DoF and Perspective Revisited

[ThierryH]

it seems from here in Switzerland too, I have been having this feeling yesterday night.

 

Thierry

... (The moon seems to have little perspective, viewed from the UK.) ...

[Wim van Velzen]

Seen from the Netherlands, the moon has even less perspective than from Switzerland, which is a few feet nearer to the moon  

[Ray]

This is just silly, how do you explain the images in post #23 of this thread.  When stitching is involved there is no defined FOV for a particular focal length.  The FOV will be whatever you want it to be. Your definition of perspective is not the commonly accepted definition.

How do I explain the images in post #23? They were all taken with the same effective focal length of lens. The images taken with the shorter focal lengths were cropped to the same FoV as the longest focal length. I accept that cropping an image from any lens effectively increases the focal length of the lens.

What happens if you don't crop the image taken with the shorter focal length? Here are two pairs of images below. Each pair of images was taken with the same physical lens, same camera, same position, same time. You could consider the cropped image as the first of a series of shots taken with an effectively longer focal length lens for stitching purposes.

If you consider that FoV has no bearing on perspective, then you would presumably consider that both images in each pair below have the same perspective. Is this correct?

[attachment=16433:Angkor_c...rop_7526.jpg]  [attachment=16434:Angkor_c...dor_7526.jpg]

[attachment=16435:4463_crop.jpg]  [attachment=16436:4463.jpg]

Your definition of perspective is not the commonly accepted definition.

I'm using the Wikipedia definition as follows:

The two most characteristic features of perspective are that objects appear, (1)Smaller as their distance from the observer increases.  (2)Foreshortened: the size of an object's dimensions along the line of sight are relatively shorter than dimensions across the line of sight.

What definition are you using?

[ThierryH]

How do I explain the images in post #23? They were all taken with the same effective focal length of lens. The images taken with the shorter focal lengths were cropped to the same FoV as the longest focal length. I accept that cropping an image from any lens effectively increases the focal length of the lens.

What happens if you don't crop the image taken with the shorter focal length? Here are two pairs of images below. Each pair of images was taken with the same physical lens, same camera, same position, same time. You could consider the cropped image as the first of a series of shots taken with an effectively longer focal length lens for stitching purposes.

You can't compare if you don't crop, as simple as that, but the perspective of all images taken with different focal length lenses and from the same distance/angle IS the very SAME: again, perspective is about converging or diverging lines, not size or reproduction scale.
Now I do accept that such images produce a different visual effect, impression or feeling.

If you consider that FoV has no bearing on perspective, then you would presumably consider that both images in each pair below have the same perspective. Is this correct?

"Strangely" they do! And without any doubt.

I'm using the Wikipedia definition as follows:

The two most characteristic features of perspective are that objects appear, (1)Smaller as their distance from the observer increases.  (2)Foreshortened: the size of an object's dimensions along the line of sight are relatively shorter than dimensions across the line of sight.

The definition of perspective is NOT the size/reproduction ratio of the subject or object photographed, but solely the converging and diverging lines/forms in this reproduction. I am sorry to say that Wikipedia or not, that is a wrong definition.

Ray, I think we all understand your point, but as said by someone else, the perspective has been explained, understood and defined as such and cannot be changed: it is SOLELY the distance/angle of shooting having an influence, nothing else

Best regards,
Thierry

[Jack Flesher]

Here is a guy who posted some pretty good examples using the same lenses at differing distances and different lenses at the same distance with comparative crops:

http://jamesmskipper.tripod.com/jamesmskip...erspective.html

[elf]

You can't compare if you don't crop, as simple as that, but the perspective of all images taken with different focal length lenses and from the same distance/angle IS the very SAME: again, perspective is about converging or diverging lines, not size or reproduction scale.
Now I do accept that such images produce a different visual effect, impression or feeling.



"Strangely" they do! And without any doubt.


The definition of perspective is NOT the size/reproduction ratio of the subject or object photographed, but solely the converging and diverging lines/forms in this reproduction. I am sorry to say that Wikipedia or not, that is a wrong definition.

Ray, I think we all understand your point, but as said by someone else, the perspective has been explained, understood and defined as such and cannot be changed: it is SOLELY the distance/angle of shooting having an influence, nothing else

Best regards,
Thierry

Excellent description.  I can only add that FOV is not really relevant when stitching. The photographer chooses the FOV of the image by the number of images he or she takes and it isn't constrained by the focal length of the lens. A wide angle lens and a telephoto lens can have the same FOV in a stitched image, but one will have a higher resolution.

[Panopeeper]

I suggest a new term of photography, rayspective, the combination of FoV and perspective. Two rectilinear images are rayspectively identical, if they have the same FoV, the same perspective and are viewed in the same size (on paper or monitor), and, of course, if they depict the same scenery.

[Jack Flesher]

I suggest a new term of photography, rayspective, the combination of FoV and perspective. Two rectilinear images are rayspectively identical, if they have the same FoV, the same perspective and are viewed in the same size (on paper or monitor), and, of course, if they depict the same scenery.

          Now that was funny!

[Ray]

You can't compare if you don't crop, as simple as that....

Precisely! This is one of the problems I have with the proof that all lenses produce the same perspective from the same distance to subject. When you crop a scene taken with a shorter lens in order to demonstrate that it has the same perspective as the image taken with the longer lens, you are effectively comparing images taken with the same focal length of lens.

It's like saying, "All lenses whatever their focal length produce the same perspective. I will now proceed to prove this point by comparing two images taken with effectively the same focal length of lens."  What sort of proof is that? As I mentioned before, this is a mere tautology.

If you consider that FoV has no bearing on perspective, then you would presumably consider that both images in each pair below have the same perspective. Is this correct?

[attachment=16435:4463_crop.jpg]  [attachment=16436:4463.jpg]

"Strangely" they do! And without any doubt.

How did you work that out? You didn't by any chance enlarge/crop on the monitor the small figure in the full scene till it was the same size as my crop in the first image above, did you; thus effectively equalising the focal length of the lenses for comparison purposes?  

Supposing I were to make large, equal-size prints of the two images above. Would it be immediately apparent that the small figure in the lower part of the full scene (taken with a 24mm lens) had the same perspective as the blown-up crop next to it, effectively taken with a 150mm lens? Or would you have to change your perspective when viewing both prints in order to determine that the perspective is the same? That is, would you have to view the print of the full scene from a distance of say 10", examine the background and the relative size of the foliage in the background, fix that impression in your mind, then walk back a few steps?
 
I think that might be more difficult than effectively equalising the focal length of the lenses by cropping the full image on the monitor to the same FoV and size as my enlargement of the crop above, thus proving that, when two images are made equal, then it logically follows that they are equal.  

The definition of perspective is NOT the size/reproduction ratio of the subject or object photographed, but solely the converging and diverging lines/forms in this reproduction. I am sorry to say that Wikipedia or not, that is a wrong definition.

What! You mean it is not true that a chief characteristic of perspective is that objects appear smaller the greater their distance from the viewer? I never realised that! Jeez!  I really must have a distorted sense of perspective.  

perspective ......... is SOLELY the distance/angle of shooting having an influence, nothing else

Distance/angle? Isn't angle related to FoV and focal length?

Now I do accept that such images produce a different visual effect, impression or feeling.

But you are quite sure that the different visual effect, impression or feeling produced by the wide-angle lens has nothing to do with perspective.

[elf]

Precisely! This is one of the problems I have with the proof that all lenses produce the same perspective from the same distance to subject. When you crop a scene taken with a shorter lens in order to demonstrate that it has the same perspective as the image taken with the longer lens, you are effectively comparing images taken with the same focal length of lens.

It's like saying, "All lenses whatever their focal length produce the same perspective. I will now proceed to prove this point by comparing two images taken with effectively the same focal length of lens."  What sort of proof is that? As I mentioned before, this is a mere tautology.






How did you work that out? You didn't by any chance enlarge/crop on the monitor the small figure in the full scene till it was the same size as my crop in the first image above, did you; thus effectively equalising the focal length of the lenses for comparison purposes?  

Supposing I were to make large, equal-size prints of the two images above. Would it be immediately apparent that the small figure in the lower part of the full scene (taken with a 24mm lens) had the same perspective as the blown-up crop next to it, effectively taken with a 150mm lens? Or would you have to change your perspective when viewing both prints in order to determine that the perspective is the same? That is, would you have to view the print of the full scene from a distance of say 10", examine the background and the relative size of the foliage in the background, fix that impression in your mind, then walk back a few steps?
 
I think that might be more difficult than effectively equalising the focal length of the lenses by cropping the full image on the monitor to the same FoV and size as my enlargement of the crop above, thus proving that, when two images are made equal, then it logically follows that they are equal.  



What! You mean it is not true that a chief characteristic of perspective is that objects appear smaller the greater their distance from the viewer? I never realised that! Jeez!  I really must have a distorted sense of perspective.  



Distance/angle? Isn't angle related to FoV and focal length?



But you are quite sure that the different visual effect, impression or feeling produced by the wide-angle lens has nothing to do with perspective.

Which of these statements are true?

1. If any portion of two images are identical, then all portions of both images will be identical when those portions are included in the field of view.
2. Focal length of a lens is unrelated to the perspective of an image.
3. Format size of the camera (film or sensor) has no effect on the perspective.
4. Stitched images can have any FOV, including greater than 360 degrees.
5. Perspective is only determined by camera position relative to the subject.

[Murray Fredericks]

Which of these statements are true?

1. If any portion of two images are identical, then all portions of both images will be identical when those portions are included in the field of view.
2. Focal length of a lens is unrelated to the perspective of an image.
3. Format size of the camera (film or sensor) has no effect on the perspective.
4. Stitched images can have any FOV, including greater than 360 degrees.
5. Perspective is only determined by camera position relative to the subject.

1, 2, 3, 5


(but something tells me that this is trick question and 4 is somehow true??)

[Ray]

Which of these statements are true?

1. If any portion of two images are identical, then all portions of both images will be identical when those portions are included in the field of view.

Only true when both images are taken with the same focal length of lens. If one of the two images is taken with a wider-angle lens it will include portions which cannot be included in the other image. If you arrange things so that both images contain the same portions, they are essentially the same images taken with effectively the same FL of lens. Then, of course perspective will be the same.

2. Focal length of a lens is unrelated to the perspective of an image

.

Only true if one excludes the perspective of the person viewing the image.

3. Format size of the camera (film or sensor) has no effect on the perspective.

Depends on the lens used. It may or may not. The sensor without lens makes some very strange images.

4. Stitched images can have any FOV, including greater than 360 degrees.

I would say this is not true.

5. Perspective is only determined by camera position relative to the subject.

Which subject? For the statement to be meaningful you have to specify the subject, so I'll rephrase this statement for you. 5. Perspective is only determined by camera position in relation to a specified subject.

The focal length of the lens, and direction pointed, specifies the subject. Change the focal length of the lens, without changing your position, then you change the nature of the subject. You've shifted the goal posts and all bets are off  

In conclusion, none of your statements are completely true.

[Panopeeper]

(but something tells me that this is trick question and 4 is somehow true??)

Not the question, but the pano is a trick. I have seen such; someone shot the frames with enough delay, that when the camera reached the starting position, the daylight scenery become late evening and then night; thus the shots covered more than 360° and the changing illumination made the impression of different scenery.

This is the only kind of pano over 360° I know of. Of course, the scenery can change on different ways, not only through time delay.

[elf]

1. If any portion of two images are identical, then all portions of both images will be identical when those portions are included in the field of view.

Only true when both images are taken with the same focal length of lens. If one of the two images is taken with a wider-angle lens it will include portions which cannot be included in the other image. If you arrange things so that both images contain the same portions, they are essentially the same images taken with effectively the same FL of lens. Then, of course perspective will be the same.

A single frame of the stitched image will have different FOV when different focal length lens are used. The FOV of a stitched image can be the same for any focal length and (here's the part you don't seem to believe) the perspective will be the same.
.

2. Focal length of a lens is unrelated to the perspective of an image
.
Only true if one excludes the perspective of the person viewing the image.

How does a person have perspective?

3. Format size of the camera (film or sensor) has no effect on the perspective.

Depends on the lens used. It may or may not. The sensor without lens makes some very strange images.

The lens has no knowlege of the sensor, it cannot change the light passing through it based on a sensor located behind it. The only thing that changes is the FOV. Some sensors will be able to resolve more, but they can't change the perspective.

4. Stitched images can have any FOV, including greater than 360 degrees.

I would say this is not true.

Actually it is.  When stitching images you can continue around and around.  Pano heads don't have stops at 360 degrees.

5. Perspective is only determined by camera position relative to the subject.

Which subject? For the statement to be meaningful you have to specify the subject, so I'll rephrase this statement for you. 5. Perspective is only determined by camera position in relation to a specified subject.

The subject can be everything in the image.  It is correct as originally stated. (Actually, your rephrasing didn't change anything)

The focal length of the lens, and direction pointed, specifies the subject. Change the focal length of the lens, without changing your position, then you change the nature of the subject. You've shifted the goal posts and all bets are off  

Sorry, your analysis is just not true.  I think the only way you will be able to realize this is to stitch several images together yourself.  I'd suggest using a 35mm and 70mm. Make a 2 image wide stitch with the 35mm and a 4x4 stitch with the 70mm (very minor cropping allowed to make each have the same FOV).  For bonus points, how will you be able to determine which image was shot with which lens or to put it another way what focal length will you say the stitched image used since the FOV is the same?  

In conclusion, none of your statements are completely true.

Actually they're all true

[filmcapture]

Now I am shocked that some dear colleagues in this forum could not understand perspective even they could have posted hundreds or thousands times. I recall I learned this basic theory of Optics in middle school, unfortunately, I really did not expect so much effort were needed to explain this term.
 [attachment=16464:focus.jpg]

[ThierryH]

That's exactly what I have said to another colleague, even with kinder terms and words, and I have been accused of insulting him: be careful!

 

Thierry

Now I am shocked that some dear colleagues in this forum could not understand perspective even they could have posted hundreds or thousands times. I recall I learned this basic theory of Optics in middle school, unfortunately, I really did not expect so much effort were needed to explain this term.

[Panopeeper]

Now I am shocked that some dear colleagues in this forum could not understand perspective even they could have posted hundreds or thousands times. I recall I learned this basic theory of Optics in middle school

I am shocked that you are mixing up the projection of a single point (or of a line segment) with perspective. Perspective is the relationship between several projected points of objects (this includes lines, etc.).

[filmcapture]

I am shocked that you are mixing up the projection of a single point (or of a line segment) with perspective. Perspective is the relationship between several projected points of objects (this includes lines, etc.).

Yes, go ahead to follow the theory and figure out how "several projected points of objects" are projected, you will find out what "perspective" means.

[filmcapture]

More examples: http://www.phototakers.com/forum/ftopic36480.html

[cmi]

Dear Ray,

I think what you are referring to is the common experience we share when using different focal lenghts. This experience of course stays the same no matter how we explain it in the end. We look at a wideangle image and say: Wow, now thats a nice perspective! And we mean the whole impression, we dont analyze it. In reality this is incorrect formulated, we should say, what a nice perspective and angle of view! Of course nobody does this! What GIVES this view, IS the perspective, is MY perspective, the point I am standing. The other part is my angle of view.

A good way to think about perspective is to think about objects occluding other objects. When you dont move your head, of course everything remains static. If you move your head around, the occlusion of objects, the direction of lines, etc, change. The actual scenery remains as it is, only your position in space changes. Its my perspective, the point where I look from, my standpoint. If I take a steelplate with a square in it and look through it, that doesnt change perspective. Also a lens at different focal lenghts and same position doesnt change anything about the occlusion of objects. Sure, I might see more things around, the same "crop" is smaller or larger, but the only difference is that I see more around. Likewise, If you dont move your head but focus your attention only at a small thing wide away, and then at the whole scene, nothing changes except your conciousness. And so you could view differnet focal lenghts just as optimized projectors for different attention-angles (funny term but only for the sake of the analogy), that is different view angles.

So yes if you are refering to our common sense - we tend to totally neglect our own effect (walking around!) on perspective and only see what different lenses do  - then yes, in common sense, a wideangle gives "a nice perspective", but if we take that apart and really try to understand HOW it works, we need to rethink some of our common phrases, and eventually come to realize that it works somehow different then common sense suggests. And yes, if you where saying different lenses give different viewing experiences (after all, we have them for a reason) if you refer to THAT well there is no dispute that this is correct.

To end here, I read your last posts and tried to understand how you where thinking about it, I hope this makes somehow sense to you.


Christian