Depth of Field in 3D

[exemplaria]

Had this thought when thinking about PC lenses. Depth of field tends to be talked about in a 2d sense - a DOF range might be expressed as acceptable sharpness starting 50 feet from the camera and ending 100 feet from the camera. But in a real world sense, when converting a 3D world into a 2D photo, is the depth of field actually a 3D segment of 2 concentric spheres? That is, if the image is wide enough or covers enough visual space, will objects on the edge of the image eventually become blurry? From my example above, if I'm photographing a row of people who are lined up perpendicular to the direction of the camera lens, and the closest person to me is 50 feet away, someone who is 87 feet away from that person will be over 100 feet from the camera. Will they be blurry, or not? I realize after playing around with a DOF calculator my numbers may not make sense in a real world application, (as the lens would need a 120 degree range) but work with me here.

Even if I'm correct in my thinking, is this ever a real world problem? Is the radius of the "DOF sphere" equal the focus subject distance, or something else? I've shot with some super-wide lenses wide open (although that was about f4 on those lenses) but didn't seem to notice this, although I can't say I really looked for it. Is this what is happening when lenses are referred to as being soft in the corners?

 

[amolitor]

Nope, it's actually a plane. It's a geometrically translated copy of the film plane. If the film is curved, now..

With a tilt lens the plane of focus is no longer parallel to the plane of the film, but it's still a fairly simply flipped-around-and-enlarged version of the film plane.

 

[exemplaria]

So technically that would be the infinite space between two parallel planes (math major here). I don't know much about optics, is there a layman's way to explain why that is? Because of the curvature of the glass lens?

 

[Big Mike]

I'm not a physics professor, but this is how I understand it.

The plane of focus is parallel to the film plane. So if you have your line of people, if they are all on the plane of focus, they would all be in focus, even though some are actually closer to the camera. I don't believe that focus works on a spherical basis....unless you're talking about re-positioning the camera.

Of course, in a real world scenario, there is another issue, and that is perspective. People who are closer or farther away from the camera will have different perspective distortion.
Also, we need to understand that while the plane of focus may be parallel, the angle of the subject, relative to the axis of the lens will affect how they look in the photo. Obviously, someone right in front of the camera would look different than someone 50 feet to the right of the camera axis.

 

[amolitor]

What now? Are you concerned that, as you approach infinity, the light from the "plane of focus" is approaching the lens essentially from the side?

That is basically true, and indeed the curvature or the lens is in play here. Those rays of light from very very very far out would hit the lens along a path very very close to parallel with the film plane. The curvature and non-zero thickness of the lens mean that there is, in fact, something for that light to strike. The light wanders through the lens, and exits along a similar pat, almost but not quite parallel to the film plane (typically slightly less parallel) and wanders off to to the film plane way way way way way out on the other side of the lens.

This for a single piece lens.

In reality most lenses have multiple pieces and it into a tube, and don't actually pass any substantial amount of light from very far off-axis. The planarity probably falls apart as well, slamming into real-world engineering problems. There's probably some crap like the lens has to be infinitely big, or maybe infinitely small, to actually render an unbounded plane of in-focus stuff onto an unbounded film plane. I dunno about that.

You will note that very wide angle lenses, that actually DO admit light from very far off to the side, are quite bubble-like. The front lens can stick out an inch or more on some truly radical designs, and looks more like a ball of glass stuck in a tube than an actual lens.

 

[exemplaria]

Ahh no concern, just intellectually curious. After playing around with an online calculator, I can also see that it's almost impossible for my concern (if it was real) to manifest itself in the real world, given the relationship between lens focal length (and therefore field of view), subject distance and DOF.

 

[KmH]

All photographic lenses have some degree of 'field curvature'. Lens focal length and lens design has a lot to do with how much field curvature a lens has.