Auto Focus problem??

[mbird]

Hi All

Now I may be getting paranoid here, but I think I may have a focus problem.

Since getting my 50D, I have been somewhat disappointed with the sharpness of the images. I had similar problems with my old 350D, but assumed that was simply because it was a lower resolution sensor etc.

I always shoot in RAW, and leave the lens in AF mode (the same lens as I used on the 350D). The screen capture below illustrates what I mean (hopefully) as the focus point used was bang in the middle of the frame at ISO 100, F18, 1/6th sec, lens set at 50mm. You can clearly see that the enlgarged view seems "fuzzy", even though the camera was on a tripod and so camera shake shouldn't be an issue. The lens is a Canon 17-85 IS EFS, which I bought new (albeit from Ebay).

The problem seems worse in lower light conditions, although even in bright sunshine the images are not what I would call pin sharp.

Any ideas guys?

[Happy Jax 3]

Hi Mark

Does it have noise reduction and noise filter features, if so turn them off and retry, as you know I have olympus cameras and this gave the same problem.

[Guest]

Have you looked at the same RAW file in PSP?

I have found that Lightroom seems to "smooth" the look of some pictures.

[Guest]

Have you turned OFF the IS if on a tripod?

[pks1702]

Hi Mark,

It could be front or back focusing although your other images have looked good.

When I bought my 1DMk111 I was similarly not over impressed with the focus. However a chat with the Canon Rep at Focus On Imaging revealed the wonders of Micro adjustment. His contention was that the manufacturing tolerances however small between each camera and lens meant you needed to potentially micro adjust. I did have a go but there was a problem with the 1DmK111 and almost at the same time Canon recalled. The best tip the canon rep gave me was to 'pal' up with someone who had a 'pro' body and sign up to Canon CPS. Part of the service is that they will have your camera back within a couple of days. Off ours went and when returned are as I had hoped.

There are all sorts of recommendation and ways of micro adjusting - do a Google but the most cost effective way is to use a White Plastic rule angled at 45 degrees and take JPEG images as you adjust from plus to minus and review the results. You should be able to see from this if the camera is back or front focusing and adjust. On mine and I suspect yours is the same you can adjust for each lens.

http://www.f64.ro/downloads/Manuale/50D/Manual50D.pdf

Page 180 - Custom Function 111 -7

It might not be this at all but worth a look

[Guest]

Now I may be getting paranoid here, but I think I may have a focus problem.

Yet your pictures from Ranworth / Cockshoot were pin sharp and beautiful.

I would have thought if you had a front / back focus issue you would still have some sharp sections of the image, they would just be slightly in front of, or behind, the intended focus point. It would normally be more obvious in shots were you are using a much narrower DoF such as portraits / macros etc. Could it just be that your lens suffers from serious defraction issues when closing down below certain apertures? This can cause the image to go soft quite quickly and some lenses suffer more than others. You said you also had the same problem with your 350D - it wasn't using the same lens was it? If the problem is lens defraction obviously it will follow you from camera to camera as long as you stick with the lens. Do you get any different results if you stop up to say f8? Lens reviews often caution that dropping below f11-13 sees a substantial softening of the image.

I have to say, I had a quick look on photozone.de for the MTF results of that lens and it's okay but far from stellar. They give it 2.5 out of 5 for optical performance and it is knocked into a cocked hat by the standard 18-55 IS lens now being pushed out in kits. Sorry to be the bearer of such information! I don't actually like spreadsheets and judge on real world performance which often doesn't back up the lab tests (the MTF results for my Zeiss 16-80 and Tamron 17-50 are similar but there's just something about the Zeiss images that the Tamron can't quite match) but in your case it appears to letting you down in the real world too.

I guess the logical thing to do is borrow another lens and do some test shots compared to the 17-85. If it is much sharper the lens is your problem, if not, the problems lies elsewhere, most likely the camera body.

[mbird]

Thanks for your input guys. The problem is that I'm not sure if it is just me being over critical. The images look fine at normal forum resolution of 1600 wide, but when viewed in LR2 at 1:1 they are horribly soft and so would be no good for big enlargements and it also limits the amount of cropping I can do.

The lens is the same one I had on my 350D, Simon, but the same sort of problem is exhibited by my 70-300. I think I will contact WE and speak to them to see if there is some way the body/lens conficuration can be tested. This morning I've tried some test shots with IS disabled and all the noise reduction gizmos turned off, but the images still aren't crystal clear......

I've already tried the microfocus adjustments with a test print I found on t'internet, and the best focus is achieved with zero adjustment. I will have a go with larger apertures though, as that is something I didn't think of!

[Guest]

I have a standard 18 -55mm non IS lens that came with my 300D you can borrow if you like?

[pks1702]

It may not be a front/back focus issue Mark which is why I made reference to this. Out of interest when you tested did you put the lens out of focus before allowing the camera to re focus and take your test shot.

If you have any concerns do as I did get Canon to check it out under warranty. You will be without your camera for a week but they will thoroughly test it to ensure there are no issues.

I got mine back to Elstree but there is an approved Canon repairer in Colchester if you are ever down that way.

Colchester Camera Repair Services

Professional Camera, Lens, Professional Lens, Flash, Analogue Compact Camera, Digital Compact Camera, Analogue SLR Camera, Digital SLR Camera

6 Kings Court

Off Newcomen Way

Severalls Business Park

Colchester

CO4 9RA

Tel: 01206 843 322

Fax: 01206 843 323

Web: http://www.camera-repair.co.uk

Probably best though if you are still unhappy after testing to see what WE have to say

[mbird]

Simon, you are a bloody genius!

I tried the same two shots in my back garden with ISO set to auto, focus on auto, Av mode, focal length 72mm. The first was shot as f32, the second at f10. You can see from the screenshot what a major difference this makes to the clarity of the image......

I think my issues may be resolved, as I tend to shoot a lot with the aperture closed right down when I want a lot of dof. Maybe I need to restrain my fingers and stick to about f18 max.

[mbird]

Paul

Thanks for the kind offer. Maybe next time I bump into you I could just try a couple of test shots with one of your lenses?

Perry,

Thanks for that info. I think SImon may have put his finger on it, but I'll shoot a few more pics I think before I contact WE. I don't mind looking a wally on here, but if I take the camera into them and it turns out to be fine I'll look a plonker .

[Guest]

Looks like your lens has some serious defraction issues Mark but there's always a sweet spot beyond which things decline, often quite quickly. But what I can't figure out is that you say you've had this issue for some time yet I have always found your images to be absolutely pin sharp so either you are seeing things as worse than me or you are genius with Lightroom, in which case please come round and have a go with some of my images!

[Guest]

I'm sure WE will not make you look a plonker.

I bought a 2x extender on ebay just after Christmas. When I tried it on my 100-400 lens I didn't have any AF facilities.

So I took it to WE and told them that the AF didn't work.

They told me it doesn't work on that lens as the aperture isnt large enough, but they did check the extender out for me free of charge and said it was all A1.

I came out very pleased even if a bit red in the face

I normally have the two main lens with my on the boat Mark so you will be more than welcome to try them.

I can not use the -S type on the 5D otherwise I would try it for you.

[Guest]

Meant to say, have you looked at this site.

http://photography-on-the.net/forum/

Canon user forum?

[pks1702]

Not a plonker at all Mark

It is one of the beauty's of this area that issues can be posted up and views given. Those views may not be right but as in this case one just might

[pks1702]

They told me it doesn't work on that lens as the aperture isnt large enough, but they did check the extender out for me free of charge and said it was all A1.

I heard tales of taping the contacts to 'fool' the lens but I personally would not recommend it. The 1.4 x works fine though but will only work down to F4.

[mbird]

Looks like your lens has some serious defraction issues Mark but there's always a sweet spot beyond which things decline, often quite quickly. But what I can't figure out is that you say you've had this issue for some time yet I have always found your images to be absolutely pin sharp so either you are seeing things as worse than me or you are genius with Lightroom, in which case please come round and have a go with some of my images!

I think the main issue is that now I have the 50D, my expectations are much higher. WIth the 350D, the image blurred at a about the same magnification as the pixels appeared, but as the 50D is much higher resolution, it is more noticable. Also, the fact you have to reduce all pics to 1600 pixels for the forum does mean they look sharper at that size as you simply cant zoom in enough to see the problems. I think certain types of photo (like the "business" of all the bluebell stems) show the problem up, together with the shots with a smaller aperture exacerbating the problem.

Now I've read the reviews on the lens, it does seem to suffer from some image sharpness issues, and it's interesting to note that none of the test shots in the reviews are ever done with an aperture smaller than about f10.

[Guest]

Perry

"I heard tales of taping the contacts to 'fool' the lens"

Strangly enough,I did try that and it made no differance, hence the trip to WE.

They brought up some spec. on the Canon web site showing me which lens it would work as fully auto. and the 100-400 was not one of them.

Still it is something I dont use very offten unless on a tripod and at a fixed point.

[Guest]

Now I've read the reviews on the lens, it does seem to suffer from some image sharpness issues, and it's interesting to note that none of the test shots in the reviews are ever done with an aperture smaller than about f10.

If it was DP Review then take it with a pinch of salt. I have only limited time for their body reviews (the sample crops etc are very good but the conclusions that go with them often leave me thinking "are we looking at the same image" and they regularly contradict their own comments) but their lens reviews are are even more Leftfield. Somehow, in most cases, they manage to come up with results totally different from all the other review sites, which generally back each other up.

Your Canon lens is a case in point. The results the DPR obtained made it look simply dreadful, though the eventual conclusions did not seem to match the data results but still were not great. Moving over to photozone.de (which is a much better review site, IMO) had the lens returning figures much closer to what you'd expect from the decent mid-tier lens it is meant to be, at least in terms of sharpness though it appeared to show up other problems. So ignore DPR and go and read somewhere that knows how to test lenses! photozone.de still only gave the lens 2.5/5 mind you so perhaps it's not one of Canon's finest.

Btw - the current 18-55 IS that ships with current Canon bodies is a whole leap forward in quality over the older version which Paul has.

[Guest]

Just catching up as usual! You've had some great advice here Mark, and are well on the way to solving the problem.

When I first saw the initial screenshot, my immediate thoughts were (1) camera shake/tripod technique, (2) diffraction, (3) lens quality. Taking them in turn:

(1) Tripod technique - were you using a cable release or remote release, or the self-timer? At 1/6 sec. you need very good technique, and using a tripod does not assure good results. Any laziness will degrade the image - especially putting the camera on the tripod but still pressing the shutter button as normal ( yes, we've all done it to save time ). Was it windy? Did you use mirror lockup (if available) or live view (live view also stops mirror slap), and then wait for a few seconds after the last touch of the camera? As has been mentioned, IS needs to be off. If all this was OK, that pretty much rules out camera shake as a possible problem. BTW forgive me if you're doing all this religiously - I'm not suggesting you don't already know, just trying to set out a methodical testing path

(2) Diffraction - Simon has covered this very well, and your experiment has shown the importance of it. It can make a huge difference to image quality. I would recommend testing all your lenses using something like the following method. Camera on tripod, observing all points made above - if available, use mirror lockup, if not use live view. Point the camera at a suitable subject - I prefer something that consists of or includes a brick wall, because they are on a single plane if you line the camera up properly, and you can use the mortar lines to test for distortion, corner softness etc. Set to manual focus and focus using live view, zoomed in to make sure you are bang on. If a zoom lens, start at the widest focal length and take a series of shots at different apertures, going through the range methodically at whole stops - e.g. 2.8. 4.5, 5.6, 8, 11, 16, 22. I personally don't bother with half stops. On an SLR I wouldn't even try f32 - it will be unusable. Then test at other focal lengths - diffraction will be different at each length, and you need to know! Make sure the focus is on the wall that you are using as your primary testing point, irrespective of the focal length. With your lens I would go for a minimum of 17mm, 24mm, 35mm, 50mm and 85mm. Take the same series of shots at the same apertures. Load into Lightroom, and have to hand a wet towel to put over your head and a glass of something strong. You know what to do next - you should emerge from the exercise with a clear idea of the sweet spots (and c**p spots) at each focal length.

I would not shoot at f18 with an SLR - you are guaranteed degradtion through diffraction. As a rule of thumb I would try to avoid going above f11 for most lenses, but your tests will tell you the parameters for each lens. You will also know how much degradation you risk if you have to go outside the sweet spot, and where it becomes a waste of time to shoot.

This is not "pixel peeping" as practised on certain photo forums - it is getting to know how your actual kit works in the real world.

(3) Lens quality - test charts and reviews are fine, but shooting with them in real life is more important - which is why you need to test your lenses. Canon unfortunately are known for producing lenses which are not consistent, and you can get a duffer, including L lenses. I recently had to send back a 100-400 L which, when tested as above, turned out to be apallingly soft across the whole range. I thought I was seeing things, but took some of the same shots on my 70-200 f2.8 L with a 1.4x Mk II extender and compared them - it looked as if fog had been wiped off the images! Make no assumptions.

As you have realised, the problem of lens quality and diffraction is more apparent with bigger and better sensors. Faults that used to be unnoticable suddenly slap you in the face. It is even more of a problem with full frame hi-res sensors, i.e the 5D Mk II and the 1DS Mk III in the case of Canon. Some of the L lenses that were fine on smaller sensors suddenly show themselves to be underperforming and in need of a more modern update by Canon, the classic examples being the 17-40 L and the 100-400 L. Very careful technique is more important than ever with these highly revelaing sensors - any sloppiness or corner-cutting will produce a poor result. This includes hand-holding technique as well as tripod technique.

Noise reduction software effects won't be a problem at 100 ISO, and in any case if you're shooting RAW it won't be having any impact. Back or front focusing can be a problem, but can be ruled out when the whole image is soft as with yours.

When trying to trace lens/camera problems, it's important to be methodical. Once you have dealt with (1) and (2) above, if you still have the problem and it is happening with both your lenses, take up Paul's kind offer of sticking one of his lenses on your camera and doing some test shots - being careful of technique and using mid -range apertures. Also, if you can, try your lenses on another body and see if the same problem is exhibited. It's only when you've done all that that you might be left with the possibility that there is a camera problem, and the methodical testing should mean you know what the problem is. However, I'd be surprised if your camera has a problem - I can see enough from your postings to know that the bluebell shot was not typical of the results you're producing.

Good luck!

Bruce

[Guest]

Obvious I have nothing to add on lens testing after that fantastic effort from Bruce but here's a thought, and maybe not a good one. Is this the lens?

http://www.warehouseexpress.com/product ... ku=1002750

Only asking because that's as much as I paid for my Carl Zeiss 16-80mm which is probably the Alpha system equivalent lens and I have been impressed - stunned even by its quality so those comparisons don't look favourable. Thing is though, in the early days after the lens came on the market (late 2007 iirc) there were frequent complaints are sample variability. It took Sony some months to nail down the quality control on the new lens. Sensible people sent them to back to Sony and told them to sort the problem out but others just Ebayed them. Since Bruce has raised the issue of sample variability, and I believe yours was an Ebay purchase, could you have been the unlucky recipient of a dud being dumped by an unhappy owner? It's not a nice thought I know, but it might be worth thinking about if you can't get it sharp at any point during the tests Bruce has suggested.

It's odd that your 70-300 is the same though.

[pks1702]

Super effort Bruce I think all bases are covered

Plesbit wrote

Only asking because that's as much as I paid for my Carl Zeiss 16-80mm

The WE price is way over the top Simon

http://www.camerapricebuster.co.uk/prod28.html

Another review: http://www.the-digital-picture.com/Revi ... eview.aspx

[mbird]

SImon, yes that is the lens, but the cash I paid for the one I got from ebay was sub £400 (though not as low as Perry's findings). It was sold as brand new, but you could well be right and it is a slightly iffy quality version, but I'll have to try Bruce's test before I come to any conclusions. The reviews I was referring to were those on photozone.de and a couple of others. They seem to all think the lens is okay, but not outstanding, but I can't afford a better one just yet . Mind you the 17-55 f2.8 does look pretty good.......

Bruce,

What can I say , that must have taken a while to type, and I appreciate your efforts and help. Reading through what you have said, I will spend some time this weekend taking some methodical series of shots with both lenses. I naively thought a small aperture would make a better image for things like landscapes simply because of the pin-hole camera theory, but this is plainly not the case, even from my quick experiment earlier today.

I also take in everything you have said regarding tripod technique, and have to hold my hands up that mine is obviously poor. IS was turned on I did manually press the shutter button . I didn't use manual focus . I am therefore rapidly coming to the conclusion that operator error is probably the largest part of my problem.

I have had both lenses checked this afternoon by Norfolk Camera in Dereham, who tested them on their "columator" (dunno what the hell that is) and have stated they are what they would expect. The guy I spoke to did seem a little anti the 17-85 for the body I now have, but then he was a salesman afterall . I think this may well turn out to be a classic case of poor workman blaming his tools, so I'll run the methodical shots and see what happens.

[Guest]

Here we go Mark

Now we all KNOW what that is lol

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