4 ms·

An important thing missed here: according to the EXIF data in them, these photos are all taken with an iPhone XS. That means they are taken with a tiny wide ang

An important thing missed here: according to the EXIF data in them, these photos are all taken with an iPhone XS. That means they are taken with a tiny wide angle lens. You'd get a very different shot if you took out a dedicated camera with a physically larger lens and a smaller angle. The overall thrust of this article touches on this when it starts discussing different focal lengths, but the fact that the engineers at Apple made a particular set of choices with their lenses that are a compromise between "what can be achieved with a tiny flat lens measured in millimeters" and "what provides a mostly-acceptable image in any situation", and are very not the choices one might make for specifically photographing a portrait or a landscape or whatever, is really never remarked on.

user-the-name 5y ago
shuntress 5y ago
The article mentions "Correct" perspective several times and seems like it is leading into an explanation of why 50mm is generally accepted as the most natural match to the human eye's perspective. But, it never really gets there.
- Forgeties79 5y ago
  Yeah, it’s sort of a semi-conventional (and somewhat controversial) statement in the film/photo world to say that 50mm lens on a full frame sensor/35mm stock is “what people see.” Some argue it’s closer to 35-40mm. Some argue it’s all moot. Maybe he ultimately just decided not to firmly stick by it haha
- thequux 5y ago
  You can figure out "natural" focal length by taking a printed photograph, holding it at a comfortable viewing distance, and measuring that distance. Now, scale all of those distances down until the photo is the size of your imaging surface (24x36mm for 135 film, better known these days as "full-frame"). Normally, for a 4x5" photo, you'll naturally hold it about 6" from your face, which is pretty close to 150mm. Ergo, the "normal" lens for 4x5 cameras was 150mm. If you scale that down to the "35mm" 24x36 frame, though, you'll discover that that's closer to 40mm. This is not a mistake; IIRC, one of the early 135 camera makers found it easier to make a 50mm lens than a 40mm lens, and it's stuck ever since. It so happens that a normal lens has a FOV of about 1 radian, so the normal focal length is approximately the diagonal of the image surface. A proof of that is left as an exercise for the reader.
  - shuntress 5y ago
    I think that "naturally" and "pretty close" in that method are carrying a lot of weight. It is easy enough to hold up an SLR with a 50mm and see that the size of an object in the viewfinder is essentially equal to what you see with your eye. It's almost like just looking through a piece of flat glass (apart from the slightly shift due to the lens being positioned below the viewfinder.)
  - paulmd 5y ago
    > This is not a mistake; IIRC, one of the early 135 camera makers found it easier to make a 50mm lens than a 40mm lens, and it's stuck ever since. "lens coverage" is the size of the image circle that's projected by a lens. You need (or at least, generally want) an image circle that's big enough to cover your negative or sensor. lens coverage is actually something that is determined by the formula of the lens - a Super Angulon 90/8 has more coverage (will throw a larger image circle) than an Angulon 90/6.3 despite the fact that they are both 90mm lenses. And moreover it is actually something that (if you are looking at it at a formula level) is measured in degrees. You have a triangle that is formed by the nodal point of the lens, the center of the film, and the edge of the image circle, and the lens coverage (in degrees) determines that angle. But, if you take the same lens formula and you make it longer, that leg of the triangle gets bigger too! At infinity, a 90mm lens will need to be 90mm from the film (backfocal distance) while a 75mm would only be 75mm from the film. So a 90mm Super Angulon will have the same angle of coverage, but that means more coverage (in mm) than a 75mm Super Angulon would have, because of the longer focal length. So a 90mm Super Angulon would cover 5x7 film, but a 75mm Super Angulon would only cover 4x5 film. https://www.graflex.org/lenses/lens-spec.html https://www.graflex.org/lenses/lens-spec.html Anyway, point being, 40mm is really tight on a Cooke Triplet-type lens (or really, even Tessars) to cover a 35mm frame of film, especially in the faster configurations with wider apertures. It works, but really you're going to get a lot of softness and maybe even vignetting wide open. But making it 50mm gives you some extra image circle to play with and increases the angle of sharp coverage as well. 50mm is generally close enough that people don't really notice the difference anyway. Some lenses actually went even farther, early SLRs struggled to get enough distance to fit the mirror in, and the way they worked around it was to make the lenses even longer (often 58mm) to give some extra room to fit the mirror. Early in the development of SLRs it was not possible at all to do wide angle lenses because (being only let's say 35mm long) they could not focus to infinity when used with a mirror box that needed 50mm of clearance. In a rangefinder design (like a Biogon), you can have glass elements that get very close to the film but on a SLR that's where the mirror has to go. The fix was the Angénieux retrofocus design, which was a reverse-telephoto design that pushed the nodal point behind the lens, so the actual glass could be farther away, but the nodal point only had to be (eg) 35mm away. This design was directly copied by a couple companies, including Pentax, who sold it for a number of years as the Pentax/Takumar 35/3.5 and the Pentax/Takumar 6x7 75/4.5. https://en.wikipedia.org/wiki/Angénieux_retrofocus https://en.wikipedia.org/wiki/Angénieux_retrofocus A few more fun tidbits: Technically long lenses and telephoto lenses are not the same thing. A telephoto lens has a focal length shorter than its physical length (or, focal length longer than its nodal distance). Generally this is done with a specific arrangement of positive and negative elements called a "telephoto group". In contrast I have a Takumar 500mm f/5 lens that is not a telephoto lens, it's actually 500mm long and has a single doublet group at the very end, so it is like a spyglass that you mount on your camera, not actually a telephoto. The correct terminology for a lens that has a narrow field of view is a "long lens", not "telephoto". The angle of coverage changes for many lenses as you stop them down (however this is not universally true). There is often also a "angle of sharp coverage" where the image gets soft, but is still illuminated towards the edge. On many 35mm lenses this is "cropped" internally using baffles but on large format you get to determine that yourself. Again, stopping down will usually improve the angle of sharp coverage. The "field of view" of a lens is dependent on the coverage and the image size. If you put a 35mm film in a 4x5 image circle projected by a 90mm lens, it works fine, but you're only getting a part of the possible field of view of the lens. And it is the exact same field of view you would get with a plain old 90mm lens from the 35mm camera manufacturer's lineup. So in large format you end up with a situation where that 90mm lens could be "wide on 4x5, superwide on 5x7", because you can pick your sensor size as well. Just like a lens can be "normal on full frame, long on crop sensor" for the smaller sensors. Most of the simpler modern lens types have been known for a long time! For example the 50mm lens on your DSLR is likely a Planar type, the Zeiss Planar was invented in 1896. However, for many of them the materials science had to catch up to really make them practical - early lens design favored fewer elements and cemented groups because every air-glass interface would lose some light and you'd end up with very low contrast. Lens coatings were developed slightly before WW2 and were a key enabler for more advanced lens types, along with the later inventions of more advanced glass compositions (particularly fluorite elements), aspheric surfaces, and computer-aided design. It's not a coincidence that there's a cambrian explosion of lens design in the 50s/60s, that's when the material science really caught up! Anyway many of these things are tangentially observable on 35mm, but of course the intent there is to provide a streamlined experience, all of this is exposed full force in large format. Hopefully this explains some of the ideas behind quirks of lenses that people might have wondered about.
deadbeeves 5y ago
The size of the lens is actually completely irrelevant. Anyone who has done any 3D work can tell you the size of the lens has absolutely no effect on perspective. The mathematical model of a camera in a 3D renderer doesn't have a parameter for lens diameter. There's no way to make an image appear as if it has been taken by a really tiny camera or a really huge camera. At the worst the huge camera will not physically fit in certain spaces. All the size of the lens affects is how much light can come into the camera, which in turn affects the quality of the photo with less than ideal lighting conditions.
- codeflo 5y ago
  Aperture size has an effect on depth of field (the degree to which points that are out of focus are blurred). I think in practical cameras, aperture size isn't constant as you vary camera size, so that makes the scale relevant. But I'm not 100% sure how photography terminology like "stops" maps to the underlying geometry, so take this with a grain of salt.
  - thequux 5y ago
    The "f/" in aperture settings literally stands for "focal length divided by ...". e.g, an f/2 aperture on a 50mm lens has a diameter of 25mm. One aperture stop corresponds to a scaling factor of sqrt(2), because the area of the aperture scales with the square of the diameter, and halving the area results in half the light (and 1/1.414... of the aperture diameter)
    - roelschroeven 5y ago
      More specifically the f/ number relates to the size of the entrance pupil (the optical image of the physical aperture stop, as seen through the front of the lens) rather than the size of the physical aperture. Depending on the construction of the lens and where in the lens the aperture is located, the two can differ quite a lot.
roywiggins 5y ago
All else being equal, switching to a longer lens won't make the relative size of the elements of the image change at all. You can simulate the results just by cropping photos taken with the wide iPhone lens. There will be no difference between the perspective of a cropped iPhone picture and an uncropped picture with a longer lens. Of course such cropping will be likely to throw stuff away from the borders of the image that you'd rather keep, because you composed before you cropped. To avoid losing stuff on the edges, you'd have had to stand further away from your subject, which happens naturally when you're shooting with a longer lens. But you could have stood further away, shot with an iPhone, and then cropped in, and the only difference between that and a "true" zoom lens is that it's probably blurrier because you are working with fewer pixels. (the above is only true for rectilinear lenses, it's the linearity of the perspective that means that zooming == cropping) The article is about how when a human looks at a scene, you'll see quite a wide angle, but when you take a photo that covers a similar angle to what you're perceiving, you end up with an image that seems to exaggerate the sense of depth that you had at the time. This is at least partly because human eyes aren't rectilinear; you can shoot with a non-rectilinear lens but then all the straight lines you expect... aren't. A skilled artist can produce a scene with 1) straight lines 2) a similar field of view, and 3) "natural" depth, as in the Turner painting in the article. To do this you have to depart from linear perspective. You couldn't get Turner's painting just by standing further away; he's doing something rather more complicated.
- MetaWhirledPeas 5y ago
  > All else being equal, switching to a longer lens won't make the relative size of the elements of the image change at all. You can simulate the results just by cropping photos taken with the wide iPhone lens. There will be no difference between the perspective of a cropped iPhone picture and an uncropped picture with a longer lens. Right. I haven't read the article yet, but perspective is about where you are not what kind of lens you are using. A new perspective cannot be simulated by cropping or distorting an image.
wyager 5y ago
Ignoring "non-geometric" effects like depth of field, the only optical parameter that matters here is the visible angle (a function of focal length and crop factor). Parameters like the size of the aperture are massively important in general, but not for determining the relative size of objects within the scene.

user-the-name 5y ago

shuntress 5y ago

The article mentions "Correct" perspective several times and seems like it is leading into an explanation of why 50mm is generally accepted as the most natural match to the human eye's perspective. But, it never really gets there.

Forgeties79 5y ago
Yeah, it’s sort of a semi-conventional (and somewhat controversial) statement in the film/photo world to say that 50mm lens on a full frame sensor/35mm stock is “what people see.” Some argue it’s closer to 35-40mm. Some argue it’s all moot. Maybe he ultimately just decided not to firmly stick by it haha
thequux 5y ago
You can figure out "natural" focal length by taking a printed photograph, holding it at a comfortable viewing distance, and measuring that distance. Now, scale all of those distances down until the photo is the size of your imaging surface (24x36mm for 135 film, better known these days as "full-frame"). Normally, for a 4x5" photo, you'll naturally hold it about 6" from your face, which is pretty close to 150mm. Ergo, the "normal" lens for 4x5 cameras was 150mm. If you scale that down to the "35mm" 24x36 frame, though, you'll discover that that's closer to 40mm. This is not a mistake; IIRC, one of the early 135 camera makers found it easier to make a 50mm lens than a 40mm lens, and it's stuck ever since. It so happens that a normal lens has a FOV of about 1 radian, so the normal focal length is approximately the diagonal of the image surface. A proof of that is left as an exercise for the reader.
- shuntress 5y ago
  I think that "naturally" and "pretty close" in that method are carrying a lot of weight. It is easy enough to hold up an SLR with a 50mm and see that the size of an object in the viewfinder is essentially equal to what you see with your eye. It's almost like just looking through a piece of flat glass (apart from the slightly shift due to the lens being positioned below the viewfinder.)
- paulmd 5y ago
  > This is not a mistake; IIRC, one of the early 135 camera makers found it easier to make a 50mm lens than a 40mm lens, and it's stuck ever since. "lens coverage" is the size of the image circle that's projected by a lens. You need (or at least, generally want) an image circle that's big enough to cover your negative or sensor. lens coverage is actually something that is determined by the formula of the lens - a Super Angulon 90/8 has more coverage (will throw a larger image circle) than an Angulon 90/6.3 despite the fact that they are both 90mm lenses. And moreover it is actually something that (if you are looking at it at a formula level) is measured in degrees. You have a triangle that is formed by the nodal point of the lens, the center of the film, and the edge of the image circle, and the lens coverage (in degrees) determines that angle. But, if you take the same lens formula and you make it longer, that leg of the triangle gets bigger too! At infinity, a 90mm lens will need to be 90mm from the film (backfocal distance) while a 75mm would only be 75mm from the film. So a 90mm Super Angulon will have the same angle of coverage, but that means more coverage (in mm) than a 75mm Super Angulon would have, because of the longer focal length. So a 90mm Super Angulon would cover 5x7 film, but a 75mm Super Angulon would only cover 4x5 film. https://www.graflex.org/lenses/lens-spec.html https://www.graflex.org/lenses/lens-spec.html Anyway, point being, 40mm is really tight on a Cooke Triplet-type lens (or really, even Tessars) to cover a 35mm frame of film, especially in the faster configurations with wider apertures. It works, but really you're going to get a lot of softness and maybe even vignetting wide open. But making it 50mm gives you some extra image circle to play with and increases the angle of sharp coverage as well. 50mm is generally close enough that people don't really notice the difference anyway. Some lenses actually went even farther, early SLRs struggled to get enough distance to fit the mirror in, and the way they worked around it was to make the lenses even longer (often 58mm) to give some extra room to fit the mirror. Early in the development of SLRs it was not possible at all to do wide angle lenses because (being only let's say 35mm long) they could not focus to infinity when used with a mirror box that needed 50mm of clearance. In a rangefinder design (like a Biogon), you can have glass elements that get very close to the film but on a SLR that's where the mirror has to go. The fix was the Angénieux retrofocus design, which was a reverse-telephoto design that pushed the nodal point behind the lens, so the actual glass could be farther away, but the nodal point only had to be (eg) 35mm away. This design was directly copied by a couple companies, including Pentax, who sold it for a number of years as the Pentax/Takumar 35/3.5 and the Pentax/Takumar 6x7 75/4.5. https://en.wikipedia.org/wiki/Angénieux_retrofocus https://en.wikipedia.org/wiki/Angénieux_retrofocus A few more fun tidbits: Technically long lenses and telephoto lenses are not the same thing. A telephoto lens has a focal length shorter than its physical length (or, focal length longer than its nodal distance). Generally this is done with a specific arrangement of positive and negative elements called a "telephoto group". In contrast I have a Takumar 500mm f/5 lens that is not a telephoto lens, it's actually 500mm long and has a single doublet group at the very end, so it is like a spyglass that you mount on your camera, not actually a telephoto. The correct terminology for a lens that has a narrow field of view is a "long lens", not "telephoto". The angle of coverage changes for many lenses as you stop them down (however this is not universally true). There is often also a "angle of sharp coverage" where the image gets soft, but is still illuminated towards the edge. On many 35mm lenses this is "cropped" internally using baffles but on large format you get to determine that yourself. Again, stopping down will usually improve the angle of sharp coverage. The "field of view" of a lens is dependent on the coverage and the image size. If you put a 35mm film in a 4x5 image circle projected by a 90mm lens, it works fine, but you're only getting a part of the possible field of view of the lens. And it is the exact same field of view you would get with a plain old 90mm lens from the 35mm camera manufacturer's lineup. So in large format you end up with a situation where that 90mm lens could be "wide on 4x5, superwide on 5x7", because you can pick your sensor size as well. Just like a lens can be "normal on full frame, long on crop sensor" for the smaller sensors. Most of the simpler modern lens types have been known for a long time! For example the 50mm lens on your DSLR is likely a Planar type, the Zeiss Planar was invented in 1896. However, for many of them the materials science had to catch up to really make them practical - early lens design favored fewer elements and cemented groups because every air-glass interface would lose some light and you'd end up with very low contrast. Lens coatings were developed slightly before WW2 and were a key enabler for more advanced lens types, along with the later inventions of more advanced glass compositions (particularly fluorite elements), aspheric surfaces, and computer-aided design. It's not a coincidence that there's a cambrian explosion of lens design in the 50s/60s, that's when the material science really caught up! Anyway many of these things are tangentially observable on 35mm, but of course the intent there is to provide a streamlined experience, all of this is exposed full force in large format. Hopefully this explains some of the ideas behind quirks of lenses that people might have wondered about.

deadbeeves 5y ago

The size of the lens is actually completely irrelevant. Anyone who has done any 3D work can tell you the size of the lens has absolutely no effect on perspective. The mathematical model of a camera in a 3D renderer doesn't have a parameter for lens diameter. There's no way to make an image appear as if it has been taken by a really tiny camera or a really huge camera. At the worst the huge camera will not physically fit in certain spaces. All the size of the lens affects is how much light can come into the camera, which in turn affects the quality of the photo with less than ideal lighting conditions.

codeflo 5y ago
Aperture size has an effect on depth of field (the degree to which points that are out of focus are blurred). I think in practical cameras, aperture size isn't constant as you vary camera size, so that makes the scale relevant. But I'm not 100% sure how photography terminology like "stops" maps to the underlying geometry, so take this with a grain of salt.
- thequux 5y ago
  The "f/" in aperture settings literally stands for "focal length divided by ...". e.g, an f/2 aperture on a 50mm lens has a diameter of 25mm. One aperture stop corresponds to a scaling factor of sqrt(2), because the area of the aperture scales with the square of the diameter, and halving the area results in half the light (and 1/1.414... of the aperture diameter)
  - roelschroeven 5y ago
    More specifically the f/ number relates to the size of the entrance pupil (the optical image of the physical aperture stop, as seen through the front of the lens) rather than the size of the physical aperture. Depending on the construction of the lens and where in the lens the aperture is located, the two can differ quite a lot.

roywiggins 5y ago

All else being equal, switching to a longer lens won't make the relative size of the elements of the image change at all. You can simulate the results just by cropping photos taken with the wide iPhone lens. There will be no difference between the perspective of a cropped iPhone picture and an uncropped picture with a longer lens. Of course such cropping will be likely to throw stuff away from the borders of the image that you'd rather keep, because you composed before you cropped. To avoid losing stuff on the edges, you'd have had to stand further away from your subject, which happens naturally when you're shooting with a longer lens. But you could have stood further away, shot with an iPhone, and then cropped in, and the only difference between that and a "true" zoom lens is that it's probably blurrier because you are working with fewer pixels. (the above is only true for rectilinear lenses, it's the linearity of the perspective that means that zooming == cropping) The article is about how when a human looks at a scene, you'll see quite a wide angle, but when you take a photo that covers a similar angle to what you're perceiving, you end up with an image that seems to exaggerate the sense of depth that you had at the time. This is at least partly because human eyes aren't rectilinear; you can shoot with a non-rectilinear lens but then all the straight lines you expect... aren't. A skilled artist can produce a scene with 1) straight lines 2) a similar field of view, and 3) "natural" depth, as in the Turner painting in the article. To do this you have to depart from linear perspective. You couldn't get Turner's painting just by standing further away; he's doing something rather more complicated.

MetaWhirledPeas 5y ago
> All else being equal, switching to a longer lens won't make the relative size of the elements of the image change at all. You can simulate the results just by cropping photos taken with the wide iPhone lens. There will be no difference between the perspective of a cropped iPhone picture and an uncropped picture with a longer lens. Right. I haven't read the article yet, but perspective is about where you are not what kind of lens you are using. A new perspective cannot be simulated by cropping or distorting an image.

wyager 5y ago

Ignoring "non-geometric" effects like depth of field, the only optical parameter that matters here is the visible angle (a function of focal length and crop factor). Parameters like the size of the aperture are massively important in general, but not for determining the relative size of objects within the scene.