Pages: 1 ... 5 6 [7]   Go Down

Author Topic: Perspective Revisited  (Read 24382 times)

Isaac

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3128
Re: Perspective Revisited
« Reply #120 on: February 14, 2012, 12:15:35 PM »

Distant objects appear small. They are perceived as being small. No-one believes they actually are small, unless one is looking at a world of totally unfamiliar objects.
I see distant objects that appear small. I perceive them to be small. I believe they actually are small and I'm looking at a world of totally familiar objects. Leaves 3"x1" - 40 yards distant.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 9380
Re: Perspective Revisited
« Reply #121 on: February 15, 2012, 03:16:59 AM »

Ray,
That flat image can now be cropped in various ways, printed at various sizes and those prints viewed at various distances by various observers (each of whom can bring any number of cultural assumptions to the experience.) In each case, their individual perception of the image may be different, but the *recorded image doesn't change*. The printed pixels do not shuffle about on the paper based on the distance of the viewer from the paper.


John,
If you crop the printed image, a good proportion of the printed pixels do something even more dramatic than shuffling about. They do a complete disappearance act. You think that has no effect of perspective?  ;D

There seem to be some very strange and logically absurd arguments presented in this thread in defense of the statement that a change in focal length of lens, or a change in FoV, has no bearing on perspective.

I get a sense that there is a disembodiment of the viewer taking place in such arguments. In other words, if we remove the viewer from the scene, the perspective doesn't change with changing FL of lens or changing FoV (whichever terminology you prefer.)

The classic proof of the argument that a change in focal length of lens does not affect perspective is to demonstrate that images taken with effectively equal focal lengths of lens are the same. Well, of course they are. No argument there.

To overcome this obvious non-proof, I get a sense of a lot of wriggling and squirming going on.

Guillermo claims that 'equivalent focal length' is a fiction. It doesn't exist. All we have are different Fields of View. However, that argument doesn't change the reality. I'm quite willing to use different terminology. If the terminology 'equivalent focal length' is not acceptable and you prefer FoV, then that's fine by me. I'll simply rephrase my argument that a change of FoV results in a change in perspective, knowing quite well that any change in FoV when shooting from a specific position must result from a change in FL of lens.

The very technically knowledgeable Bart, argues that it can all be explained by anamorphism, but he refuses to elaborate. He simply  implies that anamorphic distortion may be an intrinsic property of all lenses, not just the obvious volume anamorphic distortions that one sees at the edges of  very wide-angle shots, which can be corrected by converters from DXO Labs.

If this is the case, that the anamorphic distortion of all lenses, perhaps other than the 'standard' lens, contribute to a distortion of perspective, and that nothing can be done about this, then surely one has to admit that a change in focal length inevitabley results in a change in perspective, because nothing can be done to correct the anamorphism.

However, those who claim that such 'apparent' changes in perspective resulting from a change in FL or FoV, can be corrected by changing the viewing distance to a resulting print, really take the cake.  ;D

The very fact that one has to change viewing distance to the print in order to compensate for perspective changes resulting in changes in FL or FoV, is proof that a change in FL affects perspective.

Crikey! Get real you guys for Christ's sake!  ;D
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1564
    • http://www.guillermoluijk.com
Re: Perspective Revisited
« Reply #122 on: February 15, 2012, 05:15:50 AM »

7 pages of semantic discussion about perspective. That's an achievement by itself Ray, congratulations!  ;D
Logged

BartvanderWolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5888
Re: Perspective Revisited
« Reply #123 on: February 15, 2012, 07:17:24 AM »

The very technically knowledgeable Bart, argues that it can all be explained by anamorphism, but he refuses to elaborate.

I do not refuse to elaborate on something that has been known since the early Renaisance (15th century) and is still as valid today, I just see no point in repeating the available explanations for the umpteenth time because you choose to ignore them.

Quote
He simply implies that anamorphic distortion may be an intrinsic property of all lenses, not just the obvious volume anamorphic distortions that one sees at the edges of  very wide-angle shots, which can be corrected by converters from DXO Labs.

Here you go again, misrepresenting what others have said. Amamorphic distortion is NOT a lens property, it's a projection effect. It is very simple to understand for most, if they are willing. Even keystoning will disappear from an image, provided the output is viewed from the correct viewpoint, just like road surface marking or street paintings.

Because viewing from the correct position (distance and angle) is not alway practical, e.g. a wide angle shot would need to be viewed at a relatively short uncomfortable distance, or because people do not take the effort, we also have a choice to change the output projection somewhat with software (relatively easy with pano stitching software), in an attempt to compensate for using the wrong viewpoint. In fact we can even exploit the assumption that the wrong viewpoint is going to be used, by thus suggesting a different perspective than was actually fixed at the moment of capture. Hence the perceived distortion in wide-angle shots and tele shots, which only occurs due to the wrong viewing position/distance. It is perceived distortion, which you misinterpret as perspective, which it isn't.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 9380
Re: Perspective Revisited
« Reply #124 on: February 15, 2012, 12:22:42 PM »

Here you go again, misrepresenting what others have said. Amamorphic distortion is NOT a lens property, it's a projection effect.

Bart,
In case you didn't notice, this thread is about the change in perspective that results from a change in focal length of lens, yet you deny that this change takes place on the basis of some phenomenon called anamorphic distortion that is not a property of lenses.

Sounds to me you are engaging in obfuscation.

Here are some definitions of anamorphosis that relate to images:

(1) A distorted projection or perspective; especially an image distorted in such a way that it becomes visible only when viewed in a special manner.

(2) A distorted projection or drawing that looks normal from a particular angle or with a certain mirror.

The essential characteristic of anamorphism is a change in shape, not a change in size or apparent distance which a change in focal length of lens can produce.

As you probably know, there are types of lenses called anamorphic, which are designed to deliberately distort the shape of objects in order to fit a horizontally wide angle of view onto a 4:3 format. On projection, the reverse type of anamorphic lens is used to horizontally expand the unnaturally narrow objects as recorded, so that they look natural again in a widescreen format. A common type of anamorphic distortion is a 4:3 format TV transmission which is displayed on a widescreen TV without adjustment, causing all figures to appear unnaturally fat.

What this has to do with a change of FoV containing objects which are not misshapen, beats me. I really do think you need to eloborate on this.

Cheers!  Ray

Logged

BartvanderWolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5888
Re: Perspective Revisited
« Reply #125 on: February 15, 2012, 02:26:39 PM »

Sounds to me you are engaging in obfuscation.

LOL, you remind me of that guy listening to the traffic alert on his car radio, as he says to himself:"What are they talking about, one ghost rider?, there are hordes of them coming my way and what's worse, they have the audacity to greet me with their headlights".

Quote
What this has to do with a change of FoV containing objects which are not misshapen, beats me. I really do think you need to eloborate on this.

Again?

I'll recap:
Focal length --> magnification,
sensor dimensions --> field of view,
lens entrance pupil --> perspective,
output viewing distance --> perspective distortion.

That's all there is to it. I'm sorry if the facts confuse you.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

theguywitha645d

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 970
Re: Perspective Revisited
« Reply #126 on: February 15, 2012, 04:08:10 PM »

In each case, their individual perception of the image may be different, but the *recorded image doesn't change*. The printed pixels do not shuffle about on the paper based on the distance of the viewer from the paper.

Are you sure about that? But maybe the pixels do not need to move. Perhaps you can help. Here is an image with two objects, A and B. The image is viewed at different distances, 1 and 2. Would the ratio of angular sizes of A and B be the same at each position? Does viewing distance change the image?
« Last Edit: February 15, 2012, 04:26:56 PM by theguywitha645d »
Logged

theguywitha645d

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 970
Re: Perspective Revisited
« Reply #127 on: February 15, 2012, 04:15:40 PM »

I'll recap:
Focal length --> magnification,
sensor dimensions --> field of view,
lens entrance pupil --> perspective,
output viewing distance --> perspective distortion.

That's all there is to it. I'm sorry if the facts confuse you.

Cheers,
Bart

Focal length --> Magnification
Sensor size and focal length --> Field of view
The ratio of object distances --> ratio of image sizes --> True perspective
The difference between the "correct" viewing distance and actual viewing distance --> Apparent perspective

I think that would be more factual. It would certainly be more in line with the texts.
Logged

BartvanderWolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5888
Re: Perspective Revisited
« Reply #128 on: February 15, 2012, 06:53:01 PM »

Focal length --> Magnification
Sensor size and focal length --> Field of view

That's correct, although one could say that the focal length was already given in the first line. All that the sensor does is capture a crop of the available image circle from that lens, which results in a FOV.

Quote
The ratio of object distances --> ratio of image sizes --> True perspective

It's the same thing, using a few more words to explain. True or actual Perspective is fixed for a given viewpoint (entrance pupil), and all relative sizes result from simple geometry.

Quote
The difference between the "correct" viewing distance and actual viewing distance --> Apparent perspective

Yup, no problem with that, although distance should also encompass viewing angle of the projection plane.

Quote
I think that would be more factual. It would certainly be more in line with the texts.

The issue is that there are so many texts, and they do not all have rectilinear lens photography with planar projections in mind. They often also do not account for things like keystoning, which can be 'corrected' by using the correct viewing position (at an angle versus the planar projection). All deviations from the correct viewing position results in apparent distortions or apparent perspective.

We essentially agree on the same priciples as they were already taught in the early 1400s, I was just trying to be a bit terse.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 9380
Re: Perspective Revisited
« Reply #129 on: February 16, 2012, 01:03:01 AM »

I'll recap:
Focal length --> magnification,
sensor dimensions --> field of view,
lens entrance pupil --> perspective,
output viewing distance --> perspective distortion.

That's all there is to it. I'm sorry if the facts confuse you.

Cheers,
Bart

Bart, I don't feel at all confused. What I see are confusing statements from people defending the position that focal length of lens or FoV have nothing to do with perspective.

Consider your above statements. They're confused. Admit it. There's certainly no doubt that sensor size can sometimes determine field of view, but sometimes it can't. It depends upon whether one is increasing sensor size or reducing it. As you've admitted in a later post, the diameter of the lens image circle limits the maximum FoV.

It's clear to me that the role of the focal length of a lens is equally important in determining field of view, if not more significant in practice than is sensor size.

Given a specific focal length of lens designed for a specific size of sensor, one can reduce FoV through cropping but rarely increase it. However, given a fixed size of sensor, one can either reduce or increase FoV in accordance with the available focal lengths of lenses that fit the camera.

To avoid this confusion and false dichotomy, by ascribing magnification to the role of the lens, and FoV to the role of the sensor, I think it's better to use the 'focal length equivalent' terminology with reference to a standard format such as 35mm format.

I believe also that thousands of people in the photographic industry would agree with me. As I mentioned before, my Panasonic P&S has a marking under the lens, '28mm wide'. This describes the FoV very well in my opinion, and in the opinion of countless others it seems.

Even for many people buying a camera for the first time, who may never have used 35mm format, the equivalent FL in 35mm format terms is still a useful reference, as long as it is understood that 24mm is wider than 28mm and that 40mm is narrower than 28mm.

Below is another confusing statement from you, in reply #138, specifically "relatives sizes result from simple geometry."

Quote
It's the same thing, using a few more words to explain. True or actual Perspective is fixed for a given viewpoint (entrance pupil), and all relative sizes result from simple geometry.

Relative sizes do not result from Geometry, whether simpler or not. Geometry is a passive system of mathematical description. It doesn't cause anything. Long before the principles of Geometry were formulated, our ancestors were aware that the smaller appearance of the size of familiar objects denoted distance. They would have had to have understood this in order to survive. Geometry merely provides an explanation for this phenomenon.

There is also another aspect of additional confusion that results from attempting to join two experiences of perspective which are clearly separate, viewing the actual and real scene through a camera lens, and viewing a resulting 2-dimensional print of that scene in a completely different environment.

These are two separate activities. Whatever I do when processing an image to make a print, and whatever decisions I make as to paper type, print size, contrast and color saturation etc, etc, etc, and whatever placing of that print on whatever wall and whatever viewing distance is chosen by the viewer, are all factors that are only tenuously connected with the original experience of perspective in the viewer and the photographer as he took the shot.

Once a print is produced, it becomes a separate 2-dimensional object in a different environment. The same principles of perspective will apply to that object as to any other 2-dimenional object in the same environment, such as a closed window or a stain on the wall.

To argue that by changing viewing distance to the print, in order to correct for an 'unnatural' or 'distorted' perspective in the original image that may have resulted from the choice of a particular focal length of lens, proves that focal length of lens has no bearing on perspective, is ridiculous.

That's pure sophistry, Bart, in the modern sense of the term.

The very fact that one could jump through such hoops when viewing a print, in order to correct for unnatural perspective, rather than just viewing the print from a usual and normal distance equal to, say 1.5x the print diagonal, tends to prove that a change in focal length or FoV without changing position, does indeed have an effect on perspective. If there's no change, there's nothing to correct.


Logged
Pages: 1 ... 5 6 [7]   Go Up