Perspective Revisited

[Ray]

There has recently been a bit of controversy on 'perspective' in another couple of threads, one entitled, 'This puzzling business of 35mm lens equivalent", and the other 'Does a photo give spatial information?'

There seems to be a mantra that perspective is related only to distance beteen the observer (photographer) and subject, which is difficult to break through. It's almnost like a religion.

I really think the matter should be sorted out. There seems to be so much confusion on the issue. Below I'm attempting to enumerate all the relevant points that have occurred to me during the recent discussion in those threads I mentioned above..

(1) There's a mathematical and geometric definition of perspective which is useful and necessary for all sorts of drafting and sketching and the creation of computerised programs to convert 3-D into 2D, or to stitch images together etc etc.

(2) There's a human experience of perspective which may be at odds with that scientific definition of perspective.

(3) There's a reason for the discrepancy. Human vision is enormously complicated. Simple geometric rules cannot encompass it.

(4) I'm reminded of the current controversy over Mark Dubovoy's article in which he claims that 'Everything Matters'. This is relevant to the discussion on perspective, as experienced in the human mind.

Mark's point, as I understood it, was that seemingly trivial details can have a surprisingly significant effect. We shouldn't ignore them.

(5) Those who claim adamantly that perspective can only be changed by a change in distance to subject should make clear that they are referring to a mathematically and geometrically abstract defintion which does not necessarily encompass the human experience which is plain and simply, and unavoidably, biased in accordance with its own sense of brain-wired perspective.

(6)The classic example of proof for the statement that perspective cannot be altered without a change in position, is the cropping of a wide-angle shot to the same angle of view as a telephoto shot.

The adherents of this simplistic approach to perspective will point out that the cropped wide-angle shot will have the same perspective as the telephoto shot, as evidenced by the same broad size and shape of objects and angles apparent in both shots.

However, those of us who don't lack a subtle apprehension of detail, will notice that the cropped wide-angle shot is a bit fuzzy and lacking in detail, compared with the telephoto shot, which is sharp and clear.

Such indistinctness in the cropped wide-angle shot is indicative of greater distance. Clarity and sharpness, or to quote Mark Dubovoy, hyper-realism, is indicative of closeness.

Great painters realise this fact. If you want to depict something as being rather distant, the last thing you do is make it sharp and detailed, which a telephoto lens does.

I could go on, but I'll leave it at that for the sake of brevity.

[32BT]

Any links to the original threads? This seems to make little sense by itself. Lenses usually are sharper when cropped, but you lose some resolution. Whether one compensates the other?

First thing that comes to mind:

A lens has a curved surface. If that curvature is exactly spherical and conforms to its field-of-view then (and only then) is there a direct relation between perspective and cropping, and will it be relatively simple to mathematically correct and stitch. But most modern lenses aren't spherical and cheap lenses may even have distortions that could be visible to the casual observer. In a fish-eye lens there is usually less need to compensate distortions, so it is better suited to mathematical corrections where field-of-view remains in a known state.

Is that the type of issues that the threads where referring to?

[Jim Pascoe]

Hi Mark

I was involved in a very small way in the thread you mention, and as I remember it the original point was about using lenses on different format sensors and how that related to the apparent perspective.  I think one of the points was for example if you used a 25mm lens on a MFT camera to photograph a face so that the face fairly well filled the frame, then the nose would be a bit too big looking in the picture, caused by perspective and the fact that we are too close to the subject.  If you then put the 25mm lens on a FF camera and took the picture from the same position (of course you would have a wider field of view), and then cropped the resulting picture to show the same size face, then the pictures would effectively look identical.  This was essentially what was being said.  And I think that is right.  This simple test basically proves that perspective is affected by distance to the subject and not the focal length of the lens.  Of course a wider lens will show much more of a scene and may artificially give a much greater sense of 'depth', but for a subject that is central in the image and could therefore be cropped down, the perspective is unchanged.

The previous thread seemed to degenerate into talking about lots of different things which might loosely be described as being about perspective, but if terms are used loosely enough they can be used to mean almost anything to different people.  One example might be the term 'aerial perspective', which I think might be used to describe the gradual reduction in clarity of distant objects in a landscape photograph caused by atmospheric haze. In a wide angle shot this will give a further feel about how far away distant objects are and can be quite effective.  But even this will be unaffected if you crop down a wide shot to the equivalent of a tele shot.  As long as both lenses are focussed on the same distant point the 'fuzziness' will be the same.  Of course this is an extreme example and obviously resolution and focussing errors would be at play here too.

But I think the sort of perspective originally being discussed in the earlier thread was concerned with geometric perspective, and that is only affected by distance from lens to subject.

In my opinion.

Jim

PS Having just re-read another part of one of the mentioned threads, I think you are also stating something that is very obvious about 'apparent' perspective.  So let us say that I shoot a group of distant building with a 200mm lens so that the buildings pretty much fill the image.  If I show you that picture you have no real way of knowing how far away they are unless I tell you the focal length of the lens and you could then make an educated guess.  If I also take a picture with a 21mm lens from the same spot and show that to you, it will become immediately obvious that the buildings are quite a long way off because you can see everything that is between the lens and the buildings.  I think that is what you mean by perspective - am I right?   But that is a different scene with a different lens, and if you crop down they will look the same, and if you could project outwards the lines from the 200mm shot you would get the same perspective.

[Bart_van_der_Wolf]

Any links to the original threads? This seems to make little sense by itself.

I agree, people should be clear about what they are referring to, unless their intention is to start all over.

This is what sparked Ray's urge to start this new thread:
Does a photo give spatial information (the nose job)?
and presumably
This puzzling business of "35mm lens equivalent"

Cheers,
Bart

[Ray]

Fred,
That's an excellent reiteration of the mantra that only position affects perspective. However, I'm speaking from the human perspective.

There are lots of issues other than 'real'  distance that influence the sense of perspective in the human brain. I mentioned one of them, specifically resolution, which you completely ignored.

Do I have to repeat that sharp objects appear closer, to the human brain, than fuzzy and indistinct objects, despite the geometry.

Do I need to mention that large objects visible in the foreground tend to make other objects that are further away, seem even further away than they actually are?

Do I also need to mention if those larger objects in the foregrounds are cropped out of the image, the perspective of the remaininmg image has changed?

Fred, you are just repeating a false notion which you have been taught but which you have never questioned, it seems to me.

[32BT]

I agree, people should be clear about what they are referring to, unless their intention is to start all over.

This is what sparked Ray's urge to start this new thread:
Does a photo give spatial information (the nose job)?
and presumably
This puzzling business of "35mm lens equivalent"

Cheers,
Bart

Okay, thanks.

Perhaps the following thought may be of interest:

A 180degr fisheye is usually indicated as 8mm, but 180degr field of view has a 0mm focal point. So, I generally read the 8mm as an indication of the distance to the plane of focus. If you attach the lens to a cropped sensor, at the correct distance, it will capture a sharp image of a smaller field of view, but if you attach the lens closer to the cropped sensor, it may still capture 180degr FOV, but the image may well be fuzzy.

I believe issues like these are far more relevant than any psycho-accoustics that may be at play.

[j-land]

There has recently been a bit of controversy on 'perspective' in another couple of threads, one entitled, 'This puzzling business of 35mm lens equivalent", and the other 'Does a photo give spatial information?'

There seems to be a mantra that perspective is related only to distance beteen the observer (photographer) and subject, which is difficult to break through. It's almnost like a religion.

I really think the matter should be sorted out. There seems to be so much confusion on the issue.

Agreed, many are confused, but often due to technical misconceptions or semantic differences.

Below I'm attempting to enumerate all the relevant points that have occurred to me during the recent discussion in those threads I mentioned above..

(1) There's a mathematical and geometric definition of perspective which is useful and necessary for all sorts of drafting and sketching and the creation of computerised programs to convert 3-D into 2D, or to stitch images together etc etc.

It's also fundamental to the creation of images by thing we call "photography".

(2) There's a human experience of perspective which may be at odds with that scientific definition of perspective.

Human visual perception is in many ways very complex and not as "objective" as imaging by a machine. Witness the many optical illusions that play on the brain's processing of visual information and the fact that we exist and process things in a temporal environment (mentioned, I think, in another thread). Of course one's psychological state at any given moment will affect perception, i.e. when under stress or in a "fight or flight" situation, your brain's visual filtering will be quite different than when you're relaxed and taking in the scenery on a sunny day. Are you using the term "perspective" to mean "perception"?

(3) There's a reason for the discrepancy. Human vision is enormously complicated. Simple geometric rules cannot encompass it.

Sure...

(4) I'm reminded of the current controversy over Mark Dubovoy's article in which he claims that 'Everything Matters'. This is relevant to the discussion on perspective, as experienced in the human mind.

Please elaborate...

Mark's point, as I understood it, was that seemingly trivial details can have a surprisingly significant effect. We shouldn't ignore them.


(5) Those who claim adamantly that perspective can only be changed by a change in distance to subject should make clear that they are referring to a mathematically and geometrically abstract defintion which does not necessarily encompass the human experience which is plain and simply, and unavoidably, biased in accordance with its own sense of brain-wired perspective.

Those claims about perspective are (largely, in the thread you mentioned) addressing the questions about equivalent focal length which is fundamentally a geometric phenomenon. Besides, the geometric definition is not so abstract - when you stand looking down some railway tracks or at the bottom of a tall building, it's quite apparent that the parallel lines converge (in my experience, anyway). I would propose that it was observation and curiosity about human vision that led to the development those definitions of perspective in the first place. I don't quite know what you have in mind and are vaguely alluding to, or is it that you haven't clearly formed your ideas yet?

(6)The classic example of proof for the statement that perspective cannot be altered without a change in position, is the cropping of a wide-angle shot to the same angle of view as a telephoto shot.

The adherents of this simplistic approach to perspective will point out that the cropped wide-angle shot will have the same perspective as the telephoto shot, as evidenced by the same broad size and shape of objects and angles apparent in both shots.

However, those of us who don't lack a subtle apprehension of detail, will notice that the cropped wide-angle shot is a bit fuzzy and lacking in detail, compared with the telephoto shot, which is sharp and clear.

Such indistinctness in the cropped wide-angle shot is indicative of greater distance. Clarity and sharpness, or to quote Mark Dubovoy, hyper-realism, is indicative of closeness.

It seems like you're observing differences in resolution where, assuming the wide angle and telephoto shots are made using the same sensor or film emulsion, the cropped part of the wide angle image needs to be enlarged to match it to the telephoto image, resulting in loss of detail. If I change my telephoto lens to another lens of the same focal length that isn't as sharp and has more flare, is the fuzzy and less distinct photograph from the second lens more "distant" than that from the first, even though they encompass exactly the same field of view? What are you getting at with your use of the word "distance"?

Great painters realise this fact. If you want to depict something as being rather distant, the last thing you do is make it sharp and detailed, which a telephoto lens does.

Sounds like you're talking about aerial perspective, where distant things lose contrast (and possibly resolution) due to scattering of light in the atmosphere. It is indeed a device used by artists to give the perception of depth in an image. Exactly the same thing happens in photographs. If you're an a hill overlooking a smoggy city the bushes in the foreground will be contrastier than the distant building. When you switch from wide angle to telephoto, the telephoto has to look through exactly the same depth of smoggy air as the wide angle and tje distant building will show the same loss of contrast. Any differences will be due to characteristics of the different lenses or the aforementioned effects of enlargement. It's possible to make a photograph with just about any lens where the distant things are sharp and the foreground is out of focus, so detail or lack thereof is not necessarily indicative of distance. In art and 2-dimensional imaging there are many devices used to give cues for distance, many of which have been touched upon in the threads you mention. It's the ability to play around with these devices in sometimes contradictory ways that has resulted in interesting works of art. There are many books about the subject. If you want to call this "perspective", fine, but don't expect any consensus on that definition.

I could go on, but I'll leave it at that for the sake of brevity.

Good idea.

[Ray]

PS Having just re-read another part of one of the mentioned threads, I think you are also stating something that is very obvious about 'apparent' perspective.  So let us say that I shoot a group of distant building with a 200mm lens so that the buildings pretty much fill the image.  If I show you that picture you have no real way of knowing how far away they are unless I tell you the focal length of the lens and you could then make an educated guess.  If I also take a picture with a 21mm lens from the same spot and show that to you, it will become immediately obvious that the buildings are quite a long way off because you can see everything that is between the lens and the buildings.  I think that is what you mean by perspective - am I right?   But that is a different scene with a different lens, and if you crop down they will look the same, and if you could project outwards the lines from the 200mm shot you would get the same perspective.

Perhaps the confusion results from a lack of a clear definition of perspective. The general definition of perspective is perhaps too narrow in the sense that it only addresses distances and angles.

It seems that resolution, texture, Dof, FoV have no bearing on this geometric definition of perspective that most people on this forum seen to slavishly accept, but I suspect their eyes don't.

[Walter_temp]

If you try to cover all traps and pitfalls of human vision: Good luck!
The original thread started with the misconception of small sensor sizes and the big nose effect, which was settled in the discussion (I thought). This part was "correlation between sensor size, focal length and distance" for the picture taken by the camera.

Now you're trying to turn the whole thing upside down and - sorry to say - there are some open questions that will not be solved by you, neither any of us. The thing has been discussed a long time ago. Take a look into some books written by Andreas Feininger. He pointed out some issues where a picture and the human perception doesn't match. This one is easy to reproduce: Wide angle lens, person in front of you with arms aiming to camera. The camera will reproduce what the eye is seeing but our brain does have another picture in mind. If you remember such a scene you will not remember hand covering most of the "image" and head reduced to a pinhead. You will remember a person with arms and hands stretched to you and all proportions will be just fine ... in your imagination. But: If you are able to reproduce the scene and just look what your eyes are covering ... well ... this will resemble the picture taken by the camera (mostly).

This is not new knowledge at all. Sometimes I'm just baffled that there are photographers who aren't aware that there is a lot of difference between "eye sight" and "mind sight". There are misconceptions about what the eye sees (as in "lens and perspective") and the inner representation of the world we think we are seeing.

Another well known example is the "Moon illusion" and i think we've all been there at least once: Take a picture of the moon near horizon with a "normal" lens (which most people insist to be equivalent to human sight, another gross misconception IMO) and the moon in the picture will be look like riduced and not giving the perception you had at all.

So what? The eye is a liar and the ear, too! That's what Prof. Pauschinger (RIP) shouted in one of his remarkable entertaining lectures.

And this defines a part of the photographer's job: To know the difference between human perception in real live and the things a camera can do and take the picture that shows what *you* like to show. Beginners will fail here.
Well known example is beginner's question why a picture of a little bird sitting in the tree (bird just covering a little fraction of the picture) just looks awful and the bird is hardly recognizable at all and the picture is not showing what was seen "by the eye".


So: What's this all about? Do you want to lecture us about things as old as photography and partly older?

Ciao, Walter

[j-land]

Ok, you want to talk about distance in pictorial representation as a general definition of perspective. If there's a photo of a landscape with the distant things sharp and an out of focus tree in the foreground, the tree does not look distant despite being out of focus because of it's relative size and it overlaps the things in the background. If the tree is chopped down and another picture is taken, the spacial perception of those distant objects may be different without the tree as a relative gauge - fine. As I mentioned previously, there are many devices that can be used in an image to affect the viewer's perception of distance. This is all pretty standard stuff that has been known and used in art for centuries. There is really nothing fundamentally "false" about what Fred said, and much of what you are asserting isn't false either, but I think if you clarified your semantics a little, there would be a lot less confusion all around.

Fred,
That's an excellent reiteration of the mantra that only position affects perspective. However, I'm speaking from the human perspective.

There are lots of issues other than 'real'  distance that influence the sense of perspective in the human brain. I mentioned one of them, specifically resolution, which you completely ignored.

Do I have to repeat that sharp objects appear closer, to the human brain, than fuzzy and indistinct objects, despite the geometry.

Do I need to mention that large objects visible in the foreground tend to make other objects that are further away, seem even further away than they actually are?

Do I also need to mention if those larger objects in the foregrounds are cropped out of the image, the perspective of the remaininmg image has changed?

Fred, you are just repeating a false notion which you have been taught but which you have never questioned, it seems to me.

[Ray]

Human visual perception is in many ways very complex and not as "objective" as imaging by a machine. Witness the many optical illusions that play on the brain's processing of visual information and the fact that we exist and process things in a temporal environment (mentioned, I think, in another thread). Of course one's psychological state at any given moment will affect perception, i.e. when under stress or in a "fight or flight" situation, your brain's visual filtering will be quite different than when you're relaxed and taking in the scenery on a sunny day. Are you using the term "perspective" to mean "perception"?

Let's just address this issue above. Optical illusions certainly exist, and they are quite amazing. But one essential point about these illusions is that they seem to apply to everyone.

If you wish to make the point that various factors other than mere distance, that might affect the human brain's perception of perspective, are purely an illusion, then we are into a branch of philosophy rather than photography, which I'm quite willing to discuss.

Are you making the point that the extreme and contrived examples that trick the eye, such as people simultaneously walking up and down a staircase which is on the same level, are what we normally see in photographic images? In other woerds, everything is an illusion except the geometry of perspective?

[j-land]

My point, somewhat agreeing with you, is that tone and contrast (and not just pure geometry), among numerous other things, can affect pictorial perception of distance, and we often use these effects in images through manipulations as simple as burning and dodging. By optical illusions I don't mean just Escheresque creations, but examples like simultaneous contrast show how our brains have evolved to process and perceive visual information (at some point these things were probably useful to distinguish our potential prey and predators in the forest). These concepts can be used in creating images and "fool" the eye in order to exaggerate or force a certain illusion of spacial relations in a photograph. In some ways I think you're preaching to the converted, but just need to clarify the terms your using and what they mean to you. I would question the value of bringing this aspect of "perspective" into the topic of equivalent focal lengths... and would not be so derisive of geometric perspective in that context.

Let's just address this issue above. Optical illusions certainly exist, and they are quite amazing. But one essential point about these illusions is that they seem to apply to everyone.

If you wish to make the point that various factors other than mere distance, that might affect the human brain's perception of perspective, are purely an illusion, then we are into a branch of philosophy rather than photography, which I'm quite willing to discuss.

Are you making the point that the extreme and contrived examples that trick the eye, such as people simultaneously walking up and down a staircase which is on the same level, are what we normally see in photographic images? In other woerds, everything is an illusion except the geometry of perspective?

[Bart_van_der_Wolf]

Perhaps the confusion results from a lack of a clear definition of perspective. The general definition of perspective is perhaps too narrow in the sense that it only addresses distances and angles.

Dear Ray,

Perspective is only about distances and angles, as seen from a unique vantage point (the perspective point or entrance pupil of a lens in the case of photography). The perception of perspective can be confused for an observer by various added distortions.

It seems that resolution, texture, Dof, FoV have no bearing on this geometric definition of perspective

Correct.

... that most people on this forum seen to slavishly accept, but I suspect their eyes don't.

Optical illusions do not change geometry, only the perception of it (if at all). Just because you close your eyes, doesn't make the scene go dark for most of the other observers...

Cheers,
Bart

[Ray]

I would question the value of bringing this aspect of "perspective" into the topic of equivalent focal lengths... and would not be so derisive of geometric perspective in that context.

I'm not trying to be derisive. I'm simply searching for clarity on the issue. I could probably accept a definition of perspective along the lines, "Perspective is affected only by distance to subject. Any other factors which might suggest otherwise, exist only in the mind of the viewer and are pure illusion."

The second part needs to be stressed. The implication is, the extreme examples of perspective distortion that most of us have seen, such as two normal people in a room who look hugely different in size, are just that; extreme examples of a distortion that is normal in everyday vision. In other words, everything we see is a distortion to some degree, according to the geometric rules of perspective.

[Ray]

"Perspective is affected only by distance to subject"

Noooo..... Arrghh.

On the other hand, I might be a little worried about accepting a concept that everything I see is a distortion of reality in relation to the purity of geometry, when the science of geometry is itself a product of the human mind, and therefore subject to distortion.

[Ray]

Then you have no other option than to ask the gods themselves. (the responsible "thing" whatever it is of this universe and its laws).

Well, I do have the option of just ignoring the theoretical rules of geometry and doing what I damned well please. 

[Bart_van_der_Wolf]

Well, I do have the option of just ignoring the theoretical rules of geometry and doing what I damned well please.  

The rules of projective geometry are not a theory, they are based on axioms.
If you don't agree, well, how about ignoring the 'theory of gravity'. I wish you a speedy recovery

Cheers,
Bart

[C Debelmas]

I'm afraid to say that you are comparing pears and apples.
Forget the word perspective which is confusing because it has many meanings.
Just consider the way a 3D object is projected on 2D surface (the film or the sensor). The shapes that you obtain on the 2D surface is only affected by the "point of view", i.e. the position of the 2D surface in respect to the 3D object. As you have to consider lenses in photography (you need lense to create a projection on the film or the sensor), you may use wide angle lenses or telephoto lenses. Then what is at sake is that when you change the focal length you don't change the shape on the 2D surface (expect if you consider the possible croping and the possible distorsion of the lenses).
Then consider the "feeling of distance" that you may have from your position with respect to the 3D object : if you rely on the shape of the 3D object projected on the 2D surface, you will have no clue about the distance. If you have other informations, like the color of the object or the blur then you may have a clue on the actual distance.
One thing is the geometry of the 2D shape, another thing is the aspect (color, blur) of the 2D shape.

[32BT]

Would be funny to sent a Lytro picture of a 2D image to Lytro and ask them why the refocusing doesn't work…

[Ray]

The rules of projective geometry are not a theory, they are based on axioms.
If you don't agree, well, how about ignoring the 'theory of gravity'. I wish you a speedy recovery

Cheers,
Bart

C'mon Bart,
You haven't thought that through, have you!  One doesn't need to know anything about the theory of gravity in order to negotiate the problems of gravity on a personal level. There were circus performers long before Isaac Newton came on the scene with his theory of gravity, who were able to perform amazing balancing acts without any theoretical knowledge of gravity whatsoever.

More likely, I would probably need a speedy recovery if I were to spend too much time contemplating the mystery of the gravitational force as I walked down the street. I might fall into a pot-hole through not paying attention to what I was doing. 

As regards axioms, it's axiomatic to me that all theories are a construct of the human brain and imagination.They represent models of reality which are usually found to be, eventually, either a bit wrong or completely wrong.

Astrophysicists are currently having a hard time explaining why 95% of the matter and energy in the universe is invisible and undetectable. The euphemistic term is Dark Matter.

There's a consensus of opinion amongst Astronomers and Astrophysicists that Dark Matter and Dark Energy really does exist, despite the fact that not one particle has been discovered so far. On the other hand, there's a growing number of impatient scientists who think our theories of gravity may be wrong, and who have attempted to modify Newton's theories. The acronym for the alternative theory to Dark Matter is MOND (Modification Of Newtonian Dynamics).

Cheers!