Focus Hunting - camera or lens?

[tim]

I see reasonably frequent reports that when lens A is on a camera it hunts more than when lens B is on the camera. I always thought it was more the camera than the lens, after all the AF system's in the camera, the lens is mainly just hunk of glass. Can anyone shed any light on the subject?

PS The pun was intentional ;)

[joeseph]

Presumably lens A has a smaller aperture than lens B. Less light on a/f sensor results in less accurate autofocussing? or are you meaning same model / aperture lenses?
certainly my 75-300 hunts more than mt 100-400L... :-)

[tim]

I was meaning two different models of lenses. Larger apertures do make focusing more accurate... maybe that's the entire difference.

I'm having a stupid day today.

[foxbat]

The AF logic may be in the camera but the accuracy of the motor and gears and the quality of the image delivered to the AF program depend on the lens.

[tim]

Lack of accuracy won't make a lens seek through it's entire range, it'll just mean it doesn't focus very accurately.

[Mike H]

Tim, if you compare the performance of lenses that have the older type of Canon motor (arc form drive) to lenses with the newer ultra sonic motors there is a noticeable difference in the sound and level of hunting that occurs during focus. The older motors seemed to not have the same start and stop mechanisms in them, so along with being noisy, they tended to have jerky movements.

What you see in lenses now depends on the vintage of the design and the price range of the lens. Canon doesn't put the USM in a few consumer lenses even now, though most recent lenses, and certainly all recent L lenses, have USM.

And, of course, the higher end cameras do have better brains (computers) sending commands to the lenses to control them, so everything should focus faster and more smoothly on a 1D Mark II, for example, than on a Rebel.

SO, it's the type of motor, maximum aperture of the lens, and model of camera that control AF performance. Of course, the photog's brain can influence things by making smart choices about what to focus on (high contrast things work better than low, etc.).

Mike H

[tim]

Sure, better cameras have better AF performance, i'm talking about why one lens might seek more than another one the same body. I would have thought faster motors would make for faster seeking, but not really made any different to the number of times the lens hunts back and forward. Joesphs post makes the most sense to me so far.

[mdr]

Recently upgraded my 100mm macro to the USM version. Immediately noticed that hunting is much less with the USM version. As I had both, I tested it with same subject, lighting and settings. And indeed, the non-USM version hunts much more (and not just slower as you suggest Tim). It appears that the USM version doesn't get the focus completely right (crude motor?) and then starts to hunt, whereas the USM version seems to get it right the first time more often.

Anyone interested in me doing some comparitive shots between the two versions?

Also note that there are different USM motors, with the ring USM performing better than the micro USM. I have had both the 100-300mm ring USM and 75-300mm IS micro USM lenses, and from my memory, the 100-330mm hunts much less, although I have not done a straight comparison.

[kraterz]

The 100 macro non-USM is particularly nasty at focusing, it probably has the worst AF among all Canon lenses, not to mention the pathetically underpowered micromotor. It's not even AFD. Using the focus limiter helped me but my AF motor died three times, and I dumped the lens.

[mdr]

Kraterz - you want to buy my non-USM? ;)

[griff2]

Just a guess, but if focus is achieved by interpolation, which I suspect it is, then the slower non USM lenses will more obviously "hunt" and since they focus much more slowly, it's possible that, hand held, the field of view may have changed by the time focus is near achievement, causing yet more hunting.

I think it's more of a lens issue than the camera.

[Jon]

It's going to be a lens issue - if the motor doesn't stop with the lens focussed correctly, the camera sensor's going to send out a correction notice, so the lens will start focussing again. More precise, responsive motors and less gear lash in the lens = less hunting in the focussing process.

[robertwgross]

Part of the autofocus function is located in the camera body, and part is located in the lens. A lens with one type of autofocus motor is going to work differently from a lens with a completely different motor.

---Bob Gross---

[J Rabin]

Tim. Everyone here is likely right to some extent.
There are body issues. The IdMkII has separate focusing circuitry. A 1998 EOS-3 focusing snaps - bamm - so much faster than my 10D did, or 20D does now, with the same lens on it. Canons 1 series focusing is a different animal in terms of speed and accuracy.
On the lens side, of course there are widest aperture issues affecting viewfinder brightness and hence ability of the body to get sufficient light for fast focus. Also, it ought to depend on how FAR lens elements have to travel on their gear train. Rear element focus lenses have smaller, lighter, focusing elements that have little momentum, easy and fast and short distance to move. Lenses like the 200mm f/2.8L, the 100-300 consumer, etc. These focus fast. On the other side, we've got macro lenses with long helicals for fine manual focus, or front element focus zoom lenses that move bigger glass elements over long distances. There are exceptions. I believe my 24-70L and 70-200L are not rear focus, and yet they snap fast.

As Canon's evangelist Chuck Westfall wrote in one of his good help pieces, when you have a longer telephoto, and it goes out of focus, it REALLY goes out of focus. The elements have traveled far away and need time to come back.
So, lots of f/stop, light level, lens mechanics, and body issues. Part of the fun. J.

[Redbird_xo]

...after all the AF system's in the camera, the lens is mainly just hunk of glass...

In the EF lens world, there is ring-type USM, rear focusing full-time manual focus and there is linear extension AFD (the oldest EF AF system?). It appears that a lens has something to do with the overall AF one way or the other. Unfortunately, I am no expert to offer real technical meaning of those terms.

[Longwatcher]

Factors in focusing (lets see if I get them all)
1. Motor drive type (USM, AFD, etc..)
2. Aperture
3. Camera
4. zoom/prime
5. focal length
6. subject contrast
7. subject light/color
8. Image Stabilization (on/off)
9. subject distance from camera
10. individual lens/camera variations

I tried to put them close to the order I think makes the most difference.

[CyberDyneSystems]

We say "lens" when we refer to how well it focuses,. how much it hunts.. but the reality is it is ALLWAYS the combo of the lens and the body that added = the whole picture.

Since the body tends to remain a "constant" 95% of the posts ill be describing how different lenses effect the combo...

On occasion though you will statements to the opposite side of the coin discussing how the camera effects focus speed,...

ie: "this lens focuses MUCH faster on the MkII than it did on the 10D".

Even so, notice that the lens is still being described as the thing that s focusig faster.... it just is how we tend to discuss it.. it is part of out vernacular.

The lens plays a HUGE part in how well a combo focuses though. Aperture is just a small part of the eqution.. as mentioned above focus motors,. element design,. and logic chips play an important part.

An example,. the 400mm f/5.6L focuses significatly faster than the 300mm f/4L IS If we use aperure alone this would make no sense. But the 400mm is also a newer lens,.it's HSM assembly and more than likely the circutry are newer designs than the 300mm.

I imention element design... the focus motor has to move an element or lens group around. the engineering that matches the motor to the lens group has an impact. I don't know if this true or just a repeated online myth,.. but the slow focus of the 85mm f/1.2L is often atriibuted to the motor-to-element ratio/design.

Look up the focusing specifics for the 300mm f/2.8L IS and the "1"series and find out why it is that this particular combo is the fastest focusing that Canon (or anyone else arguably) has to offer in any lens body combo around... it may shed some additional light on how important both halfs of the combo are.

[robertwgross]

And then... there is the possibility of user error.

I'll be walking around with my Canon 100-400 lens and then find a subject at 5 meters. Sometimes my lens will hunt in and out, in and out, and never find focus. Then, if I am smart, I will look at the focus range switch on the lens. If there is a problem, it is because I had it set to 6.8-infinity (meters), and for a subject at 5 meters, I need to switch it over to 1.8-infinity (meters). Setting it for 1.8-infinity is more foolproof, but setting it to 6.8-infinity does autofocus much faster for a subject in that range, like at 20 meters.

---Bob Gross---

[tim]

This isn't the simplest of subjects eh? I'm still not convinced the type of motor in the lens makes much difference to the number of seeks, but it will seek more quickly. There are plenty of factors all contributing to it. I want to understand as much as I can about the camera, so as to take better photos.

[Longwatcher]

And then... there is the possibility of user error.


I knew I left something off my list.
0. User error
12. Murphy's Law effect
:)