Factors that affect speed of AF

[truonda]

This was a random thought that came to mind and I wasn't able to find a good article about. What factors influence how fast AF focuses? I guess it's not a big topic?

[Ken MT]

This article gives a good summary to start.

http://en.wikipedia.org/wiki/Autofocus

[Colin Southern]

Hi Daniel,

No great surprises:

- Available light

- Lens maximum aperture

- Camera processing speed

- Lens electronics and mechanicals

- Scene contrast

are the immediate ones that come to mind.

Some cameras like the Canon 1Ds3 / 1D3 / 1D X have separate CPUs just for handling AF!

[RustBeltRaw]

Expanding a bit...

- Available light

- Lens maximum aperture

...

- Scene contrast

These are all closely related. Lots of light generally means you can find a contrasty area on your subject, which will make the focus system's job much easier. Lenses focus with the aperture wide open, so a wide lens lets in more light and provides stronger light-to-dark transitions. Eye sockets or the line of a dark shirt against light skin make good targets. In extremely uniform lighting (high or low-key images), a lack of contrast or limited high-contrast targets will make focusing a little more difficult.

The aforementioned factors generally affect whether you can autofocus, and assuming you can, the following factors affect the speed.

- Camera processing speed

- Lens electronics and mechanicals

A camera with a fast AF processor will be able to parse data from more points, react more quickly to subject motion, and provide more instructions to the focus motor more quickly. The motor itself is generally one of three types. A servo, which is slow, noisy, inexpensive, and frequently found on lower-end lenses. A stepper motor, which is somewhat slow, quiet, and a little pricier, generally found on video lenses. Finally, hypersonic/ultrasonic focus motors, which are considerably more expensive, fairly quiet, and really quick. This is the motor type (Canon calls it USM) found on most pro-market lenses. Naturally, the focus motors are geared, so there's room for speed tuning outside of motor selection.

[NorthernFocus]

...Lenses focus with the aperture wide open, so a wide lens lets in more light and provides stronger light-to-dark transitions.....

This is often overlooked when considering lens purchases. Even if one rarely shoots at maximum aperture, the camera's AF system takes advantage of it every time you focus on something.

[rpcrowe]

While a wider maximum aperture will let in more light and can be a great help in fast focusing, there are (as well mentioned above) some camera/lens combinations which have exceptionally fast and accurate AF, despite the lens being of a moderately small aperture.

Case in point: the Canon 7D with a 400mm f/5.6L lens is an extremely fast and accurate focusing combination despite the lens being a maximum f/5.6. The 400mm f/5.6L will provide faster and more accurate auto focus than a lesser lens such as the 55-250mm f/4-5.6 IS, shooting at the same aperture...

While the 7D has the most capable auto focus of all Canon 1.6x cameras, I have had extremely good results with this lens on older and "less capable" cameras such as these shots of hydroplanes traveling in excess of 100-miles per hour; shot using a Canon 30D...







One thing that is seldom mentioned is that for the initial shot, having your lens focused close to the expected distance of your subject may speed the AF. As an example, if you are shooting distant subjects, starting with your lens at infinity or close to infinity, will enable you to achieve faster focus on that first shot than if your lens is focused at the MFD.

[t6b9p]

In addition to the above, and especially this one

One thing that is seldom mentioned is that for the initial shot, having your lens focused close to the expected distance of your subject may speed the AF.

reduce the number of focus points, set the focus range (if applicable), turn off IS or VR, and I swear that my 300/2.8 seems to be faster when I use a battery grip.

[RustBeltRaw]

One thing that is seldom mentioned is that for the initial shot, having your lens focused close to the expected distance of your subject may speed the AF. As an example, if you are shooting distant subjects, starting with your lens at infinity or close to infinity, will enable you to achieve faster focus on that first shot than if your lens is focused at the MFD.

This is an excellent point. Do not overlook the importance of pre-focus, good panning, and predicting the action when focusing on tricky subjects. Note that these factors have nothing to do with the camera.

[Black Pearl]

In addition to the above, and especially this one

reduce the number of focus points, set the focus range (if applicable), turn off IS or VR, and I swear that my 300/2.8 seems to be faster when I use a battery grip.

Wouldn't having the IS or VR on help as the camera is seeing a steadier image to lock onto?

[t6b9p]

Wouldn't having the IS or VR on help as the camera is seeing a steadier image to lock onto?

Generally if you are concerned about AF speed it's because you are trying to capture a fast moving object and under these circumstances IS/VR is probably the most significant contributor to slowing down AF.

[NorthernFocus]

Generally if you are concerned about AF speed it's because you are trying to capture a fast moving object and under these circumstances IS/VR is probably the most significant contributor to slowing down AF.

I've seen this mentioned multiple times in forum discussions like this but can't recall seeing any basis for it ever presented. Can you share any references in support of this?

Having VR active definitely slows overall responsiveness off the camera. But within the limits of my perception I'm not sure that I can tell the difference. So any references siting measurable test data would be appreciated. Like many others I'm always looking to optimize performance. Unfortunately I have a technical bent and tend to rely more on measurable data than my own flawed intuition.

[Black Pearl]

Generally if you are concerned about AF speed it's because you are trying to capture a fast moving object and under these circumstances IS/VR is probably the most significant contributor to slowing down AF.

Can you quote a source for this please?

[t6b9p]

Unfortunately I have a technical bent and tend to rely more on measurable data than my own flawed intuition.

We are alike in that respect. Measurable test data I have not seen, recommendations from respected photographer and equipment testers there is.

However, I was originally tipped off to this by some fellow BIF photographers and I thought I would try it on the puffins on Skomer. After the young hatch, the parent bird is returning to the nest with food while being chased by the gulls. When they finally head for the burrow they turn on the speed and may be approaching almost head on at around 50-55 mph and can approach within several feet. An incredibly tough subject - I tried it with and without VR and all I can say is I never use VR anymore unless in low light or under 1/500th second. E.J.Peiker of Naturescapes suggest turning off VR (for any subject) at shutters speeds over 1/750s for better image quality. Thom Hogan suggest turning it off over 1/500s, Lloyd Chambers (digilloyd) similar, and there are others. However, equipment and technologies change, hopefully for the better, so my advice is try yourself and see, what do you have to lose? Just remember to pick an appropriate subject (speed and direction) for your evaluation.

I have seen some comments that turning off VR may be more important for Nikon than Canon but I only have Nikon so I cannot comment on that aspect. However, my brother was with me shooting a Canon with a 500/4 and had more success without IS on.

For panning a BIF across the field of view, or large slow birds, VR may be of use however I do not consider this a "fast AF" requirement scenario.

[NorthernFocus]

...E.J.Peiker of Naturescapes suggest turning off VR (for any subject) at shutters speeds over 1/750s for better image quality. Thom Hogan suggest turning it off over 1/500s, Lloyd Chambers (digilloyd) similar, and there are others....

This whole discussion is about AF speed. I don't believe that any of the sources that you reference suggest AF speed is improved with VR off. They suggest better IQ can be achieved. Based on various subjective reasons. I don't believe any of them claim faster AF. And in your example of your own experience, what leads you to conclude that overall higher hit rate is due to faster AF?

[Colin Southern]

Here's an idea - get out there and test it

[t6b9p]

suggest turning off VR (for any subject) at shutters speeds over 1/750s for better image quality

I should have made it clearer and separated the IQ references.

It takes time for VR to "lock in" and when the subject is approaching at high speed, that is an issue. As I said, it was suggestions from fellow photographers - I tried it and found a much higher success rate, as did my brother.

If you google and search you will find this is not an uncommon technique. You don't have to believe me, it looks like you have the right equipment so go out and try for yourself.

[Bobobird]

Simple answer - me!

[NorthernFocus]

I should have made it clearer and separated the IQ references.

It takes time for VR to "lock in" and when the subject is approaching at high speed, that is an issue. As I said, it was suggestions from fellow photographers - I tried it and found a much higher success rate, as did my brother.

If you google and search you will find this is not an uncommon technique. You don't have to believe me, it looks like you have the right equipment so go out and try for yourself.

Well there you are. My point earlier. VR is slow. Has nothing to do with AF. The AF is doing its thing a lot faster than the VR.

I do think one effect that VR has on AF has nothing to do with speed. With VR on, the image floats around a bit. So you can't hold the AF point exactly where you want it. If you're really trying to get critical focus it can make a difference.

And thanks for the recommendation to get out and try it. Maybe I'll give up shooting still life images and give it a try.

For the record I shoot with VR off at high ss. It's simple logic. Why throw another variable into the equations when it's not doing any good anyway. It takes VR so long to settle down in the first place, I can't figure out what it's doing in the 1/10 of a second between frames

[NorthernFocus]

Simple answer - me!

Ditto. But it is much easier on the ego to blame the gear

[Bobobird]

One reason to love the new generation Canon lens. VR/IS does not kick in at all and only does when the shutter button is fully depressed. Have not seen that to have any adverse affect on sharpness. The older modes 1 and 2 are still there if needed.