Is this actually what I should expect from my Canon 70-200/2.8 IS II?

[Mitch.Conner]

EDITED FOR CLARITY 12/10/2014 00:33 EST.


Note (12/12/14), I've posted my own chart shots on pages 2 and 3 as per request, and have also posted additional images that may be relevant in answering this question.

The Digital Picture has crops of an ISO 12233 type chart shot via the 70-200/2.8 IS II. Link Here (pictures included inline below)

My chart shots look nowhere near that level of sharpness, especially at the corners. It does however look just like the chart shots taken by Martin Bailey and posted on his site (link here - images also posted inline with post below)

I'm just trying to figure out what I am supposed to be seeing, and whether my lens needs to go back to Canon again.

The Digital Picture unfortunately don't have a 5D3 body option, but I imagine my 5D3 should be as good as far as sharpness is concerned. The Martin Bailey pictures were shot on a 5D3 as far as I understand.

Here are the images from The Digital Picture

@70mm







@200mm








Here are the pictures (or rather animated gifs) posted by Martin Bailey on his site:



This is the part that's been bugging me as it looks perfect on the TDP chart shot, but this is much closer to what I'm seeing:



The Problem between the two: TheDigitalPicture shows me crisp clean corners at 70mm and 200mm, a tiny bit less so at 100mm and 115mm, but still WAY sharper than the gifs from the Martin Bailey Photography website, which show more or less what I'm seeing, which is very different from the crops TDP is displaying.

Which is closer to what I should be expecting?

 

[Khalai]

This lens is (supposed to be) very sharp. I am more than happy with my copy, probably sharpest lens in my bag along with 100L. Truthfully, I've never tested it on charts or diagrams, but from real scenarios, I can easily count eyelashes or hair on taken photo and they are tack sharp. I would send it for calibration IMO.

 

[jdramirez]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

 

[Mitch.Conner]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

My understanding is that live view bypasses the auto focus issues. It uses contrast based auto focus via data from the actual image sensor, so the usual auto focus system - the one responsible for auto focus problems - is not in play at all. In other words, the autofocus light path and image light path are identical with live view, which should avoid the split light paths that is at the heart of the kinds of autofocus issues resolved via micro adjustment.

Btw - since coming back a second time it does need micro adjustment (it's back focusing a tiny bit). However, I can avoid that in live view.

 

[Mitch.Conner]

This lens is (supposed to be) very sharp. I am more than happy with my copy, probably sharpest lens in my bag along with 100L. Truthfully, I've never tested it on charts or diagrams, but from real scenarios, I can easily count eyelashes or hair on taken photo and they are tack sharp. I would send it for calibration IMO.

I'm not sure what you mean by sending it for calibration. Is that a CPS service? I'm not a member. However Canon claims that both times I sent it in they tested to verify proper function (or so they claim). At least the first time, as per their own admission the results could not have been right, as it was out of spec when it went in almost immediately after receiving it from Canon.

Sending it in again would seem to be no guarantee that it will come back in spec, and as I learned the first time -could do more harm than good.

That's why I'm doing some asking around and research before I decide whether or not to box it up yet again.

 

[jdramirez]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

My understanding is that live view bypasses the auto focus issues. It uses contrast based auto focus via data from the actual image sensor, so the usual auto focus system - the one responsible for auto focus problems - is not in play at all. In other words, the autofocus light path and image light path are identical with live view, which should avoid the split light paths that is at the heart of the kinds of autofocus issues resolved via micro adjustment.

Btw - since coming back a second time it does need micro adjustment (it's back focusing a tiny bit). However, I can avoid that in live view.

What I'm saying is that if you manually focus using live view... you can digitally zoom into the image either 5x or 10x... and then you can make sure the target is in focus. Switch off live view... or leave it on, I don't think it would matter, and then hit the shutter, either with a 2 or 10 second delay, or use a shutter release.

So auto focus, as a variable is eliminated.

 

[Mitch.Conner]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

My understanding is that live view bypasses the auto focus issues. It uses contrast based auto focus via data from the actual image sensor, so the usual auto focus system - the one responsible for auto focus problems - is not in play at all. In other words, the autofocus light path and image light path are identical with live view, which should avoid the split light paths that is at the heart of the kinds of autofocus issues resolved via micro adjustment.

Btw - since coming back a second time it does need micro adjustment (it's back focusing a tiny bit). However, I can avoid that in live view.

What I'm saying is that if you manually focus using live view... you can digitally zoom into the image either 5x or 10x... and then you can make sure the target is in focus. Switch off live view... or leave it on, I don't think it would matter, and then hit the shutter, either with a 2 or 10 second delay, or use a shutter release.

So auto focus, as a variable is eliminated.

I digitally zoomed via live view to ensure that the center was in focus.

Really, the issue for me isn't the center.

The issue for me is this image, which is very similar to what I'm seeing:

Compare that to what The Digital Picture displays for the corner @70mm and @200mm:

In the very center - using Liveview - things are in focus and crisp, not quite like the center shot by The Digital Picture, but that may just be because they're using a photographic emulsion or photo printout, whereas my printout is on standard paper - but certainly the same as the center shot by Martin Bailey (@115mm):

I'm not opposed to trying again - but I'm pretty sure I'll get the same result. I've done the test several times just since receiving it. I can try to manual focus and see if that makes any difference.

 

[takesome1]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

My understanding is that live view bypasses the auto focus issues. It uses contrast based auto focus via data from the actual image sensor, so the usual auto focus system - the one responsible for auto focus problems - is not in play at all. In other words, the autofocus light path and image light path are identical with live view, which should avoid the split light paths that is at the heart of the kinds of autofocus issues resolved via micro adjustment.

Btw - since coming back a second time it does need micro adjustment (it's back focusing a tiny bit). However, I can avoid that in live view.

What I'm saying is that if you manually focus using live view... you can digitally zoom into the image either 5x or 10x... and then you can make sure the target is in focus. Switch off live view... or leave it on, I don't think it would matter, and then hit the shutter, either with a 2 or 10 second delay, or use a shutter release.

So auto focus, as a variable is eliminated.

+1 I read the OP and this is my first thought. You should be using live view MF for your comparison.

 

[East Wind Photography]

This lens is (supposed to be) very sharp. I am more than happy with my copy, probably sharpest lens in my bag along with 100L. Truthfully, I've never tested it on charts or diagrams, but from real scenarios, I can easily count eyelashes or hair on taken photo and they are tack sharp. I would send it for calibration IMO.

I'm not sure what you mean by sending it for calibration. Is that a CPS service? I'm not a member. However Canon claims that both times I sent it in they tested to verify proper function (or so they claim). At least the first time, as per their own admission the results could not have been right, as it was out of spec when it went in almost immediately after receiving it from Canon.

Sending it in again would seem to be no guarantee that it will come back in spec, and as I learned the first time -could do more harm than good.

That's why I'm doing some asking around and research before I decide whether or not to box it up yet again.

It looks like it just needs afma but short of that, it should come out tack sharp. Mine is so sharp that I can resolve dust on my lenscal scale.

If you end up sending it back to canon you should send it to a different facility. Which repair center did you send it to initially?

 

[neuroanatomist]

Had to look a bit to find an image of a basically flat subject at f/2.8 and low ISO. Shot is a wall in the Forbidden City, Beijing. Below are a couple of screenshots, the navigator pane shows the location of the 100% views, one near the center and the other at the extreme corner. Cobwebs in the extreme corner don't look significantly softer than those in the center.

 

[Mitch.Conner]

I never shot at targets... but I would suspect the lens should be on par. Have you manually focused using live view and then used a shutter release. I wouldn't use auto focus because it might need micro adjusted... might...

My understanding is that live view bypasses the auto focus issues. It uses contrast based auto focus via data from the actual image sensor, so the usual auto focus system - the one responsible for auto focus problems - is not in play at all. In other words, the autofocus light path and image light path are identical with live view, which should avoid the split light paths that is at the heart of the kinds of autofocus issues resolved via micro adjustment.

Btw - since coming back a second time it does need micro adjustment (it's back focusing a tiny bit). However, I can avoid that in live view.

What I'm saying is that if you manually focus using live view... you can digitally zoom into the image either 5x or 10x... and then you can make sure the target is in focus. Switch off live view... or leave it on, I don't think it would matter, and then hit the shutter, either with a 2 or 10 second delay, or use a shutter release.

So auto focus, as a variable is eliminated.

+1 I read the OP and this is my first thought. You should be using live view MF for your comparison.

Fair enough - but none of the images are mine in this thread. My question thus far is whether the photos posted by The Digital Picture, or those posted by Martin Bailey are closer to what this lens ordinarily delivers?

This lens is (supposed to be) very sharp. I am more than happy with my copy, probably sharpest lens in my bag along with 100L. Truthfully, I've never tested it on charts or diagrams, but from real scenarios, I can easily count eyelashes or hair on taken photo and they are tack sharp. I would send it for calibration IMO.

I'm not sure what you mean by sending it for calibration. Is that a CPS service? I'm not a member. However Canon claims that both times I sent it in they tested to verify proper function (or so they claim). At least the first time, as per their own admission the results could not have been right, as it was out of spec when it went in almost immediately after receiving it from Canon.

Sending it in again would seem to be no guarantee that it will come back in spec, and as I learned the first time -could do more harm than good.

That's why I'm doing some asking around and research before I decide whether or not to box it up yet again.

It looks like it just needs afma but short of that, it should come out tack sharp. Mine is so sharp that I can resolve dust on my lenscal scale.

If you end up sending it back to canon you should send it to a different facility. Which repair center did you send it to initially?

None of the images are mine, are you referring to the ones by The Digital Picture, or by Martin Bailey?

Had to look a bit to find an image of a basically flat subject at f/2.8 and low ISO. Shot is a wall in the Forbidden City, Beijing. Below are a couple of screenshots, the navigator pane shows the location of the 100% views, one near the center and the other at the extreme corner. Cobwebs in the extreme corner don't look significantly softer than those in the center.

The wall wasn't perfectly flat though right? It was a carving or sculpture? Where did you focus?


Note to all: I've edited the original post as there seems to be some confusion regarding the fact that none of the images were taken by me and that I'm trying to ascertain which set of chart photos are more like what you see with your lens. My chart photos look just like the ones posted by Martin Bailey - but neither his nor mine look like the ones posted by The Digital Picture.

 

[sdfreeland]

"Fair enough - but none of the images are mine in this thread. My question thus far is whether the photos posted by The Digital Picture, or those posted by Martin Bailey are closer to what this lens ordinarily delivers?"

I have the 70-200mm II and the 5d3. My experience has been more like the Digital Picture. Images are almost as sharp as my old 135L F2 at 2.8. I wouldn't be surprised if there is a loose element with Martin Bailey's lens. Just look at what Roger Cicala says about the lens. He's used practically every 70-200mm out there from every brand.

 

[Hjalmarg1]

Note to all: I've edited the original post as there seems to be some confusion regarding the fact that none of the images were taken by me and that I'm trying to ascertain which set of chart photos are more like what you see with your lens. My chart photos look just like the ones posted by Martin Bailey - but neither his nor mine look like the ones posted by The Digital Picture.

Michael, my experience is also very similar to the one on The-Digital-Picture. Images from my 70-200/2.8 (II) are as sharp as the images delivered by my 100L at all appertures from wide open. If any images isn't in focus is my fault, not the lens.
I bought it used and never experienced any problem with it.

 

[neuroanatomist]

The wall wasn't perfectly flat though right? It was a carving or sculpture? Where did you focus?

Basically bas relief decorative tile. Focused on the central flower, DoF was likely ~2". Essentially, I'm not seeing a big IQ drop from center to corner.

 

[Grummbeerbauer]

I used to do my own ISO chart shooting with new lenses, and mine also never looked as crisp as, e.g., those on TDP (I did that also with a 70-200 2.8 IS II that I got once, but I decided that it wasn't worth the extra weight and price over my 70-200 4.0 IS).
What I can tell you is that getting these test chart shots right is really tricky. The chart has to be exactly parallel to the sensor plane, otherwise your corners will be not in focus, in particular when using fast and/or telephoto lenses the DOF is razor-thin. I tried the "mirror trick", which got me probably closest to "exactly parallel" without using an optical bench. For that, you need to put a small mirror in the center of your chart (exactly parallel to the chart itself). Now point your camera at the mirror... once you see the lens front element in the center of the mirror, it should be aligned correctly.
Second, as others have said, for consistent results you need to use live view AF as phase detection AF is usually not consistent enough (and might need some AFMA).

BTW: In the end, it was never my test chart shots that made me keep or return a lens.

 

[Mitch.Conner]

Which repair center did you send it to initially?

Newport News, VA

Both times it went in it went there. I didn't pick it, Canon support told me to send it there. I've read mixed comments about it.

I used to do my own ISO chart shooting with new lenses, and mine also never looked as crisp as, e.g., those on TDP (I did that also with a 70-200 2.8 IS II that I got once, but I decided that it wasn't worth the extra weight and price over my 70-200 4.0 IS).
What I can tell you is that getting these test chart shots right is really tricky. The chart has to be exactly parallel to the sensor plane, otherwise your corners will be not in focus, in particular when using fast and/or telephoto lenses the DOF is razor-thin. I tried the "mirror trick", which got me probably closest to "exactly parallel" without using an optical bench. For that, you need to put a small mirror in the center of your chart (exactly parallel to the chart itself). Now point your camera at the mirror... once you see the lens front element in the center of the mirror, it should be aligned correctly.
Second, as others have said, for consistent results you need to use live view AF as phase detection AF is usually not consistent enough (and might need some AFMA).

BTW: In the end, it was never my test chart shots that made me keep or return a lens.

I tried the mirror trick for several of the chart tests too.

I've tried multiple methods to try to eliminate the possibility of user error as best I could.

 

[sagittariansrock]

I'm curious why you haven't attached any of your own images.
I know you're saying Martin Bailey's images are identical, but for the sake of completeness I'd expect to see your shots as well.

 

[Policar]

I never found my old 70-200mm f2.8 L to match the test charts, so I traded up to a brand new 70-200mm f2.8 II IS, but it's... not that much better. It is better wide open, less "weird," and possessing less character, but it's not in the same category as the Sigma 18-35mm f1.8 for tack-sharpness wide open...

It does seem to have good IS (for stills, not video as much) and great autofocus and build quality (so the the original) so I can't say it's a bad lens at all. The IS alone really made the difference, at least for stills, but it's no "miracle" lens as the 200mm f2 and Otus and recent Sigma Art lenses appear to be.

 

[Mitch.Conner]

I'm curious why you haven't attached any of your own images.
I know you're saying Martin Bailey's images are identical, but for the sake of completeness I'd expect to see your shots as well.

It's all part of my evil master plan... bwahahahahahahahaha.

Actually, it's just that I didn't see a point given that somebody else already had images using the exact same chart (printed similarly using the same vector graphic), camera, and lens combo. That.. plus maybe a little bit of laziness.

Here is a non chart image that I just took to examine this from a non-chart standpoint. It's just of a knock-down textured wall at 200mm.

I'll post a chart too at 200mm if you'd like. I don't have any at 70mm in LR right now for some reason. I might have deleted them (or maybe I just never imported them - not sure). I have a bunch at 200mm with various apertures, but I'm not sure where the other focal lengths are. Probably still on a CF card.

I'll also post the textured wall at 70mm in a minute.

 

[Mitch.Conner]

ISO 12233 Chart - 200mm @ f/2.8