What defines a "Fast" Lens?

[jared_kipe]

An experienced photographer can often manually focus and track focus faster than an autofocus system. However, for non-professional photographers, an auto-focus system is going to focus faster.

To be frank, I think you're fooling yourself. A good USM/HSM lens can find focus without any hunting and do so faster than I can imagine about twisting the lens barrel. In low light or other difficult situations maybe you can focus faster, but then again if you give the lens as much of a chance as your eye, meaning you point it onto an edge, I still think the lens will focus better than a human.

Unless you cheat by guessing the distance or something, but even then you'll never be as accurate without fine tuning and in the time to do that the lens would have found the focus most likely. Its not a bad thing, just something humans can't be better at. We can focus our eyes so fast because we have a direct connection with them. Same way the camera body can focus a lens.

I'm good at math, but I know a computer is going to be better a solving vector calculus DifEQs than I could. Unless I have some intuitive sense as to what the answer will look like.

 

[mrichmon]

To be frank, I think you're fooling yourself. A good USM/HSM lens can find focus without any hunting and do so faster than I can imagine about twisting the lens barrel.

To be clear, I am not claiming that I can focus faster than an autofocus lens. However, I have seen instances back when I was working for a large newspaper where the old time sports photographers who exclusively focus manually were able to get shots that the younger shooters who relied on autofocus missed.

One specific shot at a motorcycle grand prix that caught a rider in mid air while his bike tumbled above him was only caught by the manual focus shooters. The people relying on autofocus found that they had many frames of various bits of debris in sharp focus but the autofocus never settled on the rider.

Autofocus is good and fast, but there isn't a good way for the shooter to indicate what they want in focus. Even with portraits, an autofocus system will generally get the nose/cheeks or the forehead in focus when what you generally really want is for the eyes to be in focus.

I'm good at math, but I know a computer is going to be better a solving vector calculus DifEQs than I could.

But we are still struggling to produce a machine that can catch a ball, or hit a curve ball. Yet children who have no concept of the vector calc behind such feats manage to be able to do it. A photographer isn't solving differential equations when they are focusing a camera, rather they are relying on experience.

 

[jared_kipe]

But we are still struggling to produce a machine that can catch a ball, or hit a curve ball. Yet children who have no concept of the vector calc behind such feats manage to be able to do it. A photographer isn't solving differential equations when they are focusing a camera, rather they are relying on experience.

Thats what I'm saying, I can rely on my experience to have an idea of what the outcome should be. But a computer can solve the specifics much faster.

A ball's motion is quite frankly ridiculously simple (with respect to DiffEQs). Now add in some spin to make it harder, and the kids will start missing (I'm thinking tennis here where spin matters more).

We haven't made a machine that can do these well, but we haven't been working on it for millions of years. Biological systems have been being perfected for longer than humans have been considering such things.

I'm not saying I use auto focus all the time. I use both, when doing macro, autofocus isn't particularly useful because I can clearly see the details I'm focusing on. When doing a portrait, I would probably use autofocus on the subject if I cannot clearly see what I need to be focusing on, and maybe refocus after it has the lock. If I'm doing a landscape or something, particularly if I have to use a tripod. I'll just hyperfocal it, and not worry about focusing at all.