Explaining Yaw free, distortionless & Yaw free/distortionless operation of a camera..

[Theodoros]

Most people think that yaw free operation of a camera is unimportant... They think that if you manage to work around things and have planes focused well enough it's the same...

Others think that if Scheimpflug's law is satisfied, (three planes crossing a line or near that) the operation is distortionless too...

The truth is different... Scheimpflug's law does state that three planes (the image area's plane, the subject's plane and the lens plane) should cross in a line... but it also assumes (which is never stated) that the projection radius of the cone projected in the image area should remain constant as it was if the subject was focused without any movements applied and thus, only a part of it in focus...

In other words it means that your camera is both yaw free and distortionless only if you can focus it without movements, then apply the movements you want (shifts & up downs included) and never have to refocus it... Otherwise (if you have to refocus it) there will be distortions (on all directions) involved in the image captured...

The more the focus will need corrections, the more the (distortion) error... The closer the subject's plane, the more evident the (distortion) error...

https://en.wikipedia.org/wiki/Scheimpflug_principle

[Nodda Duma]

Yes - I think - but there's a simpler way to think of it.

Distortion stems from the lens and is in part a function of field angle -- or at the image surface, the distance from the optical axis. Distortion can be thought of as a change in magnification vs. field angle -- because that's exactly what it is. The other aspect of a lens-based imaging system that affects magnification is lens-to-image plane distance and corresponding lens-to-object plane distance.

Introduce tilts in the lens or back and two things happen: 1) You've tilted the optical axis away from the mechanical center of the camera, shifting farther up the field-dependent distortion (field-dependent magnification) curve. 2) You've also introduced a linear magnification change due to the varying lens-to-image plane distances. Image plane, of course, is shorthand for plane of best focus, which is perpendicular to the optical axis.

The distortion/varying magnification has thus correspondingly increased and become asymmetrical.

[Doremus Scudder]

Yaw-free just keeps you from having to reposition the camera after making a swing when the camera has already had tilts applied. Sure, it takes an iteration or two, but it's pretty easy-peasy...

As for "distortion": We're dealing with a conical projection here. Any projection introduces some distortion; a lens introduces distortion at the edge compared to the center of the image. This is most evident with short focal-length lenses (oval rendering of round objects at the edges), but it's always there; it's a property of the conical projection. And, just like mapmakers, with their many projections (Mercator, cylindrical, Eckert, Mollweide, etc.), some distortion is always present in a projection.

The only important issue for us LF photographers is how we wish to render the subject on the film, which boils down to which conic section we choose for the film plane. The one and only thing that determines this is the position of the camera back in relation to lens plane. However, how planes are projected on the film is determined by the back position relative to the subject plane. This latter and choosing the plane of sharp focus are really the driving concerns that determine camera movements.

In other words, "distortionless" here is pretty meaningless, since there is always some. Which is the distorted image: one with converging verticals taken with no movements and the camera pointing upward at the top of a building or the one with parallel verticals taken after front rise has been applied? Most would say the former, but in reality, as long as the lens plane and the film plane are parallel, the projection is the same and there is no "distortion" in either by your definition. The image circle is still a circle and the radius is the same for both.

However, any time the lens plane and the film plane are not parallel, the projection at the film plane becomes a elliptical conic section, instead of a circle, with many different radii. Nevertheless, swings and tilts allows us to focus on other planes than the one parallel to the film plane. Any time you tilt to get the foreground sharp or swing to bring the plane of focus closer to one side or the other you are "distorting" the image by your radius-based definition. But, if we purposely alter the projection from circular to elliptical to get the image closer to our perception of reality, or to emphasize a particular plane of focus, or even to render objects larger at one place in the image than they would be with a "zero-position" shot, is that really "distortion," or are we simply choosing to use a different projection? I think its just the latter. The beauty of a view camera is that the photographer decides how the projection should be.

And, yaw-free really has nothing to do with which conic section the film plane ends up in. It's simply the relationship of the lens plane position to that of the film plane.

Doremus

[Nodda Duma]

Doremus - I think distortion in your example of non-parallel lens and film is really a magnification change across the image.

Distortion isn't a conical projection effect. It's a real aberration - a change in magnification vs field angle and is entirely a function of the lens design itself. Symmetrical lenses - amongst other designs - have zero distortion. Pinholes are free of distortion.

If distortion were a conical projection effect and couldn't be eliminated, then astrometry would not be possible.

We're all probably in agreement about the same effects, it's just a difference in semantics.

[Theodoros]

Hi Doremus... Nodda has it right I'm affraid... If you intercept a sphere at the same radius but with the center of it decentered to another point, the result plane (after intersecting the same sphere) will be an "egg" shape one instead of a circle, for a circle to exist (EDIT: ...in the projection cone...), there is no other way but to have an interception on the same axis being perpedicular to the base of the image cone (which then requires of the axis to pass through the center of the sphere)... Note that the center of the sphere projected towards the image area, is always the entrance pupil of the lens...

[Tin Can]

Too bad you guys don't use drawings.

Your words obfuscate.

I wonder about visualization on a photographic site.

Often.

[Theodoros]

The wikipedia link I used in the OP has all the drawings you need Randy... Only parameter that has to be considered is that they still ignore the change of the focusing plane's position (although they show it in blue) and thus the distortion that will be induced on the image area's plane (after a cone has been projected on it from a wrong position) - which is easy to visualize yourself (Imagine a sphere being sectioned from a center being a random point off center...) ...it is an "egg-shape" the "cut", ...isn't it?

[Pere Casals]

"Austrian army Captain Theodor Scheimpflug, who used it in devising a systematic method and apparatus for correcting perspective distortion in aerial photographs"

So... suprisingly Captain Theodor invented it to un-distort things.

As the LF learner that I am, I viewed a lot of times this image:

https://es.pinterest.com/pin/150448443772966725/

I was thinking that Tilt and Swing were ajusted to compensate vertical vs horizontal stretch, now I think that there is diagonal stretch, as image circle will be also an ellipse, but with a diagonal major axis...

Is it like this?


Anyway I guess that any distortion can be compensated with the enlarger by Tilt-Swing the projection lens, or the paper. Also it can be done with the scanned image, with PS. Does not?

[Theodoros]

"Austrian army Captain Theodor Scheimpflug, who used it in devising a systematic method and apparatus for correcting perspective distortion in aerial photographs"

So... suprisingly Captain Theodor invented it to un-distort things.

As the LF learner that I am, I viewed a lot of times this image:

https://es.pinterest.com/pin/150448443772966725/

I was thinking that Tilt and Swing were ajusted to compensate vertical vs horizontal stretch, now I think that there is diagonal stretch, as image circle will be also an ellipse, but with a diagonal major axis...

Is it like this?



Exactly... The theory only includes tilt and swing movements... it doesn't "compensate" for shifts or up-down movements or "parallel planes" as makers do... According to Scheimpflug's law, yaw free includes distortionless operation (other than using movement to have a third plane all in focus).... In other words the law doesn't allow for changing the focus point and thus induce distortions... It assumes that focusing stays constant (and thus operation is distortionless)!




Anyway I guess that any distortion can be compensated with the enlarger by Tilt-Swing the projection lens, or the paper. Also it can be done with the scanned image, with PS. Does not?

[/QUOTE]

No way! Distortions induced in an image captured as stereo (three axis perpendicular to each other), there is no way that can be corrected on a (two axis) plane...

[Nodda Duma]

Too bad you guys don't use drawings.

Your words obfuscate.

I wonder about visualization on a photographic site.

Often.

We're not crowded around a white board, and words have been used for centuries to relay technical information. The mind's eye is a wonderful illustrator in and of itself.

[Bruce Watson]

All I know is, if I level and plumb the film plane I can then use front tilt, swing, and shifts to make a more-or-less artistic image. But if I do this with some subjects, like buildings, the resulting print looks somehow unnatural. All those lines perfectly straight and parallel. In those cases, I use a little back tilt to create some keystoning, to make the image look "natural". Which it's not. But that's photography for ya.

I'm just sayin' that you can argue these things to death, but that the ground glass doesn't lie. What you see is what's important. I leave the math to the people designing cameras and lenses, and of course to the people posting in this thread. ;)

[ic-racer]

If one's camera Yaws or not when making certain movements, one still needs to set up the correct focal plane. A camera that shows Yaw when doing certain movements only requires secondary movements to obtain the correct focal plane. My daughter can do it.

[Peter Lewin]

Perhaps I am missing the point, unless it is that all of us using field cameras should go out and buy yaw-free Sinar P2s or another of the relatively small number of yaw-free cameras out there. To put it differently, is there something those of us with cameras that are not yaw-free should be doing?

[Theodoros]

Perhaps I am missing the point, unless it is that all of us using field cameras should go out and buy yaw-free Sinar P2s or another of the relatively small number of yaw-free cameras out there. To put it differently, is there something those of us with cameras that are not yaw-free should be doing?

Of course there is... (that's why this discussion is for)...

1. Make sure that your lens entrance pupil is positioned exactly at the center of the circle of which the tilt arc is part off... This will ensure that when tilting or swinging, the cone projected on the image plane will be one that its "top" is always the center of the same sphere and thus you'll have a circle projected... not an "egg shape"...
2. (after you make sure 1 above is satisfied) use your side shift's mechanism, or your up-down one, only if it is positioned bellow the respective swing & tilt ones. This means that if you need to side shift, don't do it with your front standard unless the side shift is bellow the swing mechanism and don't up-down shift (do it with your rear standard) unless your front standard's respective mechanism is below the tilt one...
This is because if you tilt or swing and your shift mechanisms are above the tilt/swing ones, the lens entrance pupil will not be centered anymore to the circle of which the tilt arc is part of the radius anymore and your swing axis won't be passing anymore from the axis around which the lens rotates... (in return the lens entrance pupil will move like if it is on a perimeter around it - think of a door that opens around an axis)...
3. Make sure that your image area's plane on the rear standard, is positioned on the same plane that intersects the tilt arc exactly at the middle of it (the "0" position), if it isn't, only use side and up down shifts on the rear standard and never tilts or swings... Other wise (in combination with 1&2 before) your image plane won't "receive" your lens cone projection as being a circle anymore, but at an angle...

Ideal situation is if you do position your lens entrance pupil in the center of the circle of which the tilt arc is part of the radius (and which the swing axis penetrates anyway), then do the same for the center of your light sensitive area and you have chosen a camera that has all its side and up down shift below the respective swing and tilt mechanisms on both standards... You can then do whatever you like with both standards and enjoy another level with your photography...

[Taija71A]

This Thread is definitely beginning to remind me of a 'Comedic Play'.
--
However, I do not recognize any of the 'Main' Characters from Shakespeare.
Where are Don Pedro, Dogberry, Leonato, Borachio or Benedick?

Also and 'More intriguingly' within these conversations...
We still are not laughing -- At any of the Jokes.

Why is that?
Perhaps, it really is... "Much Ado About Nothing". :D
--
The First Law of the View Camera (The Scheimpflug Principle)... Is now 'OLD NEWS'.
How about we talk about the Second Law (The Hinge Rule) of the View Camera -- Instead?

View Camera Focus and Depth of Field -- Part I.
http://www.trenholm.org/hmmerk/VCFaDOF1.pdf
_______

[Pere Casals]

I do not understand that:




No way! Distortions induced in an image captured as stereo (three axis perpendicular to each other), there is no way that can be corrected on a (two axis) plane...





My reasoning is next:

> We have an initial circular "circle of image" (say distortion free) projected on a "large" negative. (no movements)

> For simplicity let's think that diafragm is very closed and recorded field is inside DOF.

> Also let's think that negative is way larger than image circle.


If we do any tilt-swing with the camera standards we'll "linearly" distort this projected circle into an ellipse, that will be recorded in the negative. Any tilt-swing will always result in a projected ellipse. True... Does not?


> Then always we can find and tilt-swing inclination of the paper under te enlarger than will restore a circle on the paper, from the ellipse that was recorded in the negative... thus obtaining the same image (focus plane apart) that it would have been obtained without any tilt-swing done with the camera.


This is the reasoning I was thinking it was correct... I'm not sure of it... In any case, what is it incorrect in this reasoning??

[aluncrockford]

I do not understand that:





My reasoning is next:

> We have an initial circular "circle of image" (say distortion free) projected on a "large" negative. (no movements)

> For simplicity let's think that diafragm is very closed and recorded field is inside DOF.

> Also let's think that negative is way larger than image circle.


If we do any tilt-swing with the camera standards we'll "linearly" distort this projected circle into an ellipse, that will be recorded in the negative. Any tilt-swing will always result in a projected ellipse. True... Does not?


> Then always we can find and tilt-swing inclination of the paper under te enlarger than will restore a circle on the paper, from the ellipse that was recorded in the negative... thus obtaining the same image (focus plane apart) that it would have been obtained without any tilt-swing done with the camera.


This is the reasoning I was thinking it was correct... I'm not sure of it... In any case, what is it incorrect in this reasoning??

[aluncrockford]

I am curious as to who, of all the posters on this subject has actually used a 10x8 camera using all the movements needed to produce an outcome that satisfies a client . As someone who has worked as a photographer using 10x8 for advertising clients for the last thirty years I have to say that everything I have read here is not unlike a learn how to swim instruction book, might I suggest that if you are interested in using large format cameras with or without yaw free, then the best thing to do is actually take some pictures, personally I use a P2 in the studio and Deardorff on location both options are perfect , swapping them around would be less so

[Pere Casals]

then the best thing to do is actually take some pictures,

Yes... it's very, very important to practice... but knowing the theory (Scheimpflug, Hinge...) also helps

And the other thing that also helps is to see photographs from skilled people, and to understand how they creatively used movements, and the particularities of the lenses, the film, the process and the postprocess.

A lot to learn and to practice...

[Doremus Scudder]

Doremus - I think distortion in your example of non-parallel lens and film is really a magnification change across the image. ...

We're all probably in agreement about the same effects, it's just a difference in semantics.

I think we're just using terminology from different fields. With geometric projections as in mapmaking, a change of relative size of the projected image compared to the real subject is termed distortion. In a Mercator projection, for example, Greenland appears much bigger in relation to, say, Spain than it really is. The change in magnification due to the type of projection is termed distortion. Using swings or tilts on view camera results in a similar thing (the conic section being now an ellipse) and is therefore, "distorted." Similarly, a circular conic projection will have a different magnification at the edge of the film plane than at the center; it's a function of distance from the lens exit pupil. The distortion of objects at the edge of the image with very short focal-length lenses is a good example of this, but it exists in any lens.

To be clear, I'm not talking about lens aberrations here, just the inherent distortion in geometric projections. This distortion is always there in a conic projection and is greater the greater the difference in distance from the lens exit pupil. We have to keep in mind that we LF photographers are usually only using a part of the projected image, so the optical center of the image is often displaced from center on the film. This is simply cropping.

So, yaw-free or not, or whatever, it's really only the position of the lens plane and exit pupil relative to the film plane that determines any change in relative magnification of the image and, therefore, "distortion."

@aluncrockford: I don't shoot 8x10, but 4x5. Nevertheless, I use movements often and confidently and visualize my images in terms of planes of sharp focus and parallels. I think I've got a pretty good handle on the practical aspects of applying movements, using any of base, axis or asymmetric tilts/swings and with yaw and without. I'm trying to point out that there's really no such thing as distortion-free when using a camera of any kind.

Best,

Doremus

[Pere Casals]

it's a function of distance from the lens exit pupil.



I think that (in general) LF lenses have no visible distortion related to the distance to the exit pupil. This is something that any LF lens design solves very well or near perfectly. This is in case of no Tilt-Swing !!!!


A photographed wall of bricks is well square in all of the image circle that the lens manufacturer specifies. If the bricks are farther from the front lens element... yes... magification will be lower: the magnification of an object will depended of the distance to the object to the camera(and of the focal length), but magification will not depend of the "position of the object" in the image circle.


Distance to the exit pupil has a different effect: light falloff , that can be corrected by a center filter in the wides.

So, in my opinion (please correct me if I'm wrong...) magnification depends of the distance of the object to the entrance pupil, and it not depends of the distance from the exit pupil to the projection of the object on the ground glass, because that "distortion" is well corrected by the LF lens design...

PD: I've seen your work in your web... stunning photographs !!! love it

[B.S.Kumar]

Theodoros,

If you have finished modifying your Sinar, could you:

a. post some photos of the camera, so that we can better see what you are talking about
b. post some photos taken with a standard Sinar and your modified Sinar so that we can see the distortion.

Photos are still a thousand words each, despite Brexit...

Kumar

[Doremus Scudder]

I think that (in general) LF lenses have no visible distortion related to the distance to the exit pupil. This is something that any LF lens design solves very well or near perfectly. This is in case of no Tilt-Swing !!!!

A photographed wall of bricks is well square in all of the image circle that the lens manufacturer specifies. ...
Distance to the exit pupil has a different effect: light falloff , that can be corrected by a center filter in the wides.

So, in my opinion (please correct me if I'm wrong...) magnification depends on the distance of the object to the entrance pupil, and it does not depend on the distance from the exit pupil to the projection of the object on the ground glass, because that "distortion" is well corrected by the LF lens design...

PD: I've seen your work in your web... stunning photographs !!! love it

Pere,

You are correct to a certain extent: most lenses are designed to be rectilinear, meaning that they preserve parallel lines and right angles in a focus plane parallel to the film. So yes, your brick wall will have straight lines and square corners. The distortion I'm talking about, and which is apparent when you photograph your brick wall up-close with a short focal-length lens, is the elongating of the bricks at the edge of the image. This is caused by the difference in distance from the exit pupil and the same elongation happens to images made with longer focal lengths, just to an enough lesser extent as to be practically unnoticeable with long lenses. It is simply a difference in magnification from the center to the edge of the image and is a characteristic of the conic projection. So, no barrel or pincushion distortion, but some change in size from center to corner. This is what I'm calling distortion here.

Best,

Doremus

[rdenney]

I see several axes of discussion. One is geometric distortion of a lens, which tends to either barrel or pincushion, and how that is affected by projection onto a film plane that is not normal to the lens axis. I don't typically run into that with large-format lenses, and especially the approximately symmetrical designs of plasmats, biogon derivatives, and super angulon derivatives cancel out such distortions.

I really like the notion of thinking of rectilinear projection as a distortion, which is a second use of the word distortion. Anytime the three-dimensional world is represented in two dimensions, there will some distortion of reality. But really it's projection we are talking about. Rectilinear projection causes round objects to appear non-round if sufficiently off-axis. Only with rectilinear lenses much shorter than the film diameter does it become really noticeable, because the image contains material that's much farther off-axis with respect to the focal length. And with really short lenses, parts of a round subject of any significant size in the composition are subject to more of it than other parts, pulling it into all manner of unnatural shapes. In order to keep the lines straight, radial distances are stretched. That's what Doremus is talking about.

With smaller formats, I'm a fan of full-frame fisheye lenses for many subjects that don't have straight lines because the result looks more organic and natural. But the closer we get to the edge of a projected fisheye image, the more subjects are squeezed radially, which is the opposite of being stretched radially as with rectilinear projection. A fisheye lens with a much larger image circle than the image may not produce really apparent fisheye projection, while still maintaining a more natural representation of round objects. The 35mm focal length of the fisheye on my Pentax 67 is about one-third of the nominal film diameter, and that's about as short as I go. I have even adapted 30 and 35mm medium-format full-frame fisheye lenses for 35mm to get a nicer projection of non-straight subjects.

But the fisheye effect is intentionally called "distortion" while the rectilinear effect is intentionally called "distortionless". Certainly some straight lines need to remain straight, but even with a fisheye, straight lines that go through the center of the lens remain straight. With a view camera, it's possible for the image to not be centered on the center of the lens, of course, so that those straight lines radiate from somewhere other than the center of the image. I have come to think of the fisheye effect as "round" projection versus the rectilinear effect as "straight" projection, with each having its uses in different circumstances.

All my examples are from smaller formats, because there is no fisheye-effect lens for large format. I have at times secretly wished for it. A 65mm fisheye lens would be just huge, massive fun on a 4x5 camera.

Rick "yes, changing the subject a bit" Denney

[Pere Casals]

Pere,

brick wall up-close with a short focal-length lens, is the elongating of the bricks at the edge of the image.




Ok... but going again to the wall situation, bricks whose projection is smaller is not because they are projected in the outer part of the image circle, they are smaller because ther are much farther from the camera. Consider a cylindrical wall and that the camera is in the shaft, all bricks that will be at the same distance from the lens entrance will have the same projected size, and the magnification of all those bricks will be the same, well in fact the wall has to be spherical.

[Pere Casals]

plasmats, biogon derivatives, and super angulon derivatives cancel out such distortions.

I really like the notion of thinking of rectilinear projection as a distortion,



Yes... Fisheye lenses forgot rectilinear projection, it's an exception, producing an special mapping, usualy an "equisolid angle" projection.

https://en.wikipedia.org/wiki/Lambert_azimuthal_equal-area_projection



Well, our retina has an spherical projection surface !!! for this reason human (and animal) eyes work very, very, well with a quite optically simple lens, that spherical projection is the way we see things, later brain makes additional calculations...

https://en.wikipedia.org/wiki/Retina

Sony is preparing an spherical sensor that may simplify DSLR optics a lot, in this way they will be able to strongly compete with Canon and Nikon camera systems, because optics will be of less importance. Perhaps they will wait to not damage the market size, such an innovation can blow the whole glass industry. They will process the spherical to flat projection in the same way that they correct distortion, easy. Today's DSLR lenses design do not enforce distortionless to allow other optimizations and performance, as distortion is software well corrected.

http://petapixel.com/2014/07/04/first-photo-taken-sony-curved-sensor/

(The image is not good because lens used was for flat sensors.)


There are another interesting type of lenses, used in Machine Vision, called Telecentric, with magnification constancy

https://en.wikipedia.org/wiki/Telecentric_lens

that provides an orthographic projection (not conical...)

http://www.opto-engineering.com/resources/telecentric-lenses-tutorial



But our glasses like Symmar and Sironars render things distortion free, magnification depends upon distance to the camera...

For field curvature, swirls and other fun we can trust in projection lenses, that usualy are of Petzval type, or a fast Hektor. : )

https://www.flickr.com/photos/johanbiilmann/27082703610/in/dateposted/