Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > perspective w/ 35mm lenses?

Reply
Thread Tools

perspective w/ 35mm lenses?

 
 
Nostrobino
Guest
Posts: n/a
 
      08-04-2004

"Jeremy Nixon" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Nostrobino <(E-Mail Removed)> wrote:
>
> > No, the camera is in the same position.

>
> Look, when you move it, it's not in the same position. You've pointed
> it in a different direction! That requires "movement".


No, the camera is still in the same position relative to the subject.


>
> >> Move it, and, well, you have moved it. Don't move it, and you haven't
> >> moved it.

> >
> > I haven't moved it.

>
> Then how did it end up pointing in a different direction? The image plane
> has moved. If you move it, perspective changes. If you don't, it

doesn't.

This is a new version of the there's-no-such-thing-as-wide-angle-perspective
argument? You're now saying that it's IMAGE PLANE and its relationship to
the subject that matters, not camera position?

Is this the final version, or will there be more changes as we move along?


>
> >> If you want to demonstrate that focal length changes perspective you

need
> >> to give an example where you *don't* move the camera.

> >
> > I already have, earlier today. See my reply to BillyJoeJimBobZeke.

>
> You mean the one where you're photographing a totally different object?
> That one was too silly to even talk about, wasn't it? I mean, it's at a
> different angle to the camera,


<GUFFAW!>

THAT'S WHAT WIDE-ANGLE LENSES DO!

I have already pointed this out at least a couple of times, haven't I? Come
on now, how about a little intellectual honesty here? All of your stunts to
"prove there's no such thing as wide-angle perspective" have relied entirely
on making comparisons which carefully remove or ignore the very aspects of
the image that clearly DEMONSTRATE wide-angle perspective.


> so obviously the perspective is going to
> be different; the example doesn't even make sense, and doesn't apply to
> the subject of perspective changes at all.
>
> > You haven't read the reply. Read it. The camera is not moved a

millimeter
> > and is left pointing in exactly the same direction, not altered by a

degree,
> > a minute or even a second of arc.

>
> Sure, and in order to compensate for that, you're looking at a different
> object


The objects are identical; what's the difference? But if it really bothers
you, go ahead and swap them back and forth. It won't change the fact that
the one in the wide-angle shot still shows wide-angle perspective. And by
this time I think you know it as well as I do.


> in a different position!


As I have already said: go ahead and take a long-lens shot of the same
object any way you like, any way at all, position camera, object, film plane
however you want them--and you still will not be able to duplicate the
wide-angle perspective with a long lens.



> Okay, so since this obviously wasn't
> implied, I'll spell it out: you need to not move the camera *or* the
> subject. Okay? Because changing the relative position of those things
> changes the perspective.


Which is exactly what a wide-angle lens does, hence the wide-angle
perspective.

Have we covered this enough, or do you want to go over it all another 40 or
50 times?


> Zooming in, however, does not.


Because then you are only magnifying something that never did, and never
could, illustrate the wide-angle perspective in the first place. This, too,
we have gone over and over.

Time to start working on your retreat strategy, Jeremy.


 
Reply With Quote
 
 
 
 
Jeremy Nixon
Guest
Posts: n/a
 
      08-04-2004
Nostrobino <(E-Mail Removed)> wrote:

> You are still hung up on that "spherical image" misconception. There isn't
> any spherical image where rectilinear lenses are concerned.


Of course there is. If you take every point that is a certain distance away
from the lens, they will create a sphere. You are projecting that sphere onto
a flat surface. You get distortion when you do this.

> What is projected onto the flat plane is a two-dimensional representation of
> a three-dimensional world.


Imagine that you are photographing a building. In the lower left corner of
the image, part of the structure forms a right angle. You've used a wide-angle
lens to get your "wide-angle look". Is the right angle in the world still a
right angle in the picture? No, it's not. It's not whether you used a wide-
angle lens or not, but I had you using one just to factor out that part of
the discussion.

Why is a right angle in the corner of the image not a right angle in the
picture?

Because you've flattened the spherical image onto a flat plane, and the
resulting distortion is greatest in the corners of the image.

Why is there such a thing as a "wide-angle look"? Why do wide-angle lenses
show greater distortion at the edges than telephoto lenses?

Because they are capturing a greater area of the spherical image, so there
is more curvature in the original that is projected onto the flat plane,
and the relative angle between image plane and subject is greater at the
edges with the wider field of view, so there is more distortion at the edge
of a wide-angle image. This distortion is not absent in a telephoto image,
you just can't see it as obviously because you're using a smaller portion
of the sphere.

In the above example, if you use a fisheye lens, the right angle *will* be
a right angle in the picture. You've used a different projection, one that
maintains right angles. The perspective is the same, but the projection is
different, so the distortion is different -- straight lines will not remain
straight.

So, in one image, a right angle in the world is not a right angle in the
picture. In the other, a right angle is a right angle, but straight lines
are not straight. Which one is an "accurate" representation of a three-
dimensional world?

>> this results in distortion of some kind, depending on what type of
>> projection you use. When you use a projection that keeps straight lines
>> straight (your "normal rectilinear lens") then parallel lines will not
>> remain parallel.

>
> Sure they will, if the subject is essentially two-dimensional (i.e., any
> flat surface that's perpendicular to the lens axis). If the subject is NOT
> flat and perpendicular then of course parallel lines in it will converge.
> This is NOT DISTORTION. It is an accurate two-dimensional representation of
> what is in front of the lens.


The converging parallel lines are not distortion, but are not necessarily an
accurate representation of what is in front of the lens. Why do architectural
photographers use shift lenses, if converging parallel lines are accurate?

The distortion is a product of flattening the image. The converging parallel
lines are a product of the relative angle of the image plane to the subject.
The flattening distorts everything based in part on this angle, which is why
things at the corner of your wide-angle shots are "different" from things in
the center of them. It's why right angles can't stay as right angles unless
the object is perpendicular to the lens axis, as long as you're using a
"normal" lens.

It's *all* distorted. It's just a matter of which distortion you want. It
is not possible to create a flat image as a photograph without distorting
*something*, because the lens is looking at a sphere, not a plane.

If you focus your lens to 10 feet, what are you photographing? It's not a
flat plane 10 feet in front of the camera, parallel to the image plane. It
is the inside surface of a sphere 10 feet from the lens in all directions,
and you're photographing a portion of that surface by flattening it, which
distorts the image in much the same way that creating a map on a flat piece
of paper makes Greenland look as big as Africa. They use different types
of projection in map-making, but it's the same idea -- representing a
sphere on a flat surface.

>> A flat image from a camera is *always* "distorted" in some way in order to
>> make it flat.

>
> Nonsense. If the subject is flat and perpendicular, you can take
> measurements from the image and they will (within reasonable tolerances)
> correspond to measurements taken from the subject itself. There's
> substantially NO DISTORTION.


Sure there is, it is just canceled out by the (unnatural) flat subject. If
you want to photograph a two-dimensional surface in order to prove that there
is no distortion when photographing a three-dimensional world, I guess that's
about as valid as moving the camera to prove that you can change perspective
without moving the camera, though.

--
Jeremy | http://www.velocityreviews.com/forums/(E-Mail Removed)
 
Reply With Quote
 
 
 
 
DSphotog
Guest
Posts: n/a
 
      08-04-2004
YOO-HOO !!! I can.

"Nostrobino" <(E-Mail Removed)> wrote in message
news15Qc.41$1%(E-Mail Removed)...
>
> "BillyJoeJimBob" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
> > Nostrobino wrote:
> > >
> > > No good, BillyJoeJimBob. You have to keep the object in the corner
> > > of the frame.

> >
> > As I stated in an earlier post, that is impossible without either:
> >
> > A) changing the camera's position relative to the object (which you
> > have forbidden), or

>
> Correct. The camera position remains the same. This, remember, is what

your
> side insists is the SOLE determinant of perspective--the camera position,
> nothing else.
>
>
> > B) changing the direction the camera is pointing (which changes the
> > locations of the vanishing points in the frame).

>
> Obviously the camera DIRECTION has to be changed in order to keep the

object
> in the corner of the frame. Neither you nor others on your side of the
> argument have ever said camera direction mattered as far as perspective is
> concerned. You've all said it's only camera position, only camera

position,
> only camera position that matters. If you are now at last admitting that
> other factors come into it, your whole argument begins to collapse.
>
>
> >
> > If you do either of these two things, you've changed the perspective

>
> Ah-HA!
>
> Are you now half or two-thirds of the way to finally understanding what I
> have been saying? You will remember that I said several times that
> perspective is determined by three things: camera position, focal length,
> and the direction the camera is pointing. Now it is only focal length that
> you still have to be convinced about.
>
>
> > regardless of what focal length lens you're planning on using. Thus,
> > it cannot be stated that the focal length has any effect at all on
> > perspective based on the conditions you stipulated since it is not
> > possible to separate lens effects (if any, and there are not any)
> > from effects caused by changes in the camera's relative position or
> > pointing direction.

>
> The only difference in perspective that results from keeping the object in
> the corner of the frame as focal length is changed is BECAUSE of the focal
> length change.
>
> Objects in the corner of a wide-angle shot appear radially stretched

(under
> normal viewing circumstances). This is by definition a difference in
> perspective. It is what gives the wide-angle shot its "wide-angle look."
>
> All the earlier suggestions to "prove" there is no difference in

perspective
> between lenses of different focal length, by enlarging the center of a
> wide-angle shot to the same size as a long-lens shot, etc., are invalid

and
> meaningless. Doing that proves nothing except that perspective is not
> changed by enlargement. There never was any suggestion that it would be.

OF
> COURSE you cannot demonstrate wide-angle perspective without a wide angle!
> (Duh.)
>
>
>
> >
> > > All you're doing is increasing the field of view
> > > around the object, not affecting the object itself. Obviously that
> > > will not change the perspective. It's wide-angle perspective that
> > > I'm saying a wide-angle lens produces, remember? That's what the
> > > argument is all about. Limit the object
> > > to where the lens covers it at the long setting and you can't
> > > possibly see the wide-angle perspective.
> > >
> > > Do it again, but properly.

> >
> > And that would be, how?
> >
> > > Put the object in the corner of the frame
> > > at 18mm, then do whatever you like at 55mm. While that's still not
> > > much of a range, some difference should be apparent.

> >
> > Okay, "then do whatever you like at 55mm" is absolutely not any sort
> > of experimental procedure. What, precisely, should I do at 55mm that
> > does not involve moving the camera, the object or changing the
> > direction in which the camera is pointing and yet still keeps the
> > object in the corner of the frame?

>
> What I am saying is that you can do anything you like, and you CANNOT
> duplicate the perspective of a wide-angle lens with a long lens. Go ahead
> and try. I don't care what you do at 55mm, you will never get the
> perspective you did at 18mm. Like it or not, wide-angle perspective

exists,
> it's real, anyone with normal eye-brain functioning who has not been
> bamboozled by reading nonsense about perspective can see it, and (except

by
> later manipulation of the image of course) you cannot get that perspective
> with a long lens no matter what you do.
>
> Once again (and this, too, I have repeated several times), the perspective
> of any picture is a characteristic of that picture IN ITS ENTIRETY. You
> cannot take a little piece of a picture (which is what you're trying to

do)
> and say that conveys the perspective of the original whole.
>
>
> >
> > If you can tell me (in simple terms, please) what I should do to keep
> > an object in the corner of the image at both 18mm and 55mm without
> > moving the camera, the object, or changing the direction the camera
> > is pointing, then I'll do that.

>
> Obviously it's impossible to keep the object in the corner of the frame at
> different focal lengths without changing the direction of the camera.

Gosh.
> What to do, what to do?
>
> Okay, here's the solution: You used a CD in your experiment. Leave it in

the
> same place on the floor (where it will be in the corner of the frame at

the
> longest lens setting), leave the camera pointing in the same direction
> (angled downward of course), and place ANOTHER CD on the floor where it

will
> be in the corner of the frame with the lens at its widest angle. Small

boxes
> would be better, but CDs should do.
>
> Voilą!
>
> The CDs being identical in size and shape, the only difference in
> perspective between the two of them now will be the result of difference

of
> field of view, the first one illustrating long-lens perspective and the
> second one wide-angle perspective. Enlarge the farther one to suit if you
> feel it necessary, but I doubt it will be.
>
> Of course if this gets out, the pundits who write nonsense for
> "authoritative texts" will probably hate you and may snarl at you if they
> pass you in the street, but look at what you have accomplished!
>
>



 
Reply With Quote
 
DSphotog
Guest
Posts: n/a
 
      08-04-2004

"Nostrobino" <(E-Mail Removed)> wrote in message
news15Qc.41$1%(E-Mail Removed)...
>
> "BillyJoeJimBob" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
> > Nostrobino wrote:
> > >
> > > No good, BillyJoeJimBob. You have to keep the object in the corner
> > > of the frame.

> >
> > As I stated in an earlier post, that is impossible without either:
> >
> > A) changing the camera's position relative to the object (which you
> > have forbidden), or

>
> Correct. The camera position remains the same. This, remember, is what

your
> side insists is the SOLE determinant of perspective--the camera position,
> nothing else.
>
>
> > B) changing the direction the camera is pointing (which changes the
> > locations of the vanishing points in the frame).

>
> Obviously the camera DIRECTION has to be changed in order to keep the

object
> in the corner of the frame. Neither you nor others on your side of the
> argument have ever said camera direction mattered as far as perspective is
> concerned. You've all said it's only camera position, only camera

position,
> only camera position that matters. If you are now at last admitting that
> other factors come into it, your whole argument begins to collapse.
>
>
> >
> > If you do either of these two things, you've changed the perspective

>
> Ah-HA!
>
> Are you now half or two-thirds of the way to finally understanding what I
> have been saying? You will remember that I said several times that
> perspective is determined by three things: camera position, focal length,
> and the direction the camera is pointing. Now it is only focal length that
> you still have to be convinced about.
>
>
> > regardless of what focal length lens you're planning on using. Thus,
> > it cannot be stated that the focal length has any effect at all on
> > perspective based on the conditions you stipulated since it is not
> > possible to separate lens effects (if any, and there are not any)
> > from effects caused by changes in the camera's relative position or
> > pointing direction.

>
> The only difference in perspective that results from keeping the object in
> the corner of the frame as focal length is changed is BECAUSE of the focal
> length change.
>
> Objects in the corner of a wide-angle shot appear radially stretched

(under
> normal viewing circumstances). This is by definition a difference in
> perspective. It is what gives the wide-angle shot its "wide-angle look."
>
> All the earlier suggestions to "prove" there is no difference in

perspective
> between lenses of different focal length, by enlarging the center of a
> wide-angle shot to the same size as a long-lens shot, etc., are invalid

and
> meaningless. Doing that proves nothing except that perspective is not
> changed by enlargement. There never was any suggestion that it would be.

OF
> COURSE you cannot demonstrate wide-angle perspective without a wide angle!
> (Duh.)
>
>
>
> >
> > > All you're doing is increasing the field of view
> > > around the object, not affecting the object itself. Obviously that
> > > will not change the perspective. It's wide-angle perspective that
> > > I'm saying a wide-angle lens produces, remember? That's what the
> > > argument is all about. Limit the object
> > > to where the lens covers it at the long setting and you can't
> > > possibly see the wide-angle perspective.
> > >
> > > Do it again, but properly.

> >
> > And that would be, how?
> >
> > > Put the object in the corner of the frame
> > > at 18mm, then do whatever you like at 55mm. While that's still not
> > > much of a range, some difference should be apparent.

> >
> > Okay, "then do whatever you like at 55mm" is absolutely not any sort
> > of experimental procedure. What, precisely, should I do at 55mm that
> > does not involve moving the camera, the object or changing the
> > direction in which the camera is pointing and yet still keeps the
> > object in the corner of the frame?

>
> What I am saying is that you can do anything you like, and you CANNOT
> duplicate the perspective of a wide-angle lens with a long lens. Go ahead
> and try. I don't care what you do at 55mm, you will never get the
> perspective you did at 18mm. Like it or not, wide-angle perspective

exists,
> it's real, anyone with normal eye-brain functioning who has not been
> bamboozled by reading nonsense about perspective can see it, and (except

by
> later manipulation of the image of course) you cannot get that perspective
> with a long lens no matter what you do.
>
> Once again (and this, too, I have repeated several times), the perspective
> of any picture is a characteristic of that picture IN ITS ENTIRETY. You
> cannot take a little piece of a picture (which is what you're trying to

do)
> and say that conveys the perspective of the original whole.
>
>
> >
> > If you can tell me (in simple terms, please) what I should do to keep
> > an object in the corner of the image at both 18mm and 55mm without
> > moving the camera, the object, or changing the direction the camera
> > is pointing, then I'll do that.

>
> Obviously it's impossible to keep the object in the corner of the frame at
> different focal lengths without changing the direction of the camera.

Gosh.
> What to do, what to do?
>
> Okay, here's the solution: You used a CD in your experiment. Leave it in

the
> same place on the floor (where it will be in the corner of the frame at

the
> longest lens setting), leave the camera pointing in the same direction
> (angled downward of course), and place ANOTHER CD on the floor where it

will
> be in the corner of the frame with the lens at its widest angle. Small

boxes
> would be better, but CDs should do.
>
> Voilą!
>
> The CDs being identical in size and shape, the only difference in
> perspective between the two of them now will be the result of difference

of
> field of view, the first one illustrating long-lens perspective and the
> second one wide-angle perspective. Enlarge the farther one to suit if you
> feel it necessary, but I doubt it will be.
>
> Of course if this gets out, the pundits who write nonsense for
> "authoritative texts" will probably hate you and may snarl at you if they
> pass you in the street, but look at what you have accomplished!
>
>

One of the many things that you're missing her is that camera POSITION
involves THREE planes, not just the two that you have choosen to use. 1)
Forward and backward (closer or farther from subject) 2) left and right
horizontally and 3) up and down vertically.

FYI - The film plane is the effective camera position. All the other parts
of the camera simply support what one is trying to accomplish at the film
(or image if you wish) plane.

When you turn the camera and point it down you have, in fact, changed the
distance from the object in question of the left and right part of the plane
and the top and bottom as well. Thus, you have "moved" the camera.

Regarding the above "Yoo-Hoo", ever played with a view camera??

Regards,
Dave


 
Reply With Quote
 
DSphotog
Guest
Posts: n/a
 
      08-04-2004
"All the earlier suggestions to "prove" there is no difference in
perspective
between lenses of different focal length, by enlarging the center of a
wide-angle shot to the same size as a long-lens shot, etc., are invalid and
meaningless. Doing that proves nothing except that perspective is not
changed by enlargement. There never was any suggestion that it would be. OF
COURSE you cannot demonstrate wide-angle perspective without a wide angle!
(Duh.)"

Since you now agree that enlarging and image doesn't change perspective,
what exactly do you think happens when you replace a wide angle lens with a
telephoto?
ENLAGEMENT I believe. And without changing perspective. Wadda ya know.
Thanks for your help with this.

Regards Again,
Dave


 
Reply With Quote
 
Jeremy Nixon
Guest
Posts: n/a
 
      08-04-2004
Nostrobino <(E-Mail Removed)> wrote:

> No, the camera is still in the same position relative to the subject.


No, it's not. You've moved it.

> This is a new version of the there's-no-such-thing-as-wide-angle-perspective
> argument? You're now saying that it's IMAGE PLANE and its relationship to
> the subject that matters, not camera position?


The image plane is the camera.

> All of your stunts to "prove there's no such thing as wide-angle perspective"
> have relied entirely on making comparisons which carefully remove or ignore
> the very aspects of the image that clearly DEMONSTRATE wide-angle perspective.


All of your stunts to prove that you can change perspective without moving the
camera or the subject have moved either the camera or the subject, which
everyone agrees changes perspective.

>> Sure, and in order to compensate for that, you're looking at a different
>> object

>
> The objects are identical; what's the difference?


Their relative position to the camera is the difference. Since that is
different, the perspective will be different.

I thought you were only arguing that focal length changes perspective, but
now it appears that you're saying that relative position of camera and
subject does *not* change perspective? That no matter what angle you have
on an object, the perspective will be the same? You're not really making
much sense here.

>> Okay, so since this obviously wasn't implied, I'll spell it out: you need
>> to not move the camera *or* the subject. Okay? Because changing the
>> relative position of those things changes the perspective.

>
> Which is exactly what a wide-angle lens does, hence the wide-angle
> perspective.


You're really trying to say that using a wide-angle lens moves the subject
or the camera to a different position? How does it accomplish this amazing
feat?

--
Jeremy | (E-Mail Removed)
 
Reply With Quote
 
Bart van der Wolf
Guest
Posts: n/a
 
      08-04-2004

"Nostrobino" <(E-Mail Removed)> wrote in message
news:eN5Qc.39$(E-Mail Removed)...
>
> "Bart van der Wolf" <(E-Mail Removed)> wrote in message
> news:4110e715$0$34762$(E-Mail Removed)4all.nl...

SNIP
> > Projection on a flat surface means the corners are at a
> > larger distance than the center normal.

>
> You may have something in mind there, but you have concealed it

artfully.

I'll try and keep it simple; shine a flashlight beam straight onto a
surface from short distance. You'll see a circular spot. Now without
changing the distance to the surface, shine it at a slight angle
towards an imaginary rectangle's corner. You'll see a sort of an
ellipse, even egg shape if you look carefully, but the shape depends
on distance.

Due to the projection angle the circle is distorted because it hits a
flat plane and not a spherical surface. This is the projection
distortion caused by conversion from a spherical to a rectangular
projection.

I hesitate to take it a step further, because if you read the
following whith the wrong mind set, it'll boggle the mind or (perhaps
worse) you'll draw the wrong conclusions. Nevertheless here goes:
http://www.trenholm.org/hmmerk/SHBG07.pdf
Especially focus (pun intended!) on figure 2. and realise this was
shot relatively close-up.

Does it boggle the mind already? Told you it would.
There you have a demonstration of projection distortion increasing the
size of the discs/dots (more in one dimension) and perspective
reducing it at the same time. If you call everything that may
contribute to a certain "look" perspective, you'll not be able to
unravel what happened to the "dots", and certainly not able to control
it predictably.

Bart

 
Reply With Quote
 
Nostrobino
Guest
Posts: n/a
 
      08-04-2004

"Bart van der Wolf" <(E-Mail Removed)> wrote in message
news:41112964$0$34762$(E-Mail Removed)4all.nl...
>
> "Nostrobino" <(E-Mail Removed)> wrote in message
> news:eN5Qc.39$(E-Mail Removed)...
> >
> > "Bart van der Wolf" <(E-Mail Removed)> wrote in message
> > news:4110e715$0$34762$(E-Mail Removed)4all.nl...

> SNIP
> > > Projection on a flat surface means the corners are at a
> > > larger distance than the center normal.

> >
> > You may have something in mind there, but you have concealed it

> artfully.
>
> I'll try and keep it simple; shine a flashlight beam straight onto a
> surface from short distance. You'll see a circular spot. Now without
> changing the distance to the surface, shine it at a slight angle
> towards an imaginary rectangle's corner. You'll see a sort of an
> ellipse, even egg shape if you look carefully, but the shape depends
> on distance.


Well, if I am at the position of the flashlight (which would be assumed),
I'd still see a circle. Someone more perpendicular to the wall where the
beam hit it would see an ellipse, that's true. But the person in that
position would see an ellipse where I shone the flashlight beam
perpendicularly to the wall, also (the ellipse then going in the other
direction, i.e. vertically instead of horizontally). In fact there's only
one angle from which he would see a perfect circle, and that's the same
angle at which I shone the beam on the wall, but on the other side.


>
> Due to the projection angle the circle is distorted because it hits a
> flat plane and not a spherical surface. This is the projection
> distortion caused by conversion from a spherical to a rectangular
> projection.


Okay, now I see what you mean.


>
> I hesitate to take it a step further, because if you read the
> following whith the wrong mind set, it'll boggle the mind or (perhaps
> worse) you'll draw the wrong conclusions. Nevertheless here goes:
> http://www.trenholm.org/hmmerk/SHBG07.pdf
> Especially focus (pun intended!) on figure 2. and realise this was
> shot relatively close-up.


Figure 2 is in fact an excellent example of exactly what I'm talking about,
wide-angle perspective. Note that the ping-pong balls appear to be stretched
radially, while the two-dimensional spots are not changed in any way. It is
impossible to look at that photo and NOT see the wide-angle perspective.


>
> Does it boggle the mind already? Told you it would.


No, it's mostly an article about the Scheimpflug principle, and I'm
basically familiar with that.


> There you have a demonstration of projection distortion increasing the
> size of the discs/dots (more in one dimension) and perspective
> reducing it at the same time. If you call everything that may
> contribute to a certain "look" perspective, you'll not be able to
> unravel what happened to the "dots", and certainly not able to control
> it predictably.


But NOTHING happens to the discs or dots. Look again. That's precisely the
point the author is making: it's only the SOLID objects that are affected in
this way.

This is exactly the case with any wide-angle rectilinear lens. (It is not
the case with a fisheye lens, which is a whole different ball of wax.) Only
SOLID objects have perspective of any kind. Two-dimensional objects (viewed
perpendicularly to their plane) have no perspective at all.

I think you have misunderstood something about that article. If you think I
have missed something, which is always possible, please direct me to it. I
saw nothing in it about any "spherical surface" such as you mention, but for
the sake of time I only skimmed the article quickly.

With ordinary rectilinear lenses there is no "spherical surface" to be
concerned with. The plane of front focus (i.e., the area of sharpest focus
in front of the lens) is in fact just that, a PLANE, or as close to a plane
as the lens designers and makers have been able to make it.

A fisheye lens on the other hand evidently does "see" the world as the
inside of a hemisphere. When doing lens testing with a 16mm fisheye and the
familiar 1951 USAF test targets, with the central target a few feet in front
of the lens I had to arrange the others in a semicircle around it in order
for them to be in focus. That is emphatically not the case with any
well-made rectilinear lens, no matter how wide an angle.


 
Reply With Quote
 
Jeremy Nixon
Guest
Posts: n/a
 
      08-04-2004
Nostrobino <(E-Mail Removed)> wrote:

> With ordinary rectilinear lenses there is no "spherical surface" to be
> concerned with. The plane of front focus (i.e., the area of sharpest focus
> in front of the lens) is in fact just that, a PLANE, or as close to a plane
> as the lens designers and makers have been able to make it.


A plane is accomplished through (intentional) distortion. You're still
photographing a sphere.

--
Jeremy | (E-Mail Removed)
 
Reply With Quote
 
Nostrobino
Guest
Posts: n/a
 
      08-04-2004

"Jeremy Nixon" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Nostrobino <(E-Mail Removed)> wrote:
>
> > You are still hung up on that "spherical image" misconception. There

isn't
> > any spherical image where rectilinear lenses are concerned.

>
> Of course there is. If you take every point that is a certain distance

away
> from the lens, they will create a sphere. You are projecting that sphere

onto
> a flat surface. You get distortion when you do this.


You if that's what was happening, but it isn't (except with a fisheye lens).

As I just mentioned to Bart, the surface of front focus is a plane, not a
sphere.


>
> > What is projected onto the flat plane is a two-dimensional

representation of
> > a three-dimensional world.

>
> Imagine that you are photographing a building. In the lower left corner

of
> the image, part of the structure forms a right angle. You've used a

wide-angle
> lens to get your "wide-angle look". Is the right angle in the world still

a
> right angle in the picture? No, it's not. It's not whether you used a

wide-
> angle lens or not, but I had you using one just to factor out that part of
> the discussion.


It depends. If I'm photographing the building face on, i.e. perpendicularly
to the front of the building, then any right angle in the lower left corner
(or elsewhere) is still a right angle.

If I'm photographing obliquely it is not, because horizontal lines converge
into the distance. That's perspective. A full frontal photo (if the subject
were absolutely two-dimensional, no projections etc.) would have no
perspective at all.

>
> Why is a right angle in the corner of the image not a right angle in the
> picture?


Again, whether it is or isn't depends on how the photo was taken.


>
> Because you've flattened the spherical image onto a flat plane, and the


WHAT "spherical image"? There isn't any spherical image.


> resulting distortion is greatest in the corners of the image.
>
> Why is there such a thing as a "wide-angle look"? Why do wide-angle

lenses
> show greater distortion at the edges than telephoto lenses?


Actually they don't. That is to say, there really isn't any distortion
(assuming a good-quality wide-angle lens). The apparent radial stretching of
objects that increases as they approach the corners results from a
difference in point of view. If viewed from the "proper" position (a
position analogous to that of the lens relative to the subject when the
picture was taken), the apparent distortion would disappear.

Try it. Look at a photo with that wide-angle "distortion" and move your eye
closer to the center of the picture. At some point the corners will look
perfectly normal.


>
> Because they are capturing a greater area of the spherical image, so there
> is more curvature in the original that is projected onto the flat plane,


There is no curvature.

Objects exist in three-dimensional space. A lens produces a two-dimensional
representation of that space on the film. The size, shape, angles and
distances of the objects on that two-dimensional representation are
collectively what is meant by perspective. That's all there is to it.


> and the relative angle between image plane and subject is greater at the
> edges with the wider field of view, so there is more distortion at the

edge
> of a wide-angle image. This distortion is not absent in a telephoto

image,
> you just can't see it as obviously because you're using a smaller portion
> of the sphere.
>
> In the above example, if you use a fisheye lens, the right angle *will* be
> a right angle in the picture.


No, I assure you it will not. It will be even farther from a right angle
than it was in the oblique photo taken with the rectilinear lens.


> You've used a different projection, one that
> maintains right angles. The perspective is the same, but the projection

is
> different, so the distortion is different -- straight lines will not

remain
> straight.
>
> So, in one image, a right angle in the world is not a right angle in the
> picture. In the other, a right angle is a right angle, but straight lines
> are not straight. Which one is an "accurate" representation of a three-
> dimensional world?


Your premise being wrong, your conclusion is wrong. A fisheye lens will not
reproduce your right angle in the lower corner of the image. Far from it.


>
> >> this results in distortion of some kind, depending on what type of
> >> projection you use. When you use a projection that keeps straight

lines
> >> straight (your "normal rectilinear lens") then parallel lines will not
> >> remain parallel.

> >
> > Sure they will, if the subject is essentially two-dimensional (i.e., any
> > flat surface that's perpendicular to the lens axis). If the subject is

NOT
> > flat and perpendicular then of course parallel lines in it will

converge.
> > This is NOT DISTORTION. It is an accurate two-dimensional representation

of
> > what is in front of the lens.

>
> The converging parallel lines are not distortion, but are not necessarily

an
> accurate representation of what is in front of the lens.


Sure they are. They are just what you see with your eyes.


> Why do architectural
> photographers use shift lenses, if converging parallel lines are accurate?


To remove the unwanted perspective effect.


>
> The distortion is a product of flattening the image. The converging

parallel
> lines are a product of the relative angle of the image plane to the

subject.
> The flattening distorts everything based in part on this angle, which is

why
> things at the corner of your wide-angle shots are "different" from things

in
> the center of them. It's why right angles can't stay as right angles

unless
> the object is perpendicular to the lens axis, as long as you're using a
> "normal" lens.
>
> It's *all* distorted. It's just a matter of which distortion you want.

It
> is not possible to create a flat image as a photograph without distorting
> *something*, because the lens is looking at a sphere, not a plane.
>
> If you focus your lens to 10 feet, what are you photographing? It's not a
> flat plane 10 feet in front of the camera, parallel to the image plane.


Yes, that's PRECISELY what it is, with a rectilinear lens.


> It
> is the inside surface of a sphere 10 feet from the lens in all directions,


Emphatically NOT! That is true with a fisheye lens (as I've mentioned
elsewhere), but definitely not the case with a rectilinear lens.


> and you're photographing a portion of that surface by flattening it, which
> distorts the image in much the same way that creating a map on a flat

piece
> of paper makes Greenland look as big as Africa. They use different types
> of projection in map-making, but it's the same idea -- representing a
> sphere on a flat surface.


Jeremy, this appears to be the root of your misunderstanding, or at least
part of it. Get a book on photographic optics and theory. I don't know where
on earth you got these misconceptions, but you are WILDLY in error.

Don't take my word for it. Start another thread in this newsgroup and pose
the question, "Is the field of best focus in front of a rectilinear lens a
FLAT PLANE, or SPHERICAL?"

There is no point in continuing with this until you've gotten yourself
straightened out on this.


 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
how to restore the original perspective in eclipse david wolf Java 2 12-10-2012 09:20 AM
Perspective transform on glyphs? daniel.w.gelder@gmail.com Java 0 01-02-2006 08:55 AM
Most 35mm format lenses can't cover 35mm full frame DSLR? einst_stein@yahoo.com Digital Photography 8 06-03-2005 11:25 PM
Is there any perspective filter for images? Michel HTML 9 06-16-2004 08:41 AM
Any Eclipse equivalent to WSAD server perspective? Jay Eckles Java 1 02-19-2004 08:39 AM



Advertisments