What you showed us was a comparison of two different camera (which is fine) with two different lenses (which is not fine).
Sony 70-200G:
"It is fast (although not as fast as in the Zeiss ZA 24-70mm f/2.8), near silent and very accurate. "
Canon 70-200/4L IS:
"The lens features a ring-type USM drive resulting in a near silent and very fast AF."
"fast" vs "very fast" just to highlight the important words here. That's a subjective comment to some degree but obviously yours does also qualify as such.
A while ago we discussed internally how to come to a definitive conclusion regarding the DSLR AF capabilities. Technically you require identical lenses with identically capable lens firmware (which is already a no go) under identical lighting conditions (which means indoor).
The only relatively meaningful lens for a more formal test would be a Sigma 70-200/2.8 HSM II which is available on all platforms.
Regarding the indoor test scene - our discussion ended when considering taking pictures of a Merklin train approaching the camera. :-)) Technically this may be even fine to some degree though. However, we dropped the idea because it isn't really reflecting a typical field scene which would be a point of criticism, of course.
In our opinion such burst shot comparisons are a waste of resources because you can't keep the set of parameters valid enough.
03-18-2011, 01:07 PM
(This post was last modified: 03-18-2011, 01:09 PM by boren.)
Klaus, wouldn't lighting be practically identical if you perform the test at noon on a bright day? You can also put all tested cameras one by the other on a horizontal bar (e.g. Manfrotto 131DDB), and perform the test at the same time and at what is practically the same angle. The subject of course would be the same too.
[quote name='Klaus' timestamp='1300449572' post='6929']
What you showed us was a comparison of two different camera (which is fine) with two different lenses (which is not fine).
Sony 70-200G:
"It is fast (although not as fast as in the Zeiss ZA 24-70mm f/2.8), near silent and very accurate. "
Canon 70-200/4L IS:
"The lens features a ring-type USM drive resulting in a near silent and very fast AF."
"fast" vs "very fast" just to highlight the important words here. That's a subjective comment to some degree but obviously yours does also qualify as such.
A while ago we discussed internally how to come to a definitive conclusion regarding the DSLR AF capabilities. Technically you require identical lenses with identically capable lens firmware (which is already a no go) under identical lighting conditions (which means indoor).
The only relatively meaningful lens for a more formal test would be a Sigma 70-200/2.8 HSM II which is available on all platforms.
Regarding the indoor test scene - our discussion ended when considering taking pictures of a Merklin train approaching the camera. :-)) Technically this may be even fine to some degree though. However, we dropped the idea because it isn't really reflecting a typical field scene which would be a point of criticism, of course.
In our opinion such burst shot comparisons are a waste of resources because you can't keep the set of parameters valid enough.
[/quote]
The Sony 70-200mm f2.8 G is the best the Sony platform has to offer in this area. The 70-200mm f4 L USM is not... both the f4L IS USM and f2.8L IS USM II are noticeably faster.
If it is true, that the Sony 70-200mm f2.8 G SSM is slower (how do we know that? The camera has a big influence on how fast a lens focusses too.. my 450D is slower in focussing than a 7D or an 1D for instance), that then again does not make the A55v (the one in that "test" from IR) very good at tracking, its lens being ones of the things that hold it back.
Even if the 70-200mm f4 L USM is faster, it is still mounted on my 450D, with less light and at a closer distance. And the shots still are better in focus on the whole.
I would not say that my 450D is better than average suited for sports photography, though.
You can not use a 3rd party lens and think you level the playing field, lens wise, either. The firmware of the lens will be different for each mount, and for some mounts it may be more "compatible" than others, like you mention yourself.
If you assess cameras and their tracking ability, in whatever scenario, you will have to pick a good lens for each platform, even if that means that for one platform the lens is better than the other. That is what it is about, of course, testing the camera with a mounted lens.
So, yes, there were differences...
- The distance reported by IR was not the real distance, which made me end up shooting the car closer by. Advantage for the A55v here.
- I shot very late in the day, sun was low and covered. And back lit the scene/cars. Advantage for the A55v/IR test.
- I used a f4 lens, not an f2.8 lens. This means that my camera did not use its f2.8 high precision sensor.
So, still, when not everything was the same (In fact,. the only thing that really was the same was the subject (car) traveling towards the camera at approximately the same speed), my 450D with lens seems to cope better tracking the subject in this particular scenario.
Of course, it may well be that the IR "reviewer" was even less capable than me, and completely ruined the A55v + 70-200mm f2.8 G shoot. And that that is the reason for the not so great results in that not very taxing test.
It might even also be possible that the A55v reviewer from dpreview was using the camera all wrong, making him come up with a wrong conclusion of the AF tracking capability.
[quote name='boren' timestamp='1300453620' post='6932']
Klaus, wouldn't lighting be practically identical if you perform the test at noon on a bright day?
[/quote]
On a cloudless day, maybe. And, to be nitpicking: with the sun at the same angle and strength, which means at least within the same month.
Cloudless days don't happen here too often.
But that would only solve the lighting problem. You'd still need a subject that moves reproducibly.
-- Markus
Editor
opticallimits.com
03-18-2011, 01:22 PM
(This post was last modified: 03-18-2011, 01:25 PM by Brightcolours.)
[quote name='boren' timestamp='1300453620' post='6932']
Klaus, wouldn't lighting be practically identical if you perform the test at noon on a bright day? You can also put all tested cameras one by the other on a horizontal bar (e.g. Manfrotto 131DDB), and perform the test at the same time and at what is practically the same angle. The subject of course would be the same too.
[/quote]
Best would be to not shoot on a bright day, but on an overcast day. This makes the lighting the most uniform and repeatable.
Also, you should perform the test a number of times per camera/lens combination, and middle the results.
You would need a track you can use, and a car which is easily set to a specific speed.
And that is only for a coming towards you-subject. You should also test cameras to track subjects moving irregularly across the frame and track with a very busy background and a smaller subject... The performance of AF tracking is very complex, and so is the testing of it.
[quote name='Brightcolours' timestamp='1300454565' post='6935']
Best would be to not shoot on a bright day, but on an overcast day. This makes the lighting the most uniform and repeatable.
Also, you should perform the test a number of times per camera/lens combination, and middle the results.
[/quote]
Uniform, yes, repeatable, no. The intensity still varies a lot with cloud thickness.
But as you already mention, there's a lot more to test to fully cover AF performance. That's another reason why we don't do it.
-- Markus
Editor
opticallimits.com
[quote name='mst' timestamp='1300455196' post='6936']
Uniform, yes, repeatable, no. The intensity still varies a lot with cloud thickness.
But as you already mention, there's a lot more to test to fully cover AF performance. That's another reason why we don't do it.
-- Markus
[/quote]
I fully understand why you don't test AF tracking performance.
I think it would be very helpful to note focus accuracy without nit picking between different cameras ability to track. There are plenty of cameras that seem unable to focus well (i.e, either too much tolerance for 'focus correct' or sensor mark too broad for precise focusing - which is more important for close up).
Anyways there is a bit of conflict here since it sounds like a formal analysis is very difficult and the review attempts to do a formal analysis; yet accurate focusing is such a key ingredient to any camera. Oddly with all the complaints I found the E-3 focus to be more 'accurate' than the canon 450D. What this statement really means is that I notice more shots being off with the 450D (not consistent front/back focus) than I did with the E-3 and I don't believe this is due to dof issues one would expect with the E-3 smaller sensor.
-
Hum. Oh well I guess the problem is want (good description of camera focus accuracy) versus having a practical solution on how to provide.
There is no controversy. A review here rated the camera 4 for sports and someone who has never touched the camera complained about the rating. Just bias. IR tested A55 and 7D side by side and concluded comparable performance. It's pretty simple. What's left to say?
[quote name='oneguy' timestamp='1300524882' post='6948']
There is no controversy. A review here rated the camera 4 for sports and someone who has never touched the camera complained about the rating. Just bias. IR tested A55 and 7D side by side and concluded comparable performance. It's pretty simple. What's left to say?
[/quote]
One word comes to mind: troll. "an e-mail message or posting on the Internet intended to provoke an indignant response in the reader."
|
