Auto focus - can you explain this.

FlyTVR

FlyTVR

Suspended / Banned
Messages
4,234
Edit My Images
No
I've been doubting the quality of my shots recently. Took some test shots this morning.

Lens Sigma 70-200 F2.8 HSM. Images no PP other than crop.

First shot is taken at F8 ISO200, 200mm (with on camera 580EX with some ex comp).

If you look at the anoted image, you can see that the AF point is on the pupil. However it appears that the eye lashes are the part of the eye that are perfectly focused.

f8.jpg



So, I repeated the same experiment but this time at F10 (different subject).

f10.jpg


So, what was the AF system doing in the first shot? Can anyone explain?
 
Last edited:
Backfocuing I would say..;) What camera are you using..
 
Sorry I can see now, have you tried your Micro Adjustment for this lens.;)
 
Definately back or front focussing, can never remember which way round they go.
 
Kev M said:
Definately back or front focussing, can never remember which way round they go.
Click to expand...

Back focus behind, F/Focus in front of a subject.;)
 
In that case I'd say it was front focussing because the sharp area is in front of the AF area.
 
Lens problem or body? Solution?

Id do more experiments but got to go out now.
 
probably a combination of both given they both built to tolerances rather than exactness.

Do you have micro adjustment on your camera? If not you might need to send the lens and body off for calibration.
 
I'd say that trying to test a lens on a 'live' subject will not tell you anything. You need a totally stationary subject and a rock solid camera/lens mount. Any slight movement of the subject at this shallow DOF will corrupt your findings. :)
 
The 5d ll has micro adjustment for lenses doesn't it ?
 
Yep, stationary subject at a decent distance, tripod mount and leave the flash off too. Fire with a remote if you have it and you have just about the most stable platform you can reasonably have.

Repeat the exercise with more than one lens. Generally if you change lenses and it's fine then that would indicate that it's a lens based problem. I have no experience of microadjustment but that may be well worth invesigating.
 
davekiddle said:
I'd say that trying to test a lens on a 'live' subject will not tell you anything. You need a totally stationary subject and a rock solid camera/lens mount. Any slight movement of the subject at this shallow DOF will corrupt your findings. :)
Click to expand...

It's shot at f8 though, shouldn't the dof be fine for a live subject?
 
Put the camera on a tripod and use a stationary object - i usually use a barcode on a cereal box to start, use af to focus and take a photo - switch to MF and liveview x5 or x10 take a photo -compare the 2 a see if you can see a difference. use a remote or the self timer.
 
Jayst84 said:
Also remember that the little red square isn't the same size or shape as the actual AF point.
Click to expand...

Hmmmmm. So what use is it?

Down hearted. £1800 camera - £500 lens and I don't even know if it's right or not!

No F1 tonight. Once little one is in bed it's our with the tripod and remote and test time.

Any ideas what a food static test subject would be?
 
Shoot a back focus test chart. Google will pull up the popular ones. Also map out the AF areas. They're thin strips, they are not single dots or squares. This will put you right.

Surefire way to test if its lens or autofocus:

LIVEVIEW.

It will use contrast detect autofocus (as opposed to phase detect). This SHOULD get you the sharpest image, regardless of AF fine tune (which only applies to phase detect). If it's sharp, its your AF fine tune that needs adjusting. If it's not sharp, your lens has a problem.
 
I'm no expert but might it not be the flash thats the culprit or more to the point a combination of all three.
You say they are @ 200mm with on camera flash, correct me if i am wrong but the flash wont be accurate at that range and might be throwing other things off.
Its just a suggestion and it might not be anything to do with it.
Camera, tripod and something stationary without the flash will tell i guess.
 
EyePiece said:
I'm no expert but might it not be the flash thats the culprit or more to the point a combination of all three.
You say they are @ 200mm with on camera flash, correct me if i am wrong but the flash wont be accurate at that range and might be throwing other things off.
Its just a suggestion and it might not be anything to do with it.
Camera, tripod and something stationary without the flash will tell i guess.
Click to expand...


Thanks. I first noticed issues in studio conditions. Re tried with 580 exii this am.

Going to run some more tests tonight.
 
focustestchart.com/focus21.pdf

I have used this in the past, simple and easy to use and shows more precisely what the lens is doing.

Paul
 
How would the flash throw the focus out?
 
From my very limited experience with pro tele zooms, at minimal focusing distances (ie 1-2m) they seem to usually be off by +/- 5mm with their focusing, which when you have very shallow DOF is annoying but I understand can be rectified with the in camera settings for front/back focusing.

In reality, I never shoot that close, so don't bother. I've seen it on two pro Nikon 80-200's though.
 
Just as a side issue, if I remember correctly - the little red squares are only an indication of where the sensor will try to focus. I believe that the area is slightly larger than indicated. In which case, it wll focus on the area of most contrast within its field - likely to be the eyelashes as they are nice black lines against pale skin. By contrast the iris detail is fairly subtle and will not be chosen by default.

This is why, for macro work I believe that most people focus manually.

However, I may be way off base.
 
grahamsewelluk said:
Just as a side issue, if I remember correctly - the little red squares are only an indication of where the sensor will try to focus. I believe that the area is slightly larger than indicated. In which case, it wll focus on the area of most contrast within its field - likely to be the eyelashes as they are nice black lines against pale skin. By contrast the iris detail is fairly subtle and will not be chosen by default.

This is why, for macro work I believe that most people focus manually.

However, I may be way off base.
Click to expand...

That sounds VERY plausible, thanks.

I'm too knackered tonight to do any more tests, and I've still not watched the F1. :)
 
EdinburghGary said:
I'm having a fairly major problem with my 85mm f1.4 just now, it's also mis-focusing more than I would like. Going to lace it with arsenic and send it to Nikon, needs sorting.

G.
Click to expand...

When I spoke to Fixation, they reckoned that they can't adjust any lens for focus that doesn't have an AF-S microchip. Any AF-D/AF lens uses the body to judge focus. Probs have to send the body in too...
 
Not a conclusive test at all.

There's clearly only a few mms of DoF. Can you be sure that neither you nor the subject moved between focusing and shooting? Did you use focus/recompose at all? The focus indicator only tells you which focus spot was used, which is not necessarily where it was positioned at the time the focus was set.

Having said that, if you do have a lens AF problem, it is quite likely that it could be out on extreme close up, but fine at normal distance. If you microadjuist for that, you'll throw it out for normal shooting. Very slight variance is not uncommon but is usually so small that it passes unnoticed.

If you use one of those focus test charts, be very careful and wary of what you find. They often lead people astray and you end up making things worse.

Most people don't use AF at that kind of distance, simply because it is hard to nail the focusing distance reliably. Turn off AF and manual focus roughly, then rock your body gently back and forth and watch the focus plane as it moves to exactly the position you want.
 
There is an AF microadjustment procedure documented rather well here - http://www.birdphotographers.net/forums/showthread.php?58042-AF-microadjustment-tricks.

For optimum results you do need to set up a controlled environment. The thread should cover what is required.

For a video demonstration of an alternative procedure for the assessment and adjustment of AF you may find the LensAlign videos helpful. The product is quite expensive but the procedure is worth understanding, even if you choose to fabricate your own focus test target and scale (as I do).

http://videos.lensalign.com/LensAlign/LensAlign-Video-User-Guide/6830310_fnpyr

It is also worth noting that engineering tolerances and practicalities of economics mean that there may be focus calibration errors out of the factory. Either the camera, or lens, or both may be at fault (although individually within acceptable limits). Some errors will net to zero. Others will compound to unacceptable levels. This article makes for interesting reading....

http://www.lensrentals.com/news/2008.12.22/this-lens-is-soft-and-other-myths
 
Last edited:
Ok - here are the initial results. No flash - just modelling lights. Tripod / remote. All shots fully open (either 1.8 / 2.8 depending on lens).

Firstly tried the 5D with the Sig 70-200 2.8
http://i146.photobucket.com/albums/r276/DavHughes/5dNifty50.jpg

Then thought I'd go all Canon, 5D with Nifty Fifty 1.8
http://i146.photobucket.com/albums/r276/DavHughes/5d70-200.jpg

For comparison, swapped bodies to 400d, same nifty fifty
http://i146.photobucket.com/albums/r276/DavHughes/400DNifty50.jpg

Thoughts?
 
tdodd said:
Use AF microadjustment. That what it's there for.
Click to expand...

I promise to have a proper read when I get home.

However, a quick google and I found quotes like "Adjust your lens for the normal distance you use it". This implies that by adjusting a lens for one distance may impace the sharpness at other distances. Is this the case :suspect:
 
If you micro adjust for a lens, does this mean that it will alter the focus for all your other lenses?
 
You must log in or register to reply here.
Back
Top