AF miss rate

[spacecadet]

I've always had a few misses with my cheap kit zooms, but the number seems to have been a bit higher on my recent trip, as high as a few percent.

Sometimes it looks like some sort of de-centring as there seem to be, as used to be said, "circles of confusion everywhere" rather than the plane of focus just being off.

What does the team think?

[wiskerke]

Firstly I would google that lens for known issues.

You could try finding another copy and do a side by side test.

Or you could upgrade. Cheap kit zooms are cheap. Some are better than others. Some are good/bad at very specific apertures or focal lengths.

Have a look at the DXO lens database.

AF could be the body: check it with other lenses as well. It could even be just the setting of the AF preferences.

 

wim

 

edit: all that assuming you do know the difference between AF-C; AF-S and the wide variety of picking AF points or zones. All depending on you brand and model of course.

[spacecadet]

Thinking about it, it's definitely a random rather than missed  focus problem. Some images are just off. I was wondering if it's something I should just expect.

Here's one- no reason for it not to be sharp that I can see. f5.6, 125th.

https://drive.google.com/file/d/0B2tpC1O6KZhUa2FnNUhqZFFtXzQ/view?usp=sharing

I think you can view it full size.

[ReeRay]

If the lens usually performs but periodically focus is missed then it all points to user error. Most cameras give off an audible or visual signal in the viewfinder when AF is achieved. If you shoot too quick before it locks on or use C-AF and it's hunting then that would your problem.

 

The above happens to me periodically when I see a quick grab shot and fire off before the AF is locked on.

 

It happens.

[Bill Brooks]

There is a two part series here on manual focussing that has a lot of good technical information that can apply to AF as well.

 

http://lenspire.zeiss.com/en/focusing-zeiss-dslr-lenses-part-1/

[Bryan]

I can't see that image full size Mark. It's coming up 1600 by 1065 in PS. At that size it doesn't look bad.

 

The only auto focus lens that I regularly use is a 19mm f2.8 Sigma. I have occasional problems with this lens/camera combination missing focus on the NEX 6. For example I might shoot a panoramic view and 3 of the 4 shots will be sharp and one unusable, and that's using centre point focus. For most of my stock shooting (plenty of time to mess about) manual focus is fine and 100% reliable.

 

I have also had lenses with decentring problems, including an expensive Canon 24-70 L zoom and a (relatively) cheap Sony 55-210. They both went back to meet their maker, but the Canon lens was never 100%

 

I would be inclined to test the lens using a tripod and manual focus at various subject distances and focal lengths.

[spacecadet]

If the lens usually performs but periodically focus is missed then it all points to user error. Most cameras give off an audible or visual signal in the viewfinder when AF is achieved. If you shoot too quick before it locks on or use C-AF and it's hunting then that would your problem.

 

The above happens to me periodically when I see a quick grab shot and fire off before the AF is locked on.

 

It happens.

It's not that. It won't fire without AF confirmation.

 

Bryan, I don't know how to deliver a full size jpeg then. it's not really up to scratch full size. I can get some of these through by downsizing but I'd rather not. Some of this may be down to my going from 4900 to 5400px long side last year I suppose.

Looking at some of them, they seem to be sharp to one side whereas normally they'd be a little out if the lens was working "normally" which makes me think that an element may be out of alignment on the offending images. Interesting to hear about the decentring though, that's where my money is tending.

The lens tests fine, it's just the odd duff image. I've got used to fast AF and my shooting style has got a bit, er, free is a polite way of putting it, so I don't think I would go back to routine MF. Maybe I'll try it for a day though.

 

Thanks all for your thoughts. More welcome.

 

Bill, that is an eye-opener, pun intended.

[wiskerke]

Thinking about it, it's definitely a random rather than missed  focus problem. Some images are just off. I was wondering if it's something I should just expect.

Here's one- no reason for it not to be sharp that I can see. f5.6, 125th.

https://drive.google.com/file/d/0B2tpC1O6KZhUa2FnNUhqZFFtXzQ/view?usp=sharing

I think you can view it full size.

 

Yes it's even downloadable at full size with the Exif in tact.

Google seems to sharpen a bit.

 

The sharpest point is on the bells or the somewhere at the far corner of the roof. Probably that's where the AF point was.

In general if it's an AF fault, one would expect a sharper point somewhere closer or further away.

 

Could it be you have it on AF-C and it's still hunting a bit because there's not much contrast in the middle?

However at 24mm / 5.6 the whole facade should be more or less in focus.

 

Test it some more; try to recreate the problem. Recently I had something like this with a very trusted MF lens. Eventually I went back to the exact same location twice. Still that didn't solve my problem. As usual there were more than one cause: I did not focus with closed aperture, although I thought I did and there was extreme plane of focus curvature the wrong way around, resulting in a oof spot right in the middle. Took me 3 weeks.

 

wim

[spacecadet]

 

Thinking about it, it's definitely a random rather than missed  focus problem. Some images are just off. I was wondering if it's something I should just expect.

Here's one- no reason for it not to be sharp that I can see. f5.6, 125th.

https://drive.google.com/file/d/0B2tpC1O6KZhUa2FnNUhqZFFtXzQ/view?usp=sharing

I think you can view it full size.

 

Yes it's even downloadable at full size with the Exif in tact.

Google seems to sharpen a bit.

 

The sharpest point is on the bells or the somewhere at the far corner of the roof. Probably that's where the AF point was.

In general if it's an AF fault, one would expect a sharper point somewhere closer or further away.

 

Could it be you have it on AF-C and it's still hunting a bit because there's not much contrast in the middle?

However at 24mm / 5.6 the whole facade should be more or less in focus.

 

 

AF-S, focused on the church door. I noticed the sharpest point, which made me think of either curvature or decentring.

Incidentally, San Julian de los Prados in Oviedo, if you didn't know it. Exquisite. Great town, they even have 6.5% beer.

[wiskerke]

 

 

Thinking about it, it's definitely a random rather than missed  focus problem. Some images are just off. I was wondering if it's something I should just expect.

Here's one- no reason for it not to be sharp that I can see. f5.6, 125th.

https://drive.google.com/file/d/0B2tpC1O6KZhUa2FnNUhqZFFtXzQ/view?usp=sharing

I think you can view it full size.

 

Yes it's even downloadable at full size with the Exif in tact.

Google seems to sharpen a bit.

 

The sharpest point is on the bells or the somewhere at the far corner of the roof. Probably that's where the AF point was.

In general if it's an AF fault, one would expect a sharper point somewhere closer or further away.

 

Could it be you have it on AF-C and it's still hunting a bit because there's not much contrast in the middle?

However at 24mm / 5.6 the whole facade should be more or less in focus.

 

 

AF-S, focused on the church door. I noticed the sharpest point, which made me think of either curvature or decentring.

Incidentally, San Julian de los Prados in Oviedo, if you didn't know it. Exquisite. Great town, they even have 6.5% beer.

 

 

All the more reason to return at least twice to get to the bottom of this.

;-)

 

wim

[spacecadet]

I actually did LOL.

Thanks, Wim.

[Bill Brooks]

Could it be the lens’s image stabilization?

 

I have a 70-200 F4 IS Canon L zoom that most of the time is very sharp.

 

However on occasion it will produce an image that looks like the lens is slightly decentred. Slightly softer on either one or the other side. It does not do this with IS off and on a tripod.

 

Image stabilization decentres the lens to keep the image in the same place on the sensor, so when IS makes an adjustment you may be only seeing an area outside the sharpest centre point of the lens. This is why the effect would be intermittent and look like the lens is decentred. The effect would be greater with IS on a kit lens with a smaller sharp image circle.

[spacecadet]

Thanks for the thought -it's a Sony 18-55, so no lens IS, it's in the A58 body.

[phomme]

I shoot with Canon and Olympus bodies, and in both cases I noticed that my  Pro grade lenses do focus faster than my cheaper non Pro grade lenses.

 

Of course you need to compare $1,400 compared to $199 price difference for similar focal point lenses.

[spacecadet]

I don't think I've spent $1,400 on cameras in my life yet.

[Allan Bell]

Had a look at your image Mark and was able to zoom in on it using the "+" sign at the bottom.

 

Definitely OOF but because it is all over the image I am leaning to slight (very slight) camera shake.

 

If you were focused on the door and the focus system was off then another part of the image would have been in focus.

 

JMO.

 

Allan

 

Don't know what shutter speed you were using at the time.

 

ITMA

[John Mitchell]

Thanks for the thought -it's a Sony 18-55, so no lens IS, it's in the A58 body.

 

The Sony SAM lenses have really clunky AF. The Sony e-mount lenses, even the kit zooms, have internal AF and are much faster and more accurate IME.

 

I found this article on AF enlightening.

[spacecadet]

1/125. Wim has identified the sharpest point so I don't think it's camera shake which does look a bit different on the A58. This is one of a few with a similar problem.

[M.Chapman]

I got rid of my Canon 550D and lenses because the AF was unreliable with a variety of lenses, but especially noticeable on long zooms.

In my case the failure percentage was much higher, to the extent that I always took 2 or 3 pictures of everything to hopefully get a really sharp one. In the end I found I got better results if I focussed twice on the same point. The first focus would move the lens from wherever it was last, to close to the "perfect focus" position. The second focus made a minor correction and got a better result. I saw the reason for this explained somewhere. It could be internet rubbish, but he article said that Canon DSLR lens focussing algorithm was "open loop". User presses shutter, camera determines the amount of focus error from the phase detect sensor and commands the lens to move accordingly. Shutter fires. Camera does not verify focus has been achieved after lens moves and before firing shutter. This means that the further the lens has to move, the more the system is relying on the accuracy of the lens calibration. I've no idea how the Sony works. I also found the IS caused de-centering and varying vignetting issues at wide angles, especially on the Canon EF-S 10-22mm. Combine this with front and back focussing issues, I gave up on Canon SLRs and swapped to mirrorless.

Focussing became 100% reliable

[Betty LaRue]

I thought it looked like camera shake, too. Must be that 6.5 beer. none of it really looked sharp to me. Wherever the focus point is, there's still a bit of blur. Of course the parts that weren't in the focus plane would appear worse.

 

But I do find that hard to swallow. It's not like this is your first rodeo.

If the SS was slower, maybe. 1/125 is plenty fast enough for hand-holding a 55mm at long end. I'm blaming it on the beer.

 

Betty

[Betty LaRue]

I got rid of my Canon 550D and lenses because the AF was unreliable with a variety of lenses, but especially noticeable on long zooms.

In my case the failure percentage was much higher, to the extent that I always took 2 or 3 pictures of everything to hopefully get a really sharp one. In the end I found I got better results if I focussed twice on the same point. The first focus would move the lens from wherever it was last, to close to the "perfect focus" position. The second focus made a minor correction and got a better result. I saw the reason for this explained somewhere. It could be internet rubbish, but he article said that Canon DSLR lens focussing algorithm was "open loop". User presses shutter, camera determines the amount of focus error from the phase detect sensor and commands the lens to move accordingly. Shutter fires. Camera does not verify focus has been achieved after lens moves and before firing shutter. This means that the further the lens has to move, the more the system is relying on the accuracy of the lens calibration. I've no idea how the Sony works. I also found the IS caused de-centering and varying vignetting issues at wide angles, especially on the Canon EF-S 10-22mm. Combine this with front and back focussing issues, I gave up on Canon SLRs and swapped to mirrorless.

Focussing became 100% reliable

+1

[spacecadet]

Thanks for that. I sometimes take two (very) similars for no good reason which can save me, but the only ones I notice are the ones where I didn't, of course. IS can be eliminated as it's in-body in the A58. My problem is much less than yours was fortunately.

There's a lot of AF sensor info in the EXIF which might help me if I could interpret it! The lenses tend not to be very good at edge of field which may mask the problem, but QC don't seem to mind.

Wim's point about focussing stopped-down will help me- I'd forgotten about it, if I ever knew.