Crop factor : the consensual explanation ?

[ampguy]

Ok

Ok

I understand what you're saying, and see it in your example. You are saying that one can get the same perspective, basically if you use a perfect lens for the wide, and shoot straight on.

I'm saying that in real life, shooting medium to extreme angles, etc., that the wide is not the same perspective as a longer lens cropped - it's only the same area or dimensions, Also, I'm not interested in photoshopping the images to correct them, I'm talking about real life lenses and no photoshopping.

The aspect ratio of legs to body is because you shoot down from the top and compare things that are not in the center of the frame. That's normal with a wideangle, but when you crop you crop away these portions. You have to crop from the center, just like a digital camera sensor would.

For this kind of comparison, you mustn't move the camera between shots, and you must crop from the center. Here, you would have had to kneel down so that the boy is in the center.

Philipp

 

[yanidel]

So, to reframe the problem and be more pragmatic.

I want to take a portrait that is the 75mm full frame equivalent with a camera with a 1.5 crop factor at a given distance. Framing must be exactly the same. What is the mix of focal and distance that matches what you would get on the 75mm?

- take a 50mm, get closer to the subject until you get the matching frame
- take a 90mm, get further until you get the matching frame.

 

[vieri]

So, to reframe the problem and be more pragmatic.

I want to take a portrait that is the 75mm full frame equivalent with a camera with a 1.5 crop factor at a given distance. Framing must be exactly the same. What is the mix of focal and distance that matches what you would get on the 75mm?

- take a 50mm, get closer to the subject until you get the matching frame
- take a 90mm, get further until you get the matching frame.

With both lens you would be able to match the 75 mm framing at some point, however f stop being equal the results would be very different between the 50 mm and the 90 mm, and neither would match the results from a 75 mm on 24x36 at the same f stop. Compression is a function of the lens, not of the format.

 

[ferider]

So, to reframe the problem and be more pragmatic.

I want to take a portrait that is the 75mm full frame equivalent with a camera with a 1.5 crop factor at a given distance. Framing must be exactly the same. What is the mix of focal and distance that matches what you would get on the 75mm?

- take a 50mm, get closer to the subject until you get the matching frame
- take a 90mm, get further until you get the matching frame.

Hi Yanidel,

the only way to get a similar photo on cropped format from the same focus distance is to use a faster 50. For example, if your 75 is at f2.5 on full format, on APS-C you would have to use a 50/1.4 (50mm and 75mm hera as physical focal length) wide open and from the same distance.

See the attached DOF calculation.

Best,

Roland.

 

[rxmd]

I'm saying that in real life, shooting medium to extreme angles, etc., that the wide is not the same perspective as a longer lens cropped - it's only the same area or dimensions

Sorry to be so insistent, but even the angle doesn't matter - as long as (1) you crop from the center, and as long as (2) the shot from the longer lens is at the same angle. Your example was missing condition (2). If both are met, you can shoot at any absurd angle you like and a crop from the center will look like it's from a longer lens, perspective and all.

To illustrate this, here's three shots from a 15mm lens, each with an 1:1 crop of the center. No matter how oblique the angle and how obvious the ultrawideangle distortion, the crop always looks like it was taken with a normal lens. The ultrawideangle perspective and distortion effects are completely gone:

Note also that there was no photoshopping and perspective correction involved at all. All I've done is cropping the center.

This is a fundamental law of photography and worth trying out. That's why I think every photographer should try medium or large format at some point. It's really instructive to take a 90mm Super Angulon, an extreme wideangle on 5x7", and use it as a portrait tele on 35mm with a telephoto perspective. Or experiment around with cropped wideangle shots on 35mm. It illustrates a lot about how perspective works: how it does not only depend on the focal length of the lens, but also sensor size, print size and cropping. It's counterintuitive at first, but it's worth trying even more so.

Philipp

 

[Roger Hicks]

I know the specs for the F and possibly F2 were 100% when sold, but are they still 100% when later lenses are attached, like say a 24mm or wider on an original F? It couldn't have tested, at the time of manufacture.

Why couldn't they have tested it? The mirror-up 21/4 Nikkor antedates the F, as far as I recall, and because of its construction it gives more of a 'big image' effect than a 24mm reverse-telephoto.

Yes, most ultrawides do give a larger image on most film cameras, thanks to the back of the lens being closer to the film and the image 'creeping under' the film gate. I'd be astonished if the F3 were much different from an F in this regard.

At this point, though, angels and heads of pins come to mind. Alpa used to sell cameras with 24x36mm gates and with (as I recall) 23x34.5mm gates to allow for slide frames...

Cheers,

Roger

 

[gdi]

I'm saying that in real life, shooting medium to extreme angles, etc., that the wide is not the same perspective as a longer lens cropped - it's only the same area or dimensions, .

You should should look up the definition of perspective - or just pay attention to Philipp's excellent explanation and samples...

 

[ampguy]

ok

ok

show me how you would preserve the "perspective" with this wide angle shot with your 50mm:

 

[rxmd]

show me how you would preserve the "perspective" with this wide angle shot with your 50mm:

OK. We are talking about rectilinear lenses all the time, but your shot looks like it was taken with a fisheye, which is the prototype of non-rectilinear lenses. So perspective comparisons don't really make sense. But while we are at it, it would probably look like this:


Again this is just a crop of the center section. It's small because your original picture was small. It looks like a fairly standard close-up shot of a child's nose, taken with a non-rectilinear, equisolid-angle-projection normal lens... Whether or not that is a particularly rewarding subject of photography is another question 😀

Philipp

 

[rxmd]

I know this seems counterintuitive if you are used to the 35mm world and to equating certain image properties with certain focal lengths. Let's try to explain this again by looking at the relationship between perspective and film/sensor size. I think this will be the last time I try to explain this because it won't get any more obvious by repetition. If it's not clear from the explanation, you'll have to try it out. 😀

What you are saying is basically that a wideangle picture has a certain "wideangleness" to its perspective that should be retained even when you crop it - and because 21mm "is" a wideangle, all crops from a 21mm lens should exhibit this "wideangleness" somewhere. Do I understand you correctly?

Now there's nothing intrinsic about 21mm being a wideangle. It's just a focal length, a set of distances and angles in an optical system. The "look" comes from the size of the film behind it!

Like I said, I recently spent some time photographing with a friend who has a Plaubel Peco Profia 13x18 and amongst other lenses a 90/f6.8 Super Angulon. The lens is very difficult to focus on the large screen, but it's a gorgeous ultrawideangle on 5x7", equivalent to the field of view of a 17mm lens on 35mm film. Now taking your argument, pictures with this lens should have this certain "wideangleness", after all the lens is a wideangle. By your argument this "wideangleness" should also appear on a 24x36mm crop of the center section - or if I use it on 35mm film, because there is no difference between cropping a large image or using a smaller film or sensor format. Do I understand correctly? By extension of your argument, various 90mm lenses should show different perspective depending on the kind of film; if it's a 90mm for 5x7", the pictures should be "wideanglish", but if it's a 90mm for 35mm film, the pictures should be "portraitish".

Now tell me again why this should be the case? A focal length is a focal length, it's a fixed optical property. 90mm lenses behave identically! The only difference is that the 5x7" lens casts some extra light outside because of its larger image circle, but that makes no difference at all for the 24x36mm center section; after all the extra light by design never gets to the center.

Now the comparison between 5x7" and 35mm film is of course rather extreme, but the underlying laws of optics are the same, even if you compare between 35mm and an 1.5x crop sensor instead.

Philipp

 

[ampguy]

Hi Philipp

Hi Philipp

Yes, for the most part, you do understand what I'm saying, but here's where I think we disagree. It appears that your definition of rectilinear lens is any lens that is not a fisheye, where increasing frame size never reaches 180 fov.

I believe that a true rectilinear lens is one which renders all straight lines in the subject as straight lines in the image, without any distortion. In the real world, without software correction, the best lenses, including your CV 15 or any commercial ultrawide, even made to order Nikkors, etc. still have distortion and do NOT render all straight lines in the subject as perfectly straight lines in the image, and thus are NOT TRUE rectilinear lenses.

So if you agree with the paragraph above, that some distortion, no matter how small exists in ultrawides, then you should know that he distance between close and distant objects will be exaggerated, sometimes dramatically and always reducing the size of objects in the background. This exaggerated perspective is influenced and can be exacerbated by camera angle. For example, your CV15 images show a lot of this distortion.

So when this image distortion is in the part that is cropped, whether a nose, or a dog next to bicycle wheels, it is a different perspective than with two rectilinear lenses (as defined in my second paragraph.)

To the OP - here is a useful web site I have found that may help you understand your question, and come to your own conclusions:

 

[ampguy]

addl. info.:

addl. info.:

90mm can be a fixed focal length, but the field of view depends on a) the focal length of the lens, 90mm in this case, AND the size of the film (or sensor), so fov is not a fixed characteristic of a lens, and can only be stated if the size of the film or sensor is known, and for a rect. frame, 3 fov are needed, the horiz., vert., and diagonal.

It is very difficult to make a rectilinear lens with more than about 100 deg. of horiz. coverage, and from above, we see that horiz. fov changes with formats, and this is why the amount of distortion can vary between different film sizes or digital sensor sizes.

...
Now tell me again why this should be the case? A focal length is a focal length, it's a fixed optical property. 90mm lenses behave identically! The only difference is that the 5x7" lens casts some extra light outside because of its larger image circle, but that makes no difference at all for the 24x36mm center section; after all the extra light by design never gets to the center.

Now the comparison between 5x7" and 35mm film is of course rather extreme, but the underlying laws of optics are the same, even if you compare between 35mm and an 1.5x crop sensor instead.

Philipp

 

[rxmd]

Hi Ted,

fov is not a fixed characteristic of a lens, and can only be stated if the size of the film or sensor is known, and for a rect. frame, 3 fov are needed, the horiz., vert., and diagonal.

Well technically for a rectangular frame only two are needed, the third follows from Pythagoras' theorem.

It appears that your definition of rectilinear lens is any lens that is not a fisheye, where increasing frame size never reaches 180 fov.

Well not really. I'm referring to the textbook definition of rectilinear projection: a rectilinear lens is a lens that renders straight lines as straight lines. I think we are basically in agreement here.

In the real world, without software correction, the best lenses, including your CV 15 or any commercial ultrawide, even made to order Nikkors, etc. still have distortion and do NOT render all straight lines in the subject as perfectly straight lines in the image, and thus are NOT TRUE rectilinear lenses.

That's true, but in this context it's irrelevant. Firstly, there are some pretty good rectilinear lenses. The CV 15, for example (which isn't mine BTW) is outstanding for a 15mm lens.

You can get lenses made for topographical applications and photogrammetry, such as the aforementioned Topogon wideangle, the Russian Orion-15 or some Biogons. For all practical purposes these lenses are perfectly rectilinear.

So if you agree with the paragraph above, that some distortion, no matter how small exists in ultrawides, then you should know that the distance between close and distant objects will be exaggerated, sometimes dramatically and always reducing the size of objects in the background.

That's also true. But I think you get the reason wrong. The exaggerated perspective is not due to some distortions where the lens fails to live up to 100% rectilinearity. On the contrary, these happen precisely because the lens is built for rectilinear projection. This is the perspective of ultrawideangles; it's how all ultrawideangles render things, and it's not because the lens design is somehow deficient and not up to 100% rectilinearity. For example, if you take a look at the following shot (from Flickr)...

...it shows exactly the behaviour you are speaking of, but it was shot on a Hasselblad SWC with its Biogon 38/f4.5 which is renowned for its outstanding lack of distortion.

For example, your CV15 images show a lot of this distortion.

Well but it's a freaking ultrawideangle, that's how these lenses render things. All of them, even if they're perfectly rectilinear. I haven't seen a percentage value for the CV15, but even Erwin Puts writes that "distortion is surprisingly low".

I think you are confusing two things. One is the distortion that is present in wideangle projection in general, that turns circles into ovals at the edge of the frame. That's what you see in the CV15 image. Note that straight lines are still straight if you look at the cityscape shot! You can rest assured that 99.9% of the lenses used by the RFF crowd are rectilinear.

The other is things like pincushion and barrel distortion, which are basically shortcomings of the lens formula.

Unfortunately the English language has no precise distinction between the two usages of the word "distortion"; however it's the former that gives wideangle shots their unique look, not the latter as you seem to be assuming.

Philipp