Is my lens soft??

Hi there,

I've got to ask. I have the feeling my efs 18-200 f3.5-5.6 IS lens was not as sharp as it once used to be on my 550D. But now that I'm playing with my EOS 80D I'm more and more convinced I have a problem. I know the efs 18-200 isn't regarded as a super sharp lens anyway, but I feel it used to do better.

I'm just not sure exactly what kind of problem I have. Am I just seeing things? Expecting too sharp an image from the camera and lens combo? Perhaps spoiled by the galleries on here for instance showing great and sharp shots in portrait and any other categories? Do I just need AFMA or is there really something wrong with my lens? Is it the better sensor on the 80D showing up faults that the 550D was hiding somewhat more?

I don't think the latter is the case. Pics from our last vacation with the 550D all seem to suffer the exact same issue. Sharp at first glance, but soft somehow. And when one zoomes in on eyes for instance the softness becomes more apparent. And when one has a play in photoshop with some sharpening, well then suddenly the softness is so obvious it's almost like the focus was off.

I'll include two example pictures. Pictures taken with central focus point on the eyes and confirmed in focus by the 80D. I'll include an out of camera and a photoshop sharpened image. By no means are these edits perfect as litterally they were only sharpened (on the forum they look a bit over sharpened, on my monitor in PS they look good), but they illustrate the difference and what I'd sort of expect out of camera instead after edit.

First one: 1/250 F16 ISO 400 @110mm (on purpose this time to make sure there would be plenty or even too much Depth of Field as I usually stick to the max aperture size at the chosen focal length for the most blurred background)

SOOC:

Can't really tell when you take an image and shrink it to fit for digital display. I 100% crop would be a better way to evaluate image quality. At first glance though, it looks pretty decent to me.

Glasses can throw AF off a bit. The second set looks sharper, and you should get optimal sharpness with crop at F8.
If you want sharper photos, try a good prime, like the 35F2IS or 85F1.8. Personally, I don't have a problem with the two photos.

The picture of the boy was taken at f/16 1/250 sec and the focus was on left eye. ISO was 400

The girl was taken at f/5.6 1/250 and an ISO of 160.

With apertures like f/16 there is always a possibility of introducing diffraction error. So try a series of test shots keeping the aperture down around f/5.6 to f/8. Make sure in this type testing that you don't use any filters.

Just because you used f16, that doesn't guarantee you have a wide DOF. For example, if your camera is back focusing, even 110mm at f16 depending on distance could only be about 6 inches, which is barely enough to cover an entire child's head, and you could still see issues due to focusing offsets.

Also per the child photo, the hair would have created a higher contrast point for the AF system then the glasses/eyes that were in the shadows, so that makes it difficult too.

I would do some targeted test shots to test AF to see if you are indeed seeing that the system is "focusing" on a point, but when you take the shot, you find that some other area before or after the object you focused on is really in focus, indicating front or back focus issues.

gjl711 wrote in post #18473834
Can't really tell when you take an image and shrink it to fit for digital display. I 100% crop would be a better way to evaluate image quality. At first glance though, it looks pretty decent to me.

I'll try and do that next time. Probably after some more tests as proposed.

artyH wrote in post #18473836
Glasses can throw AF off a bit. The second set looks sharper, and you should get optimal sharpness with crop at F8.
If you want sharper photos, try a good prime, like the 35F2IS or 85F1.8. Personally, I don't have a problem with the two photos.

I have the same issue with my daughter who's not wearing glasses or lenses... Front/back focussing sounds like something to try and investigate further.

As for fast primes, the 85mm f1.8 is on the short list. But I'm trying to go for the 70-200f2.8 L IS first as I think it's suitable for both portrait as well as other genres of photography.

John from PA wrote in post #18473837
The picture of the boy was taken at f/16 1/250 sec and the focus was on left eye. ISO was 400

The girl was taken at f/5.6 1/250 and an ISO of 160.

With apertures like f/16 there is always a possibility of introducing diffraction error. So try a series of test shots keeping the aperture down around f/5.6 to f/8. Make sure in this type testing that you don't use any filters.

The F16 in this shot of my son (the girl is the wife) was an exception. I usually stick to the f4.5-f8 range depending on focal length.

I have a UV filter on the lens for protection. A habit from the 35mm days... Is that a factor in this?

TeamSpeed wrote in post #18473840
Just because you used f16, that doesn't guarantee you have a wide DOF. For example, if your camera is back focusing, even 110mm at f16 depending on distance could only be about 6 inches, which is barely enough to cover an entire child's head, and you could still see issues due to focusing offsets.

Also per the child photo, the hair would have created a higher contrast point for the AF system then the glasses/eyes that were in the shadows, so that makes it difficult too.

I would do some targeted test shots to test AF to see if you are indeed seeing that the system is "focusing" on a point, but when you take the shot, you find that some other area before or after the object you focused on is really in focus, indicating front or back focus issues.

Any suggestions for taking these test shots? Convenient subject/situation(s)?

The diffraction limited aperture of the 80D is f/5.9. Using any smaller aperture (larger number) is going to create increasingly softer images, even with the best lenses in the world. Add to that the fact that the 18-200 is one of (THE ONE?) worst EF-s lenses Canon has ever made, and you've got some issues to deal with.

Better lenses will help. Larger apertures will help. The 80D is a fantastically good camera. You are not giving it much to work with.

If you want to test the lens, the best way is to shoot a resolution chart such as the ISO1223 chart. You can load it for free from HERE. Set out the chart at the distance you want to test.
Tripod mount the camera,
Place it in MLU and 10 second timer.
First shot, focus using live view zoomed in 10x then take the shot with MLU/timer.
This will establish your baseline, the maximum performance the lens can deliver.
Then switch over to AF and take a few more de-focusing the lens each time.
Compare to your baseline.

When you look at the image, you can see at what point the image quality starts to fall apart. Below is an image I have for my 100-400. You can clearly see that at 400mm with zero MFA, the best the lens can resolve is a bit better than 7. When properly MFAed, even at 10 you can see the individual lines. End result is that the images are much sharper than before. Would I like more sharpness, sure, but at least I'm confident that the lens is delivering the best it is able to.

Bassat wrote in post #18473851
The diffraction limited aperture of the 80D is f/5.9. Using any smaller aperture (larger number) is going to create increasingly softer images, even with the best lenses in the world. Add to that the fact that the 18-200 is one of (THE ONE?) worst EF-s lenses Canon has ever made, and you've got some issues to deal with.

Better lenses will help. Larger apertures will help. The 80D is a fantastically good camera. You are not giving it much to work with.

Thanks for that. As stated, I realise the reputation of the efs 18-200. But mine seems to have gotten worse. Also currently I have no other lenses to change and compare.

I'm contemplating several options on that front:

The 18-135 usm nano as standard lens supplemented by the 70-200 L f2.8 and later also the 1.4 extender. Both have good if not great reps. But there is considerable overlap.
Or perhaps something like the 15-85 usm supplemented by the same 70-200 L f2.8.

Thanks, I'll find the time to do that sometime this week. I'm very curious to find out if the lens can perform better than it is doing now...

Eurogranada wrote in post #18473849
Any suggestions for taking these test shots? Convenient subject/situation(s)?

I would remove the filter, then find a fence row, or use your pavement or low cut grass as your testbed. Put something small but brightly colored like a kids toy on any of the 3 items. Stand back, open up your aperture as much as you can, and once you just cover the center AF point, you are far enough back. Take some test shots focused on that toy. Zoom in and look to see where the grass, concrete or wood of the fence is in the most focus in correlation to the test object.

Eurogranada wrote in post #18473860
Thanks for that. As stated, I realise the reputation of the efs 18-200. But mine seems to have gotten worse. Also currently I have no other lenses to change and compare.

I'm contemplating several options on that front:

The 18-135 usm nano as standard lens supplemented by the 70-200 L f2.8 and later also the 1.4 extender. Both have good if not great reps. But there is considerable overlap.
Or perhaps something like the 15-85 usm supplemented by the same 70-200 L f2.8.

You mentioned the 18-135 USM. I've had that lens for about 2 weeks. It is surprisingly good in two regards. Optically, it is really good. And the auto-focus is every bits as quick and accurate as my 70-200 f/4L IS.

Take off the UV filter. Vent some of your current frustrations by taking a hammer to it. Or see how many times you can skip it across a local lake. Even the good ones are pointless.