70-200L autofocus on center vs. peripheral focus points

[photennek]

Hi,

I have the Canon EF 70-200 f/2.8 L IS ii USM lens and use it with my 5d3. I have tried my best to AFMA it and I am happy with its focus accuracy when using the center focus point in the single spot focus point -mode. However, I am noticing (more and more consistently) that @ f/2.8 when using peripheral focus points (for example the second column of focus points from either left or right side when in landscape orientation), the focus is often not correct. Mostly it is front focusing, though I have not made exact measurements. Is this typical to the lens, or should I send it in for some service?

And, is this more typically a lens or a body problem? I am not noticing the same with for example 24-70Lii or 100L macro, but I am not sure if there is a bit of this same with 35f2.

I've now learned to use more focus and recompose -technique with the 70-200, which is fine for stills and gives me good results, but for less still subjects peripheral focus points would be useful for nicer compositions.

Anyone having similar experiences, on this lens or another, or peripheral focus points in general? All my experiences are mainly in good light situations.

Thanks for any insight!

 

[Maximilian]

Re: 70

Hi photennek,

did you read the manual, esp. the chapter 3 about the AF?

If so, than you could have in mind that the 61 AF points are designed in different ways and that the center column is the best one for bright lenses, e.g. your mentioned f2.8 zoom. This could be the reason for the behavior you've recognized.

Not a desirable feature but a feature of the 5D3 AF system.

Edit:
Quote from page 79 about the inner 5 AF points of the central column:
"Dual cross-type AF point(s). Subject tracking is superior and the focusing precision is higher than with other AF points."

 

[Random Orbits]

It might be worth sending in both the camera and the lens in to be looked at. I use the outer points (esp. in landscape) for soccer in AI servo and portraits (landscape and portrait), and I have a good hit rate. The outermost points aren't as accurate as the points one column in, but it works much better for me than what you're describing.

 

[Mt Spokane Photography]

It can definitely be a camera issue, and if a lens is out of alignment, that can cause issues as well.

FoCal has a test for each AF point that checks its accuracy. It can be as simple as a spec of dust on the AF sensor to something not being properly aligned. Canon can fix it.

 

[Focus Studio]

I am having the same focusing problems with 5DIII and 70-200 2.8 IS LII. Tried another 70-200 2.8 IS LII on the same body - the same results. The succes rate for 24-70 2.8 LII with 5DIII is much better, in low light.

Now, I use the lens (70-200) exclusevily on 1DXII (also, almost exclusevily with servo focus and back-button focusing) - the hit rate is much better, but it also misses from time to time, by a little bit. I have to mention, though, that this behavior appears, mostly, in low light, at wedding receptions. Also, 1DXII acquires focus much-much faster in light levels where 5DIII would hunt, at best.

 

[photennek]

Re: 70

Hi photennek,

did you read the manual, esp. the chapter 3 about the AF?

If so, than you could have in mind that the 61 AF points are designed in different ways and that the center column is the best one for bright lenses, e.g. your mentioned f2.8 zoom. This could be the reason for the behavior you've recognized.

Not a desirable feature but a feature of the 5D3 AF system.

Edit:
Quote from page 79 about the inner 5 AF points of the central column:
"Dual cross-type AF point(s). Subject tracking is superior and the focusing precision is higher than with other AF points."

Maximilian, good point to refresh what the manual says. Yes it says the center points (those 5 dual cross-type AF points) are the best in accuracy, and I totally agree to that, those have the best hit rate.

Still I'm unsure if that is the sole reason for my 70-200 lens behaviour... For some two years I've used the 24-70 and 100L macro quite successfully also with those more peripheral (just normal) cross type AF points, and they also nail focus quite well, albeit not as perfectly, but I am confident in using them too. Now I am losing that confidence with the 70-200 on those peripheral normal cross type AF points, and I am slightly confused about what is the reason for this behaviour. Many (including me) regard this lens very highly, especially for its image quality, but also overall, and I've at least personally included AF accuracy into that overall.

Maybe this is just my first time I am finding 5d3 autofocus limitations (consistently)...

It can definitely be a camera issue, and if a lens is out of alignment, that can cause issues as well.

FoCal has a test for each AF point that checks its accuracy. It can be as simple as a spec of dust on the AF sensor to something not being properly aligned. Canon can fix it.

Thanks for the tip! I wasn't aware that FoCal has such functionality. Until now I've done my AFMA without it. I just went to get the software, I'll try it at home and see what results it gives me. And likely post here some results as well for you to comment.

I am having the same focusing problems with 5DIII and 70-200 2.8 IS LII. Tried another 70-200 2.8 IS LII on the same body - the same results. The succes rate for 24-70 2.8 LII with 5DIII is much better, in low light.

Now, I use the lens (70-200) exclusevily on 1DXII (also, almost exclusevily with servo focus and back-button focusing) - the hit rate is much better, but it also misses from time to time, by a little bit. I have to mention, though, that this behavior appears, mostly, in low light, at wedding receptions. Also, 1DXII acquires focus much-much faster in light levels where 5DIII would hunt, at best.

Focus Studio, nice to know you have similar experiences. Based on what you say it is not the lens alone, as it behaves differently on different bodies.

So maybe I'll end up sending those in for service, but now first I want to check the FoCal results.

If I send the body in for service, does anyone know if Canon can adjust the 5D3 body for this lens only? As I said I have lenses that work great with the body, I wouldn't want to mess their AF accuracy...

Thanks for your input already!

 

[AE-1Burnham]

WOW! Good news indeed.
...I live in Norway (originally from USA) and although I hate to knock this lovely country (although this also applies to most experience in Europe that I have had excepting Germany), one thing that is not done well here is the service industry in general. "Similar serial numbers" and yeah, duh, the lenses do look a lot alike so having a good system and good procedures would maybe ne necessary...
Happy shooting!

 

[Maximilian]

Re: 70

...
I wasn't aware that FoCal has such functionality. Until now I've done my AFMA without it. I just went to get the software, I'll try it at home and see what results it gives me. And likely post here some results as well for you to comment.

Hi photennek!

If you're starting to get used to FoCal please note that there are some really good threads here about how to use it properly.
Especially a solid ground (no vibration) and (really) a lot of light on the target (FoCal seems to love bright target) seem to be the key points.

Good luck.

 

[ksgal]

something else to consider.. if the lens is becoming a bit de-centered (micro movement of adjustment spacing/parts wearing out over time) this may also look like miss focus? Possibly? Maybe it can't focus properly or as well in that case? If it was in one particular corner? I'd also make sure it was at least a cross point you are using.. horizontal or vertical only points can be near useless in low light.