What caused this depth-of-field anomaly?

The following was shot at 15mm, f/3.5 on my EF-S 15-85mm lens (this is jus a thumbnail, click to see the full-resolution image) with IS turned off.

HTTP response: NOT FOUND | MIME changed to 'image/png'

Well it's not the best shot to demonstrate it as clearly the subjects on the right side are further away than on the left. However, I had a similar issue with my 15-85 and exchanged it [replacement arrived yesterday, haven't had a chance to fully run it through its paces].

I recommend taking some test shots of a flat surface, perpendicular to the lens axis, ie a wall shot, to see if the right sided softness is still present.

Here's some sample shots of my 15-85 issue and the tests I did: http://www.flickr.com …0/sets/72157624​077505326/

ettlz wrote in post #10377364
The following was shot at 15mm, f/3.5 on my EF-S 15-85mm lens (this is jus a thumbnail, click to see the full-resolution image) with IS turned off.

HTTP response: NOT FOUND | MIME changed to 'image/png'

(Centre AF point.) Yes, yes, what am I doing shooting this at f/3.5? Well, everything should still be inside the depth-of-field. Yet somehow it looks like the depth-of-field swung around to the left (see the blurring on the right). What went wrong here? Is it normal for an IS lens to suddenly go all T-S like this?

take the exact same shot upside down and then rotate the image to determine in the is an issue with the right side of the lens

Electric Shepherd wrote in post #10377377
Well it's not the best shot to demonstrate it as clearly the subjects on the right side are further away than on the left. However, I had a similar issue with my 15-85 and exchanged it [replacement arrived yesterday, haven't had a chance to fully run it through its paces].

But the grass to the right of the metal fence (which is sharp on the left) is rather soft and "out of focus".

looks to me like some dust being kicked up in the parking lot area, its not OOF, just scattered dust in the air.

5DamienMkII's suggestion is an excellent one. Looking at your image it does seem to be softer on the right than the left at equivalent ranges, but it's hard to tell for sure. It mght also be worth trying the same shot with IS off, just to remove variables.

To be honest, I'm scratching my head on this one. It seems like whatever's getting thrown off is intermittent; I've got photos here, some of which are OK, some soft down the left, but if there's softness it's mostly down the right (including some taken upside down).

In pursuit of something more scientific, though, I rigged up some squares on a book-case. This is the "reference" image, taken with the 18-55 IS (on a tripod, with IS enabled; Mk.IV IS is supposed to detect this) at 18mm (processed in DPP to remove CA and vignetting):

HTTP response: NOT FOUND | MIME changed to 'image/png'

HTTP response: NOT FOUND | MIME changed to 'image/png'

HTTP response: NOT FOUND | MIME changed to 'image/png'

Just bear in mind that the DOF calculated is a measure of the depth of ACCEPTABLE sharpness, based on normal viewing size and distance. When you start pixel peeping, you are violating those conditions and the depth of acceptable sharpness could be a good bit shallower.

That sure looks like a de-centering issue in the first image.

ettlz wrote in post #10380737
To be honest, I'm scratching my head on this one. It seems like whatever's getting thrown off is intermittent; I've got photos here, some of which are OK, some soft down the left, but if there's softness it's mostly down the right (including some taken upside down).

In pursuit of something more scientific, though, I rigged up some squares on a book-case. This is the "reference" image, taken with the 18-55 IS (on a tripod, with IS enabled; Mk.IV IS is supposed to detect this) at 18mm (processed in DPP to remove CA and vignetting):

HTTP response: NOT FOUND | MIME changed to 'image/png'

To my eyes, this seems pretty much the same in all corners. (And not too bad, either.) Now this is the 15-85 at 15mm, wide open at f/3.5:

HTTP response: NOT FOUND | MIME changed to 'image/png'

and slightly more marginal here at 24mm, f/4.5:

HTTP response: NOT FOUND | MIME changed to 'image/png'

Seems like the bottom right is out of kilter.

I'm not sure if these account for the first image I posted (which was taken with IS off). In its defense, I will point out that the DoF at 15mm when focusing on something 30m away is basically 3m to infinity, so the entire image should be in focus. I can only guess that maybe the IS element isn't parking properly. Grr. CAAAANNNNNOOOOONNNN...

That's not a perfect setup, so you can't really conclude anything from this. Parts of the pieces of paper are not entirely flat, and neither is the camera aligned properly. The lower edge is generally less sharp than the top edge.

Neither can you conclude anything from the first picture. The wind seems to be playing in the tops of the trees, and the tree at the left middle isn't sharp either. And as someone else states, there seems to be some dust in the air on the right hand side.

To my eye there could possibly a slight decentering issue, mostly on the right hand side, towards the bottom, but you do need to do more rigorous tests to proves this.

I suggest you get a double spirit level to attach to the camera hot shoe, mount the camera on a high enough tripod, point it straight down and make sure the spirit level indicates it is perfectly aligned with the ground, put a newspaper down between the legs of the tripod, or your test images, in such a way that text or images cover the corners and edges of the frame, take a few photographs with a remote switch of the newspaper or images on the floor with mirror lock up engaged and at least 10 seconds between first click and second click, turn the newspaper or images around 180 degrees and do this again, and then compare the images. Of course, you also need to make sure the floor is perfectly level too.

Without any sophisticated equipment, this is really the only way to check properly for decentering.

HTH, kind regards, Wim

Tried the aligned looking down test, as suggested. Pretty much the same result: soft towards the bottom right.

Ah, well. Looks like it's going to get shipped to Canon tomorrow with a CD-ROM of test images...

It's just come back from Canon for warranty service. They claim that there is nothing wrong with it. No, Canon, it won't look wrong if your test is a laser print on A4 paper.

This is simply not good enough, I did not pay £600 for a lens should not be outperformed by a kit lens costing less than a quarter of that price. I'm going to contact the retailer now for another one or get my money back and do what I should've done in the first place --- buy the 17-40. Goods have to be fit for purpose, this lens ain't.

Suffice to say, I have lost faith in Canon.

Did you focus on the fence railing?

To me it appears as if you were tracking the silver car on the highway. Why do I say that? If you look at it at 100%, the license plate is almost perfectly focussed. Also, the silver car is in focus. In contrast, the red car in the center is out of focus. While not unequivocally from motion blur, it would go a long way in explaining the photo.

Perhaps you were unconsciously tracking the silver car while taking the photo, or perhaps you simply turned the camera as you pushed the shutter button. In any case, to me it seems as if you were panning toward the left when you took the photo.

Just my input...

On your first picture the camera is making focus on the fence, and it's normal because the fstop is f3.5. For a deep Depth of field (and make focus on most things) you need f8, f10, etc.