skibum5 wrote:
to be honest though that is abou the ebst copy i have ever seen of that lens
i rarely ever see samples like that
and the two I tried were already weaker in the corners than that.... on a 20D!
EDIT: then again where is the corner crop? that is an almost center crop and and a semi-outer edge bottom crop. show us the beef!
No, that is a corner crop, 100%, from a 1Ds3.
I have owned 3 copies of this lens and all have been the same. I have no reason to believe my current copy is exceptional. In fact, I decided to downgrade to it from the 16-35 mk 2 for the sake of price/size/weight/40mm. There certainly have been some bad copies, judging by hubsand's 16-9.net experience, but it seems that Canon sharpened up its QC act a couple of years ago.
You mean, any full-frame sensor? Cameras with a crop factor don't tax the lenses quite as much, and you don't have any corner sharpness and vignetting problems. Also 12mp FF cameras are not as demanding as 21mp FF cameras.
RDKirk wrote:
That's true of any sensor.
And the vice versa is true of any lens.
Improve either, and you get more from the other. We have not "maxed out" either yet.
brainiac wrote:
There certainly have been some bad copies, judging by hubsand's 16-9.net experience, but it seems that Canon sharpened up its QC act a couple of years ago.
I've had generally good experiences with my Canon lenses, but some gaps in Canon QC clearly remain; check out this 24L...(now that's just ugly)...
The concept of "taxing' lenses--or sensors--is an erroneous way to understand what is happening.
Think of this for a second: If you handhold either a 24x36mm or a 15x22mm camera at, say, 1/15 second with a 50mm lens of unknown optical quality, you'll likely suffer camera shake, right? The image will be greatly blurred. Maybe some part of that blur is from the unknown quality of the lens. Maybe the lens suffers from horrendous chromatic aberration in the corners...but you can't tell because the motion blur obscures everything.
Okay, so you put the camera on a tripod to eliminate (pretty much) the motion blur and take another picture. Now the entire picture is much sharper with the motion blur removed--sharp enough that you can now clearly see the lens is suffering from chromatic aberration in the corners.
So, does the use of a tripod "tax" the lens? If you use a tripod and reduce motion blur as a factor of the total system blur, does that mean you must now use a better lens? Of course not...even though you can now discern that the remaining blur is mostly lens faults, you still have an overall sharper picture than before. Nobody would suggest not using a tripod because it enables you to see the lens faults more clearly.
The lens and the sensor are each mutual limitations in total system resolution. An improvement of either--all else being equal--will improve the total. Returns diminish drastically, however, as you approach the theoretical maximum of the weaker, and the total system resolution can never actually reach the maximum of the weaker of the two (the curve steepens infinitely).
We aren't nearly at the maximum visible point of visible improvement yet, though. Already have a great deal of experience using the EF lenses with sensors of even higher resolution than 21mp: Kodak Technical Pan film (a thin-emulsion, high accutance black and white film). Tech Pan is considered to be equal to a 35mp 24x36mm sensor...but of course, limited to black and white.
This is hard proof (not debate, not mere evidence, but proof in hardcopy print) that sensor resolution up to at least 35mp can result in visible improvements in resolution.
I loved the H&W film I used, though again in B/W. It was ASA 32 and was a high contrast copy film developed in a one use low contrast developer. What film and when done with my Leica.. . . .
jamato8 wrote:
I loved the H&W film I used, though again in B/W. It was ASA 32 and was a high contrast copy film developed in a one use low contrast developer. What film and when done with my Leica.. . . .
I never had a chance to use H&W, but I certainly heard of it. I never used Adox, either. Digital sensors aren't even there yet in absolute resolution (but we have[/b[ passed the best resolution of color film).
RDKirk wrote:
The concept of "taxing' lenses--or sensors--is an erroneous way to understand what is happening.
Think of this for a second: If you handhold either a 24x36mm or a 15x22mm camera at, say, 1/15 second with a 50mm lens of unknown optical quality, you'll likely suffer camera shake, right? The image will be greatly blurred. Maybe some part of that blur is from the unknown quality of the lens. Maybe the lens suffers from horrendous chromatic aberration in the corners...but you can't tell because the motion blur obscures everything.
Okay, so you put the camera on a tripod to eliminate (pretty much) the motion blur and take another picture. Now the entire picture is much sharper with the motion blur removed--sharp enough that you can now clearly see the lens is suffering from chromatic aberration in the corners.
So, does the use of a tripod "tax" the lens? If you use a tripod and reduce motion blur as a factor of the total system blur, does that mean you must now use a better lens? Of course not...even though you can now discern that the remaining blur is mostly lens faults, you still have an overall sharper picture than before. Nobody would suggest not using a tripod because it enables you to see the lens faults more clearly.
The lens and the sensor are each mutual limitations in total system resolution. An improvement of either--all else being equal--will improve the total. Returns diminish drastically, however, as you approach the theoretical maximum of the weaker, and the total system resolution can never actually reach the maximum of the weaker of the two (the curve steepens infinitely).
We aren't nearly at the maximum visible point of visible improvement yet, though. Already have a great deal of experience using the EF lenses with sensors of even higher resolution than 21mp: Kodak Technical Pan film (a thin-emulsion, high accutance black and white film). Tech Pan is considered to be equal to a 35mp 24x36mm sensor...but of course, limited to black and white.
This is hard proof (not debate, not mere evidence, but proof in hardcopy print) that sensor resolution up to at least 35mp can result in visible improvements in resolution.
Improve either, and you get more from the other. We have not "maxed out" either yet.
I don't quite agree with the way you have put this.
With any given combination of lens and sensor (and aperture) either the lens or the sensor puts a ceiling on image detail. There comes a point where there is no benefit in improving lens performance because the sensor won't record it (e.g. Leica M8), and there can also come a point when there is no point making the sensor denser because it is already above the resolution limit of the lens (16-35 mk2 corners on 5D2).
A cost effective combination has parity of resolution. Alternatively, a cost effective lens is one that mostly outresolves any sensor put behind it during its lifetime. Of course, a lens which limits resolution may not matter for a particular photographic purpose (e.g. low light), so lens choice can be made on many other criteria.
The 5D really is more forgiving of lenses than the 5D2, if you want to see more than 13 megapixels of detail. There exist lens/aperture combinations where the 5D will show its maximum detail, yet the same lens/aperture on the 5D2 will not deliver the camera's maximum detail. That's why the statement "To extract the most out of the 5D II you want to use the best glass that you have" isn't true of any camera. Some cameras won't show you the different levels of detail transmission between two lenses.
The 5D really is more forgiving of lenses than the 5D2, if you want to see more than 13 megapixels of detail. There exist lens/aperture combinations where the 5D will show its maximum detail, yet the same lens/aperture on the 5D2 will not deliver the camera's maximum detail. That's why the statement "To extract the most out of the 5D II you want to use the best glass that you have" isn't true of any camera. Some cameras won't show you the different levels of detail transmission between two lenses.
This is like Aristotle's arguments "proving" that heavier objects always fall faster than lighter objects. His arguments were logically rock solid...but empirically wrong.
I've seen sharper images from EF lenses out of Technical Pan film than any current Canon DSLR sensor can produce, and the images I'm creating from my 5D2 contain more detail than those I obtain from my 5D1 with the same lenses--including my 20mm f/1.8 Sigma.
My primary point is: There is a practical limit out there somewhere, but we aren't there yet. Every time a manufacturer gives us more resolution, people cry "It's too much," and real results prove them wrong.
"I've seen sharper images from EF lenses out of Technical Pan film than any current Canon DSLR sensor can produce"
you're assuming that you know absolute limits of canon dslr technology, which you can't claim to know, unless you were one of the sensor designers (or all of them for that matter).
RDKirk wrote:
This is like Aristotle's arguments "proving" that heavier objects always fall faster than lighter objects. His arguments were logically rock solid...but empirically wrong.
I never mind being compared to Aristotle! ;-)
>I've seen sharper images from EF lenses out of Technical Pan film than any current Canon DSLR sensor can produce, and the images I'm creating from my 5D2 contain more detail than those I obtain from my 5D1 with the same lenses--including my 20mm f/1.8 Sigma.
At what aperture, and in which part of the frame? Certainly not at f1.8 in the corner, unless you have a highly anomalous copy of the 20 f1.8. If you print the corners from that 20mm shot at f1.8 (or uprez to equal numbers of pixels for on screen comparison), you are very unlikely to see more image detail recorded in the 5D2 than the 5D. If you set the aperture to f8 and look at the centre, you are quite likely to see more image detail in the 5D2 than the 5D files. That's sort of my point. Feel free to verify that result.
>My primary point is: There is a practical limit out there somewhere, but we aren't there yet. Every time a manufacturer gives us more resolution, people cry "It's too much," and real results prove them wrong.
But when the manufacturer's own best lens isn't fully using the camera's corner detail at optimal apertures, then we have passed the lens' limit. With the 16-35 mk2 corners on full frame that happens at around 3-5 megapixels. That lens never produces really clean detail in the extreme corners on a 5D1, in my experience (compared to a Zeiss 21). That matters a lot to some landscape shooters, for whom the corners are no less important than any other part of the frame. We really are there now, and we were there with the 5D1, because many people noticed poor performance in the corners from some quite expensive lenses at quite favourable apertures. I would venture to say that a 24-105 doesn't use the 5D2's maximum detail at most apertures, and perhaps all, over most of the frame. There is a difference between resolving something and resolving it cleanly, and between frame centre and frame corner. Many people (fusspots!) want lenses which can cleanly fill all of their camera's pixels with detail in a wide range of situations. If such people want to enjoy the extra detail of 21 Mpixels, they will need to be choosy about lenses.
i wait patiently for 3 months 2 get 5D MKII comes in Tues. wife picks it up Wed ships same day over night early delivery next morning UPS. Some SOB steals camera from UPS high security area. they say sh.. happens. I hate Murphy........
I agree with Richard, the 16-35ii is really a dog when it comes to corner sharpness. I dont have yet a 5Dii, but we can see the poor performance already on a 5D. Here is a sample I shot recently
but at 100% its already clear that this lens cant give us the performance we really need urgently, as we all know, the most precious things in our images are always in the extreme corners http://www.andidietrich.com/elements/fm3/1x.jpg
when we move up to 400% we see the corner performance of this lens is a joke, imagine you make a photograph in the living room and you want to take a photograph of an insect which is sitting on a plant. I am only guessing here but surely you could not see the eyes of that little beast http://www.andidietrich.com/elements/fm3/4x.jpg
