08-11-2010, 03:50 PM
Sorry for the rather short reply, I should have been a bit more clear.
[quote name='genotypewriter' timestamp='1281538488' post='1673']
Not sure how narrow AOV can make things worse. Narrow AOV means the lens is long and therefore the nodal point of the lens is further away from the sensor and doesn't this mean the light hits fairly straight-on (just like in the case with long APS-C lenses being usable on FF than wide APS-C lenses?)[/quote]
Yes, you are right, the larger the AoV, the more vignetting you get towards the edges of the sensor, with one of the reasons being that because the sensitive areas of the site wells are not at the surface of the sensor, but literally in a small well, which means that unless you have a sensor with perfect little lenses over th esites in exactly the right position for a certain AoV, you will miss part of the light. However, I would expect the effect of the cosine rule to have a larger effect with large AoV lenses, IOW, normal optical rules causing the largest part of the effect.
The AoV part also comes into this when light gets reflected of the sensor, or rather of the site well bottom. But that I would expect to cause the infamous fringing effects. towards the sides of a sensor.
Regarding the size of the sensor, well that also has an effect. The larger the sensor, the wider the AoV will be for the same FL, and vignetting is directly, almost exponentially, related to AoV. That's the cosine rule I was referring to above. Whether a lens is designed for APS-C or for FF doesn't matter in this regard, the only dfiference will be the FL number on the lens for a similar AoV.
Well, AoV is in principle independent of aperture, so that isn't true, actually. However, there are some designs lately, where, by means of having a separate image projecting part as part of the design in a lens, the light rays are kept as parallel as possible. Still with a distinct AoV, but much less than one would expect from such a lens. In a way this is a modified retrofocus type of design, and AFAIK, really only used in WA lenses so far, which for slrs have to be of a retrofocus design anyway. It may introduc all kinds of other optical problems, however, such as higher barrel distortion levels.
The reason for vignetting at large apertures, apart from the effects of the AoV, is twofold, namely that a barrel doesn't have infinite width, and there also is the effect of light rays refracting differently towards the edges of a lens than they do in the centre, although the first effect is the strongest.
The cats eyes effect in specular highlights, f.e., is actually caused by mechanical vignetting of the lens barrel, and is much more common with large aperture complex lenses than with small aperture simpler lenses (which tend to be a lot shorter in build). That is a very rough generalization, I know, but provided a manufacturer doesn't create an absolute minimalistic design but overdesigns the glass and barrel a little, you should have no more vignetting than optical theory allows for.
There are a few lenses that actually do better in this regard, the new TS-Es come to mind: they do better at least for the unshifted or untilted part of an image ... <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. But then, they aren't very fast lenses either.
Well, the lack of distortion is IMO the only redeeming feature of that lens, for me anyway. I tested two for half a day, took the best one home, and was horrified by the vignetting in real life shots (at any aperture and FL), the CA, the behaviour with regard to flare and ghosting, and the blur or rather smearing towards the edges and in the corners. The only good part in the images was about a 1/3 crop, ok, maybe a 1/2 crop. And that was the better lens of the two. A third one I tried wasn't any better either.
I exchanged it for a Nikkor 14-24, which is very good, but a little awkward to work with on a Canon body, and later on replaced that with a TS-E 17L, which proved to be even better again and is quite easy to work with. Ok, the Nikkor with G-EOS adapter is twice the price of the Sigma, and the TS-E almost 3X the price, but I guess that is what it costs to get a really good lens in that focal range. So, granted, for the money the Sigma is still very good, as there isn't anything else in its price class, or when you need 12 mm and can't do a two-image shift panorama, but personally, I couldn't live with it. Fortunately, I am fine with 17 mm and the occasional stitched shot <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.
Kind regards, Wim
[quote name='genotypewriter' timestamp='1281538488' post='1673']
Not sure how narrow AOV can make things worse. Narrow AOV means the lens is long and therefore the nodal point of the lens is further away from the sensor and doesn't this mean the light hits fairly straight-on (just like in the case with long APS-C lenses being usable on FF than wide APS-C lenses?)[/quote]
Yes, you are right, the larger the AoV, the more vignetting you get towards the edges of the sensor, with one of the reasons being that because the sensitive areas of the site wells are not at the surface of the sensor, but literally in a small well, which means that unless you have a sensor with perfect little lenses over th esites in exactly the right position for a certain AoV, you will miss part of the light. However, I would expect the effect of the cosine rule to have a larger effect with large AoV lenses, IOW, normal optical rules causing the largest part of the effect.
The AoV part also comes into this when light gets reflected of the sensor, or rather of the site well bottom. But that I would expect to cause the infamous fringing effects. towards the sides of a sensor.
Regarding the size of the sensor, well that also has an effect. The larger the sensor, the wider the AoV will be for the same FL, and vignetting is directly, almost exponentially, related to AoV. That's the cosine rule I was referring to above. Whether a lens is designed for APS-C or for FF doesn't matter in this regard, the only dfiference will be the FL number on the lens for a similar AoV.
Quote:Similarly for the aperture, I think a large aperture lens would have less issues than a small aperture lens because the light doesn't have to bend as much? The only problem would be mechanical vignetting from the rear because the exit pupil won't be a perfect circle when viewed from the far corners.
Well, AoV is in principle independent of aperture, so that isn't true, actually. However, there are some designs lately, where, by means of having a separate image projecting part as part of the design in a lens, the light rays are kept as parallel as possible. Still with a distinct AoV, but much less than one would expect from such a lens. In a way this is a modified retrofocus type of design, and AFAIK, really only used in WA lenses so far, which for slrs have to be of a retrofocus design anyway. It may introduc all kinds of other optical problems, however, such as higher barrel distortion levels.
The reason for vignetting at large apertures, apart from the effects of the AoV, is twofold, namely that a barrel doesn't have infinite width, and there also is the effect of light rays refracting differently towards the edges of a lens than they do in the centre, although the first effect is the strongest.
The cats eyes effect in specular highlights, f.e., is actually caused by mechanical vignetting of the lens barrel, and is much more common with large aperture complex lenses than with small aperture simpler lenses (which tend to be a lot shorter in build). That is a very rough generalization, I know, but provided a manufacturer doesn't create an absolute minimalistic design but overdesigns the glass and barrel a little, you should have no more vignetting than optical theory allows for.
There are a few lenses that actually do better in this regard, the new TS-Es come to mind: they do better at least for the unshifted or untilted part of an image ... <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. But then, they aren't very fast lenses either.Quote:Actually, I've seen exceptionally good performance from that lens... unfortunately the zoom misalignments and sample variations put people off too much. Otherwise, its distortion free and the CA is very very low.
Well, the lack of distortion is IMO the only redeeming feature of that lens, for me anyway. I tested two for half a day, took the best one home, and was horrified by the vignetting in real life shots (at any aperture and FL), the CA, the behaviour with regard to flare and ghosting, and the blur or rather smearing towards the edges and in the corners. The only good part in the images was about a 1/3 crop, ok, maybe a 1/2 crop. And that was the better lens of the two. A third one I tried wasn't any better either.
I exchanged it for a Nikkor 14-24, which is very good, but a little awkward to work with on a Canon body, and later on replaced that with a TS-E 17L, which proved to be even better again and is quite easy to work with. Ok, the Nikkor with G-EOS adapter is twice the price of the Sigma, and the TS-E almost 3X the price, but I guess that is what it costs to get a really good lens in that focal range. So, granted, for the money the Sigma is still very good, as there isn't anything else in its price class, or when you need 12 mm and can't do a two-image shift panorama, but personally, I couldn't live with it. Fortunately, I am fine with 17 mm and the occasional stitched shot <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />.Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....