Focus problems with my Sigma 10-20

[mumrar]

I purchased a Sigma 10-20 f4-5.6 EX DC HSM some time ago, but it doesn't always get that many outings. When it has, I have noticed a little bit of blur always seems to creep in on to the right side. Not only that, but the AF through the viewfinder, whilst giving a confirm beep/flash, nearly always is a fraction off being 'dead on'. However, if I use Live View (painfully slow of course), it seems to nail the focus right on. I've attached three 100% crops of the same shot taken using the two different focus methods.

Viewfinder focus - top right

Live view focus - top right

Viewfinder focus - bottom left

Live view focus - bottom left

Viewfinder focus - centre (just below focus point)

Live view focus - centre (just below focus point)

It's obvious to me there's a problem, but my question is where do I send it, is there anyone who'll look at it before chardging me? I also don't want to be without a lens for a while, just to hear all the 'within perametres' response so often mentioned with this sort of problem. I would want the decentering analysed as well. Just to add, I use 2x450D bodies, so no micro-adjustment settings to help fudge around the problem.

Thanks for looking, and I hope people have some advice for me to use to help fix this as the price of a new one is well over £100 more than I paid for it.

 

[_andyjb]

I know it's not much help, but manual focus through the viewfinder on my 40d is quite a bit off whereas when using liveview it nails it. But i see this problem is with autofocus.

What's the body and focus like with other lenses. is it only this lens that is out?

andy

 

[squizza]

You'll get distortion on a lens like this, especially high f-stops and especially wide open (10mm). Can we see the original uncropped?

 

[neil_g]

Can we see the original uncropped?

agreed..

also the thread title is harsh.

 

[mumrar]

Well, as requested here are the uncropped versions, although it's hard to see any problems at this size of course. Shots were taken at 15mm f/5 ISO100. I'm well aware of the expected distortion with a wide angle lens, but the point is, ignoring the decentreing for one moment, I get a sharper overall image using live view, than with focus through the viewfinder the traditional way. All that changes is the method of focus, not the f-stop, the ISO, the shutter speed, my position, or the position focussed on. Therefore, surely that's not right??

Viewfinder view

Live view

 

[mumrar]

agreed..

also the thread title is harsh.

Well, you may think so, but the number of people having back/front and god knows what else focus issues with their Sigmas, that come back perfect after receiving attention, I don't think it paints a wholly inaccurate version of events. To answer an earlier question I forgot, all other lenses perform faultlessly focussing through the viewfinder, and two of those ARE Sigmas, so it's not like I have a vendetta

 

[dinners]

If you're saying the lens produces decent results when using live view to focus - then I would have thought the lens must be OK.

How is autofocus on your camera with other lenses ?

How is manual focus through this lens ?

 

[squizza]

Belive me I don't mean this is a harsh way at all but a landscape at f5? I would stick it on a tripod, f16 or even lower, longer exposure and you'll get the sharpness you want.

Edit: Just because a lens opens up very wide doesn't mean it should be used

Edit 2: You've got to remember that your focus points are extremely small (leaves, bricks etc..). Which would also explain the blurs.

 

[Nifkin]

Well, you may think so, but the number of people having back/front and god knows what else focus issues with their Sigmas, that come back perfect after receiving attention, I don't think it paints a wholly inaccurate version of events. To answer an earlier question I forgot, all other lenses perform faultlessly focussing through the viewfinder, and two of those ARE Sigmas, so it's not like I have a vendetta

Not wanting to have a pop at the OP, but it's thread titles like this one that proliferate the myth about Sigma having terrible QC. Probably, for every 1000 Siggies sold, 950 people are happy with them, and don't post about it, whereas the 50 who have/think they have focussing issues shout loud. There seems to be no end of threads like this where technique appears to be the problem issue, but, because the lens is a Sigma, it's immediately assumed to be a duff copy. I wonder how many perfectly decent Sigmas get returned for just this reason?

I've got a few myself (including the 10-20), both 2nd hand and brand new; none have been recalibrated and all work wonderfully as long as you know how to use them right

 

[neil_g]

Not wanting to have a pop at the OP, but it's thread titles like this one that proliferate the myth about Sigma having terrible QC. Probably, for every 1000 Siggies sold, 950 people are happy with them, and don't post about it, whereas the 50 who have/think they have focussing issues shout loud. There seems to be no end of threads like this where technique appears to be the problem issue, but, because the lens is a Sigma, it's immediately assumed to be a duff copy. I wonder how many perfectly decent Sigmas get returned for just this reason?

I've got a few myself (including the 10-20), both 2nd hand and brand new; none have been recalibrated and all work wonderfully as long as you know how to use them right

exactly my point. we've got 4 sigmas here over varying focal lengths and theyre all fine (other than user error lol)

 

[Willid1]

But i see this problem is with autofocus.

I agree with this, I had the same issues with my Canon 10-22 and was ready to blame the lens.

Now I just manually focus it to the infinity mark if my subject is further than arms length away (especially at 10mm).

I don't think the 40D AF is good enough to focus at F3.5/F4.5 especially when presented with the tiny details you get when using UWA lenses.


HTH

David

 

[lostsoulal2]

Hello Matt,

I would imagine that if you printed out either version of your photos at ~A4 size, that both would be "acceptably sharp". Looking at a photo at 100% on a screen is always going to show some softness.........

But looking at your subject matter, the actual focus point (assuming it was just the one centre point) does not have an identifiable high contrast area to lock onto. Live view works in a slightly different manner and achieves focus by contrast edge detection.

A wide angle lens should be "acceptably sharp" from a few feet / metres away and what may be better practice for you is rather than trying to focus on a low contrast subject, that you focus on something near to hand with high contrast at the "hyperfocal distance" (a fancy term that means the distance setting at any aperture that produces the greatest depth of field) and then recompose.

If you do a search on here, there have been many threads on the subject or go to the DOFmaster site where this is explained in more detail.

Good luck, but the lens appears to be working correctly to me, it seems its the subject matter which is causing the lens/camera to struggle to find an acceptable high contrast area to focus onto.

Regards

Roy

 

[fracster]

exactly my point. we've got 4 sigmas here over varying focal lengths and theyre all fine (other than user error lol)

Same here.............

 

[mumrar]

Thread title has been changed. I can understand all about what is being said, with the exception if recommending an f/16 aperture. As the live view shot shows, DOF is perfectly acceptable at f/5, but absolutely nothing in the standard AF is sharp, whereas the entire live view image is. I have shot plenty of things with less contrast that I've shot with my 70-200 @f2.8 and it's nailed it spot on. So how, when working with such narrow DOF on a telephoto can it get is right with moving subjects, then mess up with a stationary wide angle shot with infinitely more DOF? Can anyone think of what would be an ideal subject to perform another test with on a tripod? After all, the subject is being blamed here, so enlighten me as to what would provide a fair test.

 

[lostsoulal2]

After all, the subject is being blamed here, so enlighten me as to what would provide a fair test.

Anything in good ambient light conditions with a high contrast would be suitable, like a wine bottle with a white label / large black writing or a Breakfast Cereal box or even just print out on a sheet of A4 some words, (T / I / L) at a large font size~ 140-200 and then fix it to a fence about 3-5 metres away.

Set the camera parallel to the subject and focus using the "One Shot AF Mode - "Centre Point" only.

Hopefully this will prove the lens to be okay.

 

[Trixster]

Looks sharper than the 10-20 copy I had, particualry on the right-hand side of the frame.

 

[trencheel303]

That lens clearly has a problem - you will need it calibrated by sending it to Sigma. If it is not in warranty, this will cost you money. How much I don't know.

I had wondered about your copy of the lens before because as I've mentioned once or twice on TMR, it produces bagloads more CA than mine did as well.

The fact that is focussing really badly when using phase detect (but is acceptably sharp on live view contrast detect) further goes to strengthen the point. My own 10-20 never had these problems, and to suggest that it's an AF problem, or because it was at f/5, is stupid.

Belive me I don't mean this is a harsh way at all but a landscape at f5? I would stick it on a tripod, f16 or even lower, longer exposure and you'll get the sharpness you want.

This lack of sharpness is not due to camera shake, f/5 or the fact that it's not on a tripod.

Edit 2: You've got to remember that your focus points are extremely small (leaves, bricks etc..). Which would also explain the blurs.

No it wouldn't. Even wide open at the distance he is shooting (to infinity) DOF will cover EVERYTHING.

Seriously, you people need to wake up and smell the coffee. There is nothing wrong with his thread title because of the word 'Sigma' is in it (it's called being descriptive) and to suggest that this is all user error is alarming; but sadly a thing that is becoming more and more common on TP these days. Just becuase XXX amount of people have 43 sharp Sigmas doesn't automatically mean Matt's is a faultless copy either.

 

[fracster]

Can`t be bothered.