing and one would have a little piece of metal that wrapped over the shutter release with threading for a standard cable release.

I bet I could make it.

 

[barzune]

This is a fascinating thread, and I'm getting all kinds of ideas about how it works.
The pictures on #14 snapped so clearly for me that I intend to try the two-shot technique for myself.
One idea that I didn't see expressed in the discussions, though, is that the subject angle should maybe shift, along with the pupilary shift.
I'm thinking that, rather than a direct horizontal shift to provide focal diversion, there should also be an angular shift, centred on the point of interest.
The focal plane would, therefore, divert by a small segment of arc, rather than by a mere horizontal shift, which would naturally incur a change of central dimension.
I'll try the two techniques, and try to see whether there's a useful difference.

 

[Donald Qualls]

This is a fascinating thread, and I'm getting all kinds of ideas about how it works.
The pictures on #14 snapped so clearly for me that I intend to try the two-shot technique for myself.
One idea that I didn't see expressed in the discussions, though, is that the subject angle should maybe shift, along with the pupilary shift.
I'm thinking that, rather than a direct horizontal shift to provide focal diversion, there should also be an angular shift, centred on the point of interest.
The focal plane would, therefore, divert by a small segment of arc, rather than by a mere horizontal shift, which would naturally incur a change of central dimension.
I'll try the two techniques, and try to see whether there's a useful difference.

That was already discussed above with the suggestion that it hasn't been found necessary in the past. Millions of Viewmaster disks and viewers (and consumer cameras, never mind other stereo viewers and cameras) were based on parallel lens axes/film planes.

On the other hand, I don't know that anyone has actually done it -- perhaps you can show your results?

 

[barzune]

I'll have to try that out, but I'll experiment with digital first and see whether it makes sense in practice.
In theory, the difference between theory and practice is small. In practice, the difference between theory and practice is large.