Romy's (liquidstone) thoughts on the 7d2...

O-kay, now there's a good bunch of examples! Very cool, thanks Romy!

Do click for the larger samples, they are not suffering from the software image reduction, and if you click again you get the gallery thumbnail page.

CyberDyneSystems wrote in post #17211449
O-kay, now there's a good bunch of examples! Very cool, thanks Romy!

Jake, for my personal birding style, the 7D MII is the ultimate DSLR.... can't wait for the release of the production models. The beta version is already exceeding my expectations AF and IQ-wise.

High praise indeed.
It's the first APS-C I've even noticed in a long time,. the 7D was great, but it never was in a position to sway a 1D owner back at launch.,.. this guy on the other hand looks very well poised to beat the last remaining in production 1D body at some very specific games, birding being one of them.

Around here we get a little fella called the "least tern" they are fast like a swallow, and erratic, and as the name implies, tiny. Very hard to get a shot of. I am imagining your whiskered tern is similar, and am impressed with your shots tracking it!

Thanks for adding more shots and your thoughts on the camera Romy.

I am seeing more and more proff that my pre order is going to make very happy when it finally arrives.

Another thanks for posting the pictures. Noticed an ever so sweet lens attached

Cool.

liquidstone wrote in post #17211444
Just uploaded to my webpage a bunch of burst shots at BIFs. If one wishes to see more detail, just click on the larger 1500x1000 versions.

looks fantastic. Thanks!

CyberDyneSystems wrote in post #17211443
If we are talking about AF performance, we'd want to use a lens that won't slow it down to test the camera, as opposed to the lens. (or any other bottleneck)

That's a very important (perhaps the most important) data point, but it's also useful to see how the camera does with a lens that doesn't have lightning-fast autofocus. A good predictive autofocus system is going to account for the speed of the autofocus mechanism itself, and a test with a slower-focusing lens is needed to test that aspect of the autofocus system.

Even so, I would expect birders and others who depend the most on the autofocus system to be using lenses with relatively fast mechanisms, so it's likely that this test is more relevant than one with a slower-focusing lens.

If you are going to perform a test to wring out the best a camera can do, one does not do that with cheap glass which might compromise the camera's true capabilities.

I'm not quite sure I follow. A good predictive autofocus system is not going to change the characteristics of a slow focus lens. It can only achieve response as fast as that lens is capable of. The AF algorithm that provides for tracking in AI Servo mode is based on initial focusing result and subsequent points to accurately predict movement. The system will only be as fast as its weakest link.

Drat! Romy likes it. So no excuse for me not to upgrade.

Nethawked wrote in post #17212078
I'm not quite sure I follow. A good predictive autofocus system is not going to change the characteristics of a slow focus lens. It can only achieve response as fast as that lens is capable of. The AF algorithm that provides for tracking in AI Servo mode is based on initial focusing result and subsequent points to accurately predict movement. The system will only be as fast as its weakest link.

The system is predictive. It attempts to predict where the subject will be and to ensure that the lens' focus point will be moved to the proper point at the time the subject reaches that location.

Moving the lens' focus point to the proper point takes time. How much time depends on the lens. And that's the point. If the autofocus system doesn't properly account for the amount of time it takes to move the focus mechanism, then it will be behind the curve, and the end result is that it will miss focus.

A good autofocus system thus accounts for (and preferably automatically adapts to) the speed of the autofocus mechanism, and issues commands to the mechanism in sufficient time as to ensure that the focus will be correct at the predicted time.


Clearly, a slower lens will be slower to achieve lock and will thus not be capable of keeping up with more erratically moving subjects. No autofocus system can properly compensate for that. But it at least can maximize the reliability of achieving focus on a moving subject. A proper test of the autofocus mechanism must (in addition to using a fast-focusing lens) include the use of a lens with a slower mechanism precisely because it's the only way to test the predictive capability of the system as described above.

^^ If you will, please point me to similar tests conducted on previous Canon (or other manufacturer) bodies specifically related to predictive capability across a wide range of lens types. As stated before, the predictive nature of phased AF is based on an algorithm which is not likely to change due to the age or performance of the attached lens, but rather the speed at which that lens achieves focus between point A and B.

From my understanding (apparently rudimentary!) lens focus is static - there are no focus points within the lens but rather in the AF system, based on the selected AF point matrix. I'd love to be corrected if wrong!

Nethawked wrote in post #17212755
^^ If you will, please point me to similar tests conducted on previous Canon (or other manufacturer) bodies specifically related to predictive capability across a wide range of lens types. As stated before, the predictive nature of phased AF is based on an algorithm which is not likely to change due to the age or performance of the attached lens, but rather the speed at which that lens achieves focus between point A and B.

The speed at which the lens achieves focus between points is precisely what I'm referring to. The lens in Romy's tests has a very fast autofocus mechanism. How well the camera does with that lens tells us how responsive and quick the camera's autofocus system is, but it does not tell us anything about the quality of the predictive algorithms that would be in play for lenses with slower autofocus mechanisms. You really need both in order to get a complete picture of how good the camera's autofocus system really is.

I wasn't sure what terminology to use to refer to the point at which the focus mechanism in the lens has achieved proper focus on the subject. Apologies for confusion.