Perspective (was Re: Has ultrawide angle become an overused cliche?)
in [Digital Cameras]
|
From: David J Taylor on 13 Aug 2010 01:50 "Neil Harrington" <nobody(a)homehere.net> wrote in message news:v_6dnZGdNcTKA_nRnZ2dnUVZ_sOdnZ2d(a)giganews.com... [] > I don't know about Greek, and I don't know whether "camera oscura" is > still current in Italian. But it is still called a camera obscura today, > within the last few decades there were still several in operation and my > guess is they still are. They are fascinating devices in the modern > form, and at least one was built in a rotating tower so that it could > scan a good part of the local town That one used a reflex system which > projected the image down onto a large flat table. We have one locally in Edinburgh, for example: http://camera-obscura.co.uk/camera_obscura/camera_obscura.asp Cheers, David
From: David J Taylor on 13 Aug 2010 01:56 "Neil Harrington" <nobody(a)homehere.net> wrote in message news:i_qdnRQSpt9vU_nRnZ2dnUVZ_tWdnZ2d(a)giganews.com... [] > Can you explain that a little more? Or direct me to a source? I'm having > a problem grasping that "reversed perspective" and "telecentric" > business. It looks very interesting but I'm really floundering here. See: http://en.wikipedia.org/wiki/Telecentric_lens For digital use, the idea is that the rays from the lens hit the sensor at a vertical angle of incidence, thus reducing the variations in the image due to any variations of the sensor with angle of incidence. Some manufacturers claim their lenses are "telecentric", but whether they actually work any better in a system I don't know. A pointer to comparative tests would be helpful. Cheers, David
From: David Ruether on 13 Aug 2010 07:15 "David Ruether" <d_ruether(a)thotmail.com> wrote in message news:i41jlu$i18$1(a)ruby.cit.cornell.edu... [...] > YEAH!!! Judge Williams has just > followed up his ruling declaring Prop. 8 in California unconstitutional with > a judgment that his own stay on implementation will be of very limited > duration. A BIG Oooopser above! That was Judge Walker, not Judge Williams! --DR
From: David Ruether on 13 Aug 2010 07:32 "Paul Furman" <paul-@-edgehill.net> wrote in message news:xLydnf4ncKnVEPnRnZ2dnUVZ5rGdnZ2d(a)giganews.com... > David Ruether wrote: >> as I have pointed out earlier, with ever narrower angles of view, ALL the >> several perspective *types* approach each other in appearance. You ["NH"] are confusing the "look" of a photograph with the >> definition of "perspective", >> and leaving out defining "angle of view". For more, see -- >> http://www.donferrario.com/ruether/lens_perspective_types.htm >> http://www.donferrario.com/ruether/lens-angle-of-view-and-perspective.htm >> http://www.donferrario.com/ruether/seeing_and_perspective.htm >> http://www.donferrario.com/ruether/lens_distortion_types.htm >>> This seems like mostly an argument about semantics but here's an example that supports the idea of real differences in wide >>> angle 'perspective'. >> Um, best to correct that to "angle of view"...;-) > OK, "angle of view" is a better term. Semantics solved! :-) > http://www.google.com/#hl=en&source=hp&q=site%3Adonferrario.com+angle+of+view 8^) >>> > Take a pic with a super-wide, then crop the*corner* of the image. Now try to reproduce that perspective with a telephoto >>> > lens. It >>> > could be accomplished with a stitched pano, if the stitching software stretches things but not without software distortion. >>> > >>> > Check the drawing on the bottom of this page, of a distorted sphere in the cropped corner of a wide angle view: >>> > http://www.donferrario.com/ruether/lens-angle-of-view-and-perspective.htm I don't think that could be reproduced with a >>> > telephoto lens. >>> > Put a striped ball on the left side of your desk and try it. >> Thanks for that drawing for my article.;-) It also clearly shows that within >> planes parallel with the sensor plane (with rectangular perspective) that >> there is no "distortion" no mater how wide the angle of view is. > Well, there is distortion in the egg shaped silhouette of the sphere because it's at the far corner of a super-wide. If you swing > the same lens around to center the sphere, it'll have a round outline again. The topographical lines parallel to the image plane > are perfect circles though. More than this is that the axes running through the parallel-plane-cut sections of all the hemispheres are on the vanishing lines/point for the rectangular perspective, so there really is no "distortion", although there may be "unfamiliar visual effects due to viewing 3D objects imaged near the edges of a wide angle of view" since the "stacking" of their sections on the off-axis vanishing lines will cause offsets that change the overall shape of the 3D object... --DR
From: David Ruether on 13 Aug 2010 08:00
"Paul Furman" <paul-@-edgehill.net> wrote in message news:qYOdnQoW6_rQLPnRnZ2dnUVZ5oCdnZ2d(a)giganews.com... > Neil Harrington wrote: >> "Paul Furman"<paul-@-edgehill.net> wrote in message >> news:xLydnf4ncKnVEPnRnZ2dnUVZ5rGdnZ2d(a)giganews.com... >>> David Ruether wrote: [Answering "NH"...] >>>>> This seems like mostly an argument about semantics but here's an example >>>>> that supports the idea of real differences in wide angle 'perspective'. >>>> Um, best to correct that to "angle of view"...;-) >>> OK, "angle of view" is a better term. Semantics solved! :-) >> But that doesn't solve it at all, Paul. Again: if you take a very wide-angle >> shot of the distant horizon, with nothing else in the scene to lend >> perspective, then there is NO perspective no matter how wide the "angle of >> view" is. (Everything is at infinity for all practical purposes.) And >> because there is no perspective, someone looking at the photo has no clue as >> to whether its "angle of view" is wide, normal or narrow. Neil is still confusing angle of view, distance, subject characteristics, image "look", etc. with perspective. But, in his narrow example, it is true that the characteristics of various lens perspective types would likely not show (unless the lens were pointed up or down relative to the horizon, especially with a lens of fisheye type, and/or the subject foreground contained detail, hard to avoid...;-). > Yep. Near the center angle, it doesn't matter much. Using a fisheye lens, you can put a person in the middle of the frame (with a > little distance) and they look perfectly normal. Yep! ;-) >> Only when objects are reasonably close *and* three-dimensional do you get >> perspective. Your drawing of a half-sphere that you linked to is a good >> example of this. The half-sphere's surface at the corner of what would be a >> very wide-angle photo shows the apparent distortion you'd get with a very >> wide-angle lens. But the sections of the half-sphere, being flat, are still >> round because they are two-dimensional. > > Right. And those same sections would appear as ovals once you turn the lens toward them because they are no longer parallel to the > rectilinear projection (perpendicular to the axis of the lens). Yep, again! As with tilting a photo relative to our visual axis...;-) >>> http://www.google.com/#hl=en&source=hp&q=site%3Adonferrario.com+angle+of+view >>>>>> Take a pic with a super-wide, then crop the*corner* of the image. >>>>>> Now try to reproduce that perspective with a telephoto lens. It >>>>>> could be accomplished with a stitched pano, if the stitching software >>>>>> stretches things but not without software distortion. >>>>>> >>>>>> Check the drawing on the bottom of this page, of a distorted sphere >>>>>> in the cropped corner of a wide angle view: Ahem! NOT "distorted" if you are strict about definitions! Although it may look "unfamiliar"...;-) >>>>>> http://www.donferrario.com/ruether/lens-angle-of-view-and-perspective.htm I don't think that could be reproduced with a >>>>>> telephoto lens. >>>>>> Put a striped ball on the left side of your desk and try it. >>>> Thanks [Paul] for that drawing for my article.;-) It also clearly shows that within planes parallel with the sensor plane >>>> (with rectangular perspective) that there is no "distortion" no mater how wide the angle of view is. >>> Well, there is distortion in the egg shaped silhouette of the sphere >>> because it's at the far corner of a super-wide. If you swing the same lens >>> around to center the sphere, it'll have a round outline again. The >>> topographical lines parallel to the image plane are perfect circles >>> though. >> Right. And those differences are the stuff of which perspective is made. > On further thought, perspective has two properties: distance and angle. The word 'viewpoint' might encompass both. I agree that > changing the angle changes the perspective. No. Perspective is NOT a subject characteristic (nor any other characteristic), but it is ONLY a lens/imaging-type characteristic! > I learned to draw isometric in high school drafting class then perspective in college: 1 point, 2 point and even 5 point fisheye > perspective constructed on paper with a 3D grid and vanishing points. When you get into macro beyond 1:1, it's not uncommon for > some lenses or lens combinations to exhibit reversed perspective where the vanishing point is behind you, or a telecentric lens > where there is no vanishing point. For telecentric, you can only capture an image as wide as the front element of the lens. Long > telephoto perspectives come close. I cover some of these in my articles...;-) And, as you point out with the above, these *perspective types* have individual characteristics, and names...;-) --DR |