garyknrd said:9VIII said:jrista said:The wide aperture affects AF speed and AF point precision and capability. Remember, particularly in the 61pt AF system, there are f/2.8 double cross type points, f/4 cross type points, f/5.6 line points, and the center expansion f/8 points. With an f/4 lens, you ALWAYS AF at f/4, no matter what you stop down to for actual shooting. The extra stop of light allows the AF system to operate more quickly and more accurately. When f/4 AF points are used, they tend to be more precise than f/5.6 points, which need larger pixels in order to sense as well as f/4 pixels.
The point of an f/4 lens isn't that you always shoot wide open (although in the evening, it isn't uncommon...I tend to be around f/8 aperture for shooting during daytime, and f/4-5.6 for shooting around sunset, for wildlife.) It's that you ALWAYS AF wide open (by design.) And yes, with an f/4 lens, when you slap on a 1.4x TC, you still AF at f/5.6, which is still better than AF at f/8, no question.
Jrista, thank you.
I actually completely forgot about the difference between f4 AF points and f5.6 ones. The high precision f2.8 AF points are referenced so much I guess I clumped all the others into the same group. (After reading multiple articles detailing all the AF points, and watching the entire B&H Canon AF seminar [ ww.youtube.com/watch?v=iAx86nblZ2g ][great video BTW], you would think that someone would remember something like that. I guess a guy can only fill his head with so much.)
That changes my perception of the TCs quite a bit. I've been assuming that you get "worse AF" when using a TC because of a combination of optical performance and some kind of interference in the circuitry, if on the other hand it's actually just due to the change in the type of AF points used, then the 600f4+1.4xTC is actually going to AF exactly the same as the 800f5.6. I was assuming the latter would have an advantage.
Indeed that does make a very good case for the 600f4, and makes the 300f2.8 more appealing now that I know it's not some mystical interference from the teleconverter making AF worse at 600mm.
IMO a teleconverter, no mater how good it is ( and how sharp the lens is ) degrades the image.
I don't think it is quite as simple as that. Yes, adding more optical elements has an impact, however those additional optical elements also enlarge the subject. Optical elements tend to be pixel-level issues...the two primary ones are increased distortion (which primarily affects the corners), and increased CA. CA, while it usually shows up and looks bad in ISO 12280 test charts, is a PIXEL-LEVEL issue, and those ISO 12280 test charts are shot at consistent framing, so the benefit of increased subject size is LOST.
In a real-world situation, the theory here does not actually demonstrate actuality all that well. The subject increases by a factor of (840/600)^2, or 1.4x squared, two fold. In the case of a 2x teleconverter, the subject increases by a factor of four. That puts a LOT more pixels onto the subject. They may be slightly softer pixels with a little more CA...but it is still FOUR TIMES the pixels on subject. Assuming a normalized comparison scenario, one could crop the 600mm image and downsample the 840mm image to the same cropped dimensions, and the 840mm image will always win in terms of IQ. CA can be removed easily (if it even shows up as a problem.)
garyknrd said:And that also has an effect on AF accuracy and speed. As well as the loss of 1 or two stops of light.
I would love to see a auto-focus test using the 600 II with a 1.4 against the bare 800 F/5.6 lens. With the same camera it is my guess! The 800 would win. Not only being faster ( though very slightly ) but more important more accurate. Looking at the MTF charts of the two lenses it would be a very close race for sure.
Yeah, I would agree, a AF at a lenses native maximum aperture is probably going to perform a bit better than AF with a TC attached. I think there is a bit of a myth about Canon TCs intentionally slowing down AF, but only in comparison to the likes of third party TCs like Kenko. The problem with a Kenko TC is they effectively cheat. They trick the body into thinking the attached lens has a different maximum aperture. For example, with my 600+1.4x Kenko, the camera thinks the maximum aperture is still f/4. That is generally what leads to the sometimes funky behavior of AF with a third-party TC, it attempts to utilize AF points that MAY not have enough light to operate properly. You would get roughly the same kind of quirky behavior if you pin-taped a Canon 14x TC. Canon TCs report everything properly, so the AF system doesn't even try to use f/4 AF points when the relative aperture is actually f/5.6.
garyknrd said:Look at this link 1.4 tc III on the 200 f/2.8 lens. http://slrgear.com/reviews/showproduct.php/product/1367/cat/62
And here for the 2.0tc III: http://slrgear.com/reviews/showproduct.php/product/1366/cat/62
Looking at that data. Pretty much clears things up for me when using a t.c. I wish these guys would test the 600 II and 500 II using T.C.'s.
But even if that is true. I would much rather have a 600 II with a teleconverter any day... ;D
Again, as I mentioned before...these tests are done for identical framing. Of course that will put any lens with a teleconverter attached at a disadvantage, because you are basically ignoring the improvement in reach. It would be like switching from a 300mm f/2.8 to a 300mm f/2.8 + 2x TC, then doubling your distance to the subject by getting up and walking farther away from it. No one does that. You use a longer lens from the same vantage point.
Standardized test charts only exhibit the worst qualities of a lens or teleconverter, but never demonstrate their benefits. IMO, the proper way to demonstrate the true qualities of a lens in a visual, standardized lens test would be to shoot the test scene with A) identical framing and then B) at an identical subject distance, focusing the central point on the same location then sampling the results to the same final image dimensions. (i.e., in the test charts from the reviews you linked above, the little circular Proportional Scale ruler would be an IDEAL test case for standard distance comparisons.)
In every case, the longer lens, used at the same subject distance, will always resolve more detail and increase magnification. Unless you are using a TC with cheap plastic lenses that don't even qualify as optical grade, I cannot imagine that attaching a TC would ever result in lower final output IQ than using a lens without a TC. You either put fewer pixels on the subject and crop, or put more pixels on the subject and don't crop. More pixels on the subject, even if some of them have more CA, is almost always going to be preferable. It's really just a bummer that standardized tests never demonstrate that.
Upvote
0