AF Speed - What does it mean?

I've seen tests where you focus a lens at infinity then observe how long it takes to zip to MFD and we call that focus speed. I wonder though if it really means anything.

I shoot sports. I very rarely have to push a lens through a long focus change (infinity to 10 feet). Most of the time, the focus changes are measured in just a few feet as a player moves up / down the field / court / track.

Is a lens' ability to quickly move from infinity to MFD any indication at all of its ability to track a moving player accurately? I tend to think they are two different things. I tend to think that AF speed is really secondary to AF accuracy but I don't know how to measure it.

I have circumstantial "evidence" (gut feel) accumulated over the years as I've used various lenses on 40D, 1D and 7D bodies:

• 85 f/1.8 - Excellent. Very consistent & accurate
  
• 50 f/1.4 - Good . Fairly consistent and accurate
  
• 24-70L - Good. About the same as the 50 f/1.4
  
• Sigma 50 f/1.4 - Fair. Not as good as the above.
  
• Tamron 28-75 f/2.8 - Poor. Misses often
  
• Tamron 17-50 f/2.8 - Poor. Misses often
  
• Sigma 85 f/1.4 - Good. Maybe even Excellent.
  
• 70-200 II - Exellent, rarely misses
  
• 70-200 II + 1.4 TC - Excellent, rarely misses
  
• 400 f/5.6 - Excellent
  
• Sigma 100-300 f/4 - Good. Misses occasionaly
  

You have bad copy of 28-75 2.8.
My focuses quick enough and missing focus not very often.
My fastest lens to focus is 17-40L.

When the 28-75mm Tamron was introduced, I tried the lens at my local photo store. At that time I was using a 28-135mm f/3.5-5.6 IS lens as my mid-range zoom.

The Tamron's a/f speed and accuracy did not seem to match well against the 28-135mm (which has USM focusing) so I stayed with the 28-135mm until I purchased a 24-70mm f/2.8L lens. Of course, this was not a scientific test and I have read many posts which both support and conflict with my first impressions of the 28-75mm Tamron.

Now I use a 17-55mm f/2.8 IS and a 70-200mm f/4L IS as my primary lens duo. Both of these lenses focus fast and accurately. My 300mm f/4L IS and 400mm f/5.6L lenses are also fast and accurate in their autofocus.

However, my 90mm f/2.8 Tamron AF SP Macro lens does not autofocus fast but, it is quite accuate when it locks on. This doesn't bother me since I usually don't need speedy auto-focus when doing maro or close-up work. I mostly use manual focus for maro shots...

Speed and accuracy are important to me as an event photographer. One image might be right in front of me and another might be halfway across the room. I need my lenses to find them when I do. Nothing is more frustrating than a great moment missed merely because focus couldn't be achieved in time.

I have found that my 85mm is the fastest focusing lens I have at the moment, but it has the shortest focus throw. Now when I had my 17-55mm F2.8 IS that is still considered to be the fastest focusing lens on a 40d and above. It has a shorter throw and can move insanely fast.

One thing I have noticed though with my Tamron lenses is that they seem much slower then you think they are because of the noise. If you take that away they are rather fast and will keep up with my 50mm F1.4 no problem.

Nick

kf095 wrote in post #15086758
You have bad copy of 28-75 2.8.
My focuses quick enough and missing focus not very often.
My fastest lens to focus is 17-40L.

I had a good copy (two, actually) of the Tamron 28-75. My comments related to hitting focus for indoor basketball / sports photography.

Is your's performing well in that situation or are you talking about AF in general from that lens?

I'm not using Tammy for sports, it just general use zoom for me. Walkaround and portraits.

I have a tamron 17-50 and its great lens for walkaround.....however it would not be my 'go to' lens for sports. The 17-55 performs alot better IMO, for outside use in same focal range. For focus speed, especially inside sports, I always liked the 85 1.8. Just depends upon the situation I guess.

n1as wrote in post #15086748
I've seen tests where you focus a lens at infinity then observe how long it takes to zip to MFD and we call that focus speed. I wonder though if it really means anything.

There are a lot of variables in these Youtube videos too. For instance if I do the test in one shot with my 7D and then switch to servo, servo mode makes it much faster. These Youtube videos never mention that setting it seems. Then I can take the same lens and switch from my 7D to my 350D and the results get much slower. So basically when watching those videos, unless you see a bunch of lenses all tested by the same person using the same settings and same camera, they are pretty useless.

Now I can test all my lenses at home this way and the results are pretty much what I get when using them as far as far as locking on speed goes. Tracking ability is a different story. Those tests don't really mean much for that. My Sigma 30/1.4 is super fast at locking on, but it doesn't update the focus distance as often (frequently) as my Canon USM glass and thus gives a higher probability that the subject has moved outside of the focus plane when actually taking the picture and my real world shooting results do show a lower hitrate with it. A slight one but it is still measurable.

Also, I used to have the Tamron 28-75 and I will echo what you said about it. Almost horrible servo performance.

Of the other lenses I have:
17-50 fast and good accuracy. An occasional miss
24-105 excellent, almost perfect
70-200/4 IS excellent, almost perfect
18-55 slow but very accurate
55-250 slow but very accurate
30/1.4 fast a little more misses than 17-50 in servo. Excellent in one shot
50/1.8 slow and bad. A total crapshoot and this has been calibrated by Canon
60/2.8 fast and accurate but will often hunt in macro land for portrait shots
85/1.8 fast and very good. I get a few more misses than the L's but less than 30/1.4

Other lenses I've tried (for indoor basketball):

• EF-S 60 Macro - Too slow
  
• Sigma 70 f/2.8 macro - way too slow
  
• 17-55 f/2.8 IS - Great
  
• 35L - No problems
  
• 135L - Great, but a bit long