[rparmar]

I'd put it this way: there is no doubt that, as the measurement of the diagonal of a frame of 35mm film, 43mm makes sense as a definition of *something*. And maybe it even makes sense to call that thing "normal". However, whether or not it this definition of "normal" has any of the other properties sometimes claimed of it - "closest to the FOV of our eye", "most natural perspective as viewed through the viewfinder", "yields most natural perspective when printed at typical sizes and viewed at typical distances", etc - that's an entirely different matter.

And that's the part I substantiate empirically. If a lens is close to what we call "normal" then the view of the finished picture (forget viewfinder) appears natural, without undue perspective distortion or narrowing of the field (which appears cramped).

Certainly this has nothing to do with the total FOV of the eye itself, which is much greater. It may be more to do with our "binocular field". To quote from myself some time back:

Our stereoscopic field of view is 100°; including peripheral vision it is 120°. Outside this we see with one eye or the other for a complete field of view up to 180°. The binocular field, within which both eyes observe an object simultaneously, has been defined as between 50 and 60°. That's a closer approximation to what looks "natural".

Many great photographers preferred the unmediated look of 35mm on full-frame. Check out the history of Leica for examples. This focal length produces a FOV of 53°, which fits with the binocular field theory.

[Marc Sabatella]

When you crop the photo, it's the same thing as using a smaller film/sensor size, which changes the perspective just like changing focal lengths does.

You'll have to clarify your usage of the word "perspective" here, then. The usual definition that has been around for several hundred years most definitely is *not* affected by any such thing. The relative positions and sizes of objects don't change, the angles formed by converging parallel lines don't change, the location of the vanishing points don't change, etc.

Angle of view, and therefore perspective, is based entirely on the ratios of focal length to sensor size and viewing distance to print size. When these two ratios match (when the angle of view of the capture matches the angle of view of the presentation), then the perspective looks normal.

You are definitely talking now about something other than perspective. You are referring to the field of view of the print itself and comparing it compares with the field of view of the scene represented. This is a legitimate and interesting comparison to make, but it isn't perspective.

When you make the ratio of focal length to film/sensor size higher (assuming the other ratio is constant), either by using a longer focal length or a smaller sensor (or cropping the photo and blowing it up, which is using a fraction of the sensor), then you make everything in the photo look the same percentage larger, which skews the geometry and changes the perspective, creating compression distortion.

This is quite simply 100% false. I absolutely guarantee you there is no difference whatsoever in the geometry within a picture shot from a given location with one focal length versus another, any more than cropping a picture with a pair of scissors changes its geometry. A square remains square, a rectanlge with dimensions 3:2 remains a rectangle with dimensions 3:2, a right triangle remains a right triangle, etc. Feel free to do the test if you doubt it. But centuries of physics and art instruction are not wrong about this. The above is a myth, pure and simple.

The only thing that changes is the field of view spanned by the print, and indeed, if you print both pictures at the same size and view them from the same distance, one might appear to match the natural field of view of the scene itself better than the other. Again, that's a legitimate and interesting comparison, but it isn't perspective.

[Class A]

55mm focal length (physical) will provide you with the perspective of human vision at any distance or field of view.

As others have already said perspective -- as captured on the sensor/film -- solely depends on subject distance. Your magnification (and also the effective perspective) also depend on viewing magnification (either through the viewfinder or by viewing distance to a print).

A "normal" length lens is a lens that captures photos that when viewed from the "normal viewing distance" give the same perspective as human vision. It changes with the size of the film/sensor.

Exactly. For 35mm that's ~43mm and for APS-C ~28mm. BTW, Sigma makes a 28mm f/1.8 MACRO that I think is great.

You'll have to clarify your usage of the word "perspective" here, then. The usual definition that has been around for several hundred years most definitely is *not* affected by any such thing. The relative positions and sizes of objects don't change, the angles formed by converging parallel lines don't change, the location of the vanishing points don't change, etc.

They don't change on the sensor/film. But you need to consider the viewing context. Imagine a certain scene with certain angles formed by converging parallel lines. Let's say you are looking at a print of this scene from the perfect distance, i.e., the angles look completely natural. If you increase the viewing distance, the angles of the print don't change, but in order to retain the same effective perspective they should. The angles will now look exaggerated. If you look at the print too closely, the angles will look not acute enough for the expected change in perspective.

A print, taken with a "normal" focal length (diagonal of the sensor/film), viewed from the "normal" viewing distance (diagonal of the print) will give the impression as if that frame had been cut out of reality and placed in front of you. Change the focal length or the viewing distance and that quality disappears, even though the perspective as captured on the print obviously doesn't change and only depends on camera to subject distance.

Again, that's a legitimate and interesting comparison, but it isn't perspective.

Perhaps it is more precise to call it effective perspective.

[Wheatfield]

Point being, it's all about angle of view, not focal length. In whatever sense 50mm was "normal" or "natural angle of view" for 35mm film, it is *not* that for APS-C - 33mm would be the equivalent focal length.

Gosh....You think???

FWIW, I do *kind of* agree with the assessment of this field of view as being somewhat "natural", although in kind of an artificial way. That is, if I *don't* ignore my peripheral vision, I can very clearly see *far* more than this.

The concept is, though, that people don't pay a lot of attention to their peripheral vision. The reptile part of our brain does, but only to the extent it is determining movement, not detail.
If you can discern much more than movement with your peripheral vision, you have probably moved your eyes slightly.

Like, my 18-55 on APS-C doesn't go wide enough capture everything I can see. at the other extreme, if I look at a scene and think about what part of it am I most focused on, something longer than the sort of focal length we are talking about makes most sense to me. I think the 40mm nails it pretty well most of the time on APS-C. But if I think about how much of the scene I am really *aware* of - the part I am interested in plus the surrounding area I can't help but notice - then somewhere in the 28-35mm range on APS-C seems to capture this best. And yes, I have actually sat down and thought about this in a few different contexts, looking at a scene, forming mental boundaries to represent what I can see, what I am most aware of, and what I am most focused on, then putting various lenses on to see which match those best.

Do you find it as strange as I that Pentax, after five years of making APS-C cameras, has not released a normal lens for the format?
The 31 fits the focal length, but is ridiculously expensive, and isn't really designed for the digital format.

And again, it's not that the focal length does this - all lenses have the same perspective. It's that the *field of view* of an 85mm lens on 35mm film is such that it encourages you to stand at a *position* relative to your subject that makes the above true. Standing from the same place, you could shoot with any rectilinear lens, crop accordingly, and get precisely the same degree of "flattening".

The department of redundancy department just called asking for you....

[jsherman999]

........

The department of redundancy department just called asking for you....

It's OK if you missed it, they always call twice.


.

[Marc Sabatella]

Imagine a certain scene with certain angles formed by converging parallel lines. Let's say you are looking at a print of this scene from the perfect distance, i.e., the angles look completely natural. If you increase the viewing distance, the angles of the print don't change, but in order to retain the same effective perspective they should. The angles will now look exaggerated.
If you look at the print too closely, the angles will look not acute enough for the expected change in perspective.

I hear what you're saying, but again, now we're talking about something that is a function of print size and viewing distance - not simply of focal length, or even of the focal length / sensor size ratio. It's true the print won't look "natural" at this distance, but it would be obviously be wrong to say the perspective of the image changed just because you moved. What's the perspective of the image if there are two people at different places in the room? If there is no one in the room?

If a tree falls in the forest and no one is around to see it, does the perspective change?

A print, taken with a "normal" focal length (diagonal of the sensor/film), viewed from the "normal" viewing distance (diagonal of the print) will give the impression as if that frame had been cut out of reality and placed in front of you.

This much has never been in question. One could quibble what constitutes "normal" viewing distance for a given print size and how useful this is as an actual definition of "normal" focal length, but there is no denying that for a print size and viewing distance, there is one and only one field of view that produces the effect you are talking about, and this field of view is a function of both focal length and sensor size.

Change the focal length or the viewing distance and that quality disappears

True. I'm just pointing out that neither of these changes the intrinsic perspective of the image. This does put my point into pretty good focus: changing focal length doesn't change the actual perspective of the image itself any more than walking toward or away from a print of that image does. Both have a similar effect on the quality you are talking about; both have no effect on the actual perspective of the image.

Perhaps it is more precise to call it effective perspective.

Actually, I wouldn't be surprised if there weren't already a term for it, but I don't know what it is.

[Marc Sabatella]

Do you find it as strange as I that Pentax, after five years of making APS-C cameras, has not released a normal lens for the format?

I do. Although the DA35 isn't too far from being a 50mm equivalent, and the upcoming DA*30 will be closer to the 43mm equivalent.

The department of redundancy department just called asking for you....

I'm afraid I don't understand what you are seeing as redundant. I thought my point was quite straightforward: the thing that changes persepctive is position, and the only reason people associate particular focal lengths with particular perspective effects is that different focal lengths encourage one to shoot at different distance. I think most of us are on the same page regarding this fact, but considering how often someone tries claiming that, for instance, a 50mm lens on APS-C has the FOV of a 75mm lens on FF but still has the perspective of a 50mm lens, I think it bears repeating.

[Class A]

now we're talking about something that is a function of print size and viewing distance - not simply of focal length, or even of the focal length / sensor size ratio.

Yes, but one cannot talk meaningfully about a "normal" focal length without going beyond the parameters you mention above.

One can turn this into one's benefit and just use a different viewing distance to turn whatever lens into a normal lens. Some of these viewing distances will be pretty extreme though.

What's the perspective of the image if there are two people at different places in the room?

If you are talking about "effective perspective" (or whatever the correct term for it is) then this is an ill-formed question, because there will be two such perspectives. We agree that the image perspective as such doesn't change (and I'd say it even exists if no one is looking, but this takes us somewhere else ).

One could quibble what constitutes "normal" viewing distance for a given print size

As CFWhitman said, one typically assumes the diagonal of the print.

As you mentioned works of art, it would be interesting to know whether the "diagonal of the image" rule also works for paintings. Surely, there is an intended and optimal viewing distance (range) for paintings.

[axl]

I have both, DA40 and F50 f1.7.
While 50 is faster and better indoors with more space to move around, I generaly prefer FOV from DA40. Which btw, is still very good indoors. So 50 really comes into play under specific circumstances only: portraits, fast moving not too close not too far subjects, extreme low light, moving subjects in lowlight. Anyway, I like them both, and now I'm out for 28-30-31 (I really wisht tha last one ) fast lens to bracket "normal FOW" from wider end, and sit nicely between my 40ltd and Sig 10-20.
BR

[Class A]

I'm just pointing out that neither of these changes the intrinsic perspective of the image.

I thought about this again and I think my "effective perspective" is not a good choice of words. After all, perspective really refers to the relationship between different parts of a scene. And this is viewing distance independent.

I guess there is just an appropriate viewing distance for a given perspective captured by an image and a so called "normal" focal length just ensures that the appropriate viewing distance is convenient / practical.

I have both, DA40 and F50 f1.7.

Lucky you.

While 50 is faster and better indoors with more space to move around, I generaly prefer FOV from DA40.

Same here. The 28mm I'm using definitely more corresponds to what I'm seeing than a 50. With the later I always need to back up to frame the way I saw something, whereas with the 28mm it is pretty much 1:1 and if at all moving in is often easier than moving back (e.g., indoors).

So 50 really comes into play under specific circumstances only: portraits, fast moving not too close not too far subjects, extreme low light, moving subjects in lowlight.

I'd love to get a F/FA 1.4/1.7, mainly because of the reputation of the lenses. But then I wonder again, whether I'd really could put them to good use. 50mm is just a bit too short for portraits, AFAIC. With a 58mm I'd be much less reserved.

I guess the right lens for me is the Nokton (if it weren't for the price tag and the lack of AF) but then I still wouldn't have one of the legendaries. Its a tough life, if one is still occupied with equipment worries like these instead of shooting and pondering about the next composition. I know photography should be about making images instead of tormenting oneself over equipment questions, but I just find enjoyment in both kind of activities.

[m8o]

I'm late to this discussion, I know. All points have been made well, deliberated, some going off into the weeds, some concise... but again, the important points have all been made. With that said, I, myself, having gone through this same learning curve not too long ago, still didn't fully get it until some annotated examples. These two pages served to provide me the examples that finally solidified the concept in my mind (aka picture's worth a 1000 words).

1) Idea & effect of lens perspective in relation to position vs. crop to keep the primary subject the same size: Perspective: Optical: Glossary: Learn: Digital Photography Review
2) Crop was used above (the red square in the 1st image making the 2nd image), but here's another page on it a bit more (it's just a smaller window on the image compared to if you used film): Focal Length Multiplier: Optical: Glossary: Learn: Digital Photography Review

Hope it does the same for any that need it that it did for me.

[m8o]

This is quite simply 100% false. I absolutely guarantee you there is no difference whatsoever in the geometry within a picture shot from a given location with one focal length versus another, any more than cropping a picture with a pair of scissors changes its geometry.

I thought I was going to disagree with this assertion, but then realized I'd read wrong (even after reading it thrice! <grin>) Thus in support, this assertion is proved out by comparing images A thru C in the 1st link in my previous post.

[CFWhitman]

You'll have to clarify your usage of the word "perspective" here, then. The usual definition that has been around for several hundred years most definitely is *not* affected by any such thing. The relative positions and sizes of objects don't change, the angles formed by converging parallel lines don't change, the location of the vanishing points don't change, etc.

The relative positions of things appear to change because the relative sizes of their images should change but don't. Perspective is all about appearance. When you have two objects of the same size, and one is further away, it appears smaller. It isn't smaller; it only appears that way. That is perspective. When it doesn't appear enough smaller to account for the actual distance between the objects, then perspective is distorted. The angles formed by converging parallel lines don't change, but distances of points along those lines do change, and not the same way as would be expected with normal perspective.

You are definitely talking now about something other than perspective. You are referring to the field of view of the print itself and comparing it compares with the field of view of the scene represented. This is a legitimate and interesting comparison to make, but it isn't perspective.

This is quite simply 100% false. I absolutely guarantee you there is no difference whatsoever in the geometry within a picture shot from a given location with one focal length versus another, any more than cropping a picture with a pair of scissors changes its geometry. A square remains square, a rectanlge with dimensions 3:2 remains a rectangle with dimensions 3:2, a right triangle remains a right triangle, etc. Feel free to do the test if you doubt it. But centuries of physics and art instruction are not wrong about this. The above is a myth, pure and simple.

It doesn't skew the geometry from what it is with the other lens; it skews it from what it should be if it were actually closer. This distinction makes the statement 100% true, and it should have been clear by the context. Perspective changes in a certain way with distance. This is normal. Perspective changes in an entirely different way with enlargement or reduction. This is perspective distortion. There is no myth. (Edit: Incidentally, since point of view is part of perspective, the geometry visible on the print is only half the story. The geometry to the point of view is the other half. According to the concept of normal or ordinary viewing distance, doing this most definitely does skew the geometry. Of course, the counter argument would be that the point of view of correct perspective is the one that matters, and it just changes where that is.)

The only thing that changes is the field of view spanned by the print, and indeed, if you print both pictures at the same size and view them from the same distance, one might appear to match the natural field of view of the scene itself better than the other. Again, that's a legitimate and interesting comparison, but it isn't perspective.

Perspective is about perception. If you change the perception, you change the perspective. So yes, it is perspective. Perspective is about how the apparent relative sizes of things relate to distance. If you change the apparent distance by changing the apparent sizes of things without changing their apparent relative sizes, you change the perspective.

Looking at it the way that you do, you could say that the apparent point of view changes, while the perspective does not, and that is what causes the distortion (distance is still not what causes perspective distortion). However, the definition of perspective is not nearly as specifically and exactly what you are saying as you seem to suppose.

[CFWhitman]

This is why this is a bad way to explain perspective distortion.

From the article at DPReview:

Images C and D show that a tele compresses perspective (makes subjects look closer to one another), while a wide angle exaggerates perspective (makes subjects look more separated) compared to the "normal" way we see things with the naked eye. As mentioned earlier, this change in perspective is a direct consequence of the change in subject distance and thus only an indirect consequence of the change in focal length.

This is not accurate (it is at the very least misleading). If you change lenses without changing distance, perspective will still be compressed or extended with different focal lengths. While if you change distance without changing lenses, perspective will not be compressed or extended. Therefore, the change in perspective (the compression or extension) is not a direct consequence of the change in subject distance. It is a direct consequence of the change in angle of view, and an indirect consequence of the change in focal length (indirect because it can be compensated for by changing the angle of view in another way). It is not affected by the change in distance at all (the change in distance helps to make the effect more obvious, but it does not alter it).

The best argument you could make for this being accurate is that compression or extension (or 'exaggeration,' as they say in the article) is not a change in perspective, but an illusion which is the result of perspective not changing while the apparent point of view does. To me this seems like it is standing the definition of perspective on its head, but you could make the argument.

To put it another way, if you have the definition of perspective that Marc is using, but you still believe that perspective can be changed by being compressed or extended, then you are contradicting yourself.

[Sean Nelson]

If you change lenses without changing distance, perspective will still be compressed or extended with different focal lengths. While if you change distance without changing lenses, perspective will not be compressed or extended.

It seams you are using a different meaning for the word "perspective" than others.

This is the effect of moving the camera:




This is the effect of zooming (ie, changing focal lengths) without moving the camera:




When the TV screen gets larger faster than the background it's reflecting, that's a change in perspective.

But when you just zoom in, the perspective does not change.