Shot wedding with 5DIII, dissapointed in AF

[Mr Simpleton]

Interesting. it would be interesting to try to see how the results from the Mk3 compared when forced to fire after the same time as the 5D2 had. One would hope it would do at least as well in terms of hit rate.

Agree it seems like fun test, but that means controlled lighning and such... right now not feasable here. I'm sure some with the two cams can do this.

Still the "wait for confirmation" is buried deep into my spine... never ever mash the shutter. Time to re-learn?

 

[Z]

First up - I'm not a 5D3 or 5D2 shooter - I use a 7D so take my opinions with a grain of salt. However, a lot of the autofocus features that are 'new' to 5D3 users (AF point expansion, zone AF etc.) who have upgraded from the 5D2 are features that 7D users have been playing with for years. Personally, the only time I use single AF point expansion, as OP did, is for AI Servo tracking of [fast] moving subjects. In this scenario it reduces the odds of the subject moving outside the main AF point and causing the focus to hunt. For this purpose I find it particularly effective. If I were using One Shot (as OP was), a single autofocus point always suffices for me and I would never turn on point expansion for fear of overcomplicating things. As others have stated, I would first try turning off the expansion in that scenario.

The 7D manual at least agrees with me, describing AF point expansion thus:

The manually-selected AF point <S> and adjacent AF points <w> are used to focus. Effective when it is difficult to track a moving subject with just one AF point.

The manual describes how AF point expansion works in One Shot mode, but doesn't state why it would be useful. I don't mean to add to the tirade of hysterical "5D3 is awesome, you must be doing it wrong!!1" (after all it wouldn't be the first time Canon has screwed up an autofocus system) or to tell you your business, but it is a new piece of kit and teething problems are to be expected.

 

[Kernuak]

You would think something like that would be mentioned in all those reviews, or in the documentation, but I haven't seen anything like that.

The problem is, most of the reviews spend all their time shooting still life setups in their studio, instead of using the camera in real life situations. They simply don't spend the time getting to grips with things like the AF system, so can't really give a good indication of what it is like, let alone getting the best out of it in various situations.
I'll try to find the CPS article on the MkIII AF to see if it sheds any light.

 

[ew20]

I shot some surfing this weekend, just trying to get the hang of the 5D3. Servo mode is amazing with the expanded points. Because the motion is so erratic, I usually get about 40-50% in very good focus. With the 5D3, that jumped to 90%. It was hard to find shots that weren't in focus.

Right after, I swapped to one-shot to take a picture of the wife, and like some others here the focus confirmation seemed slow. There was plenty of light, it just took an extra second for the confirmation to flash. I was expecting pretty much immediate focus so I accidentally took a shot before confirmation, then took another right after to make sure I had a good image. What was interesting is the focus on the two images was practically identical, I couldn't tell them apart.

I did have point expansion on and used the points on the left hand side of the frame for both images. I'm wondering if maybe the shot does focus quickly, but the confirmation is lagging behind. Or perhaps the central point you choose focuses quickly, but the AF system takes a little time to check the surrounding points to make sure everything is fine and dandy before giving confirmation. I'll try with expansion off later and see if anything comes of it.

 

[Tim Larsen]

I just retested this 5dII vs 5dIII w/ the 70-200 f2.8 II (the 35 and 85 are at the studio and I'm at home). The 5dII and 5dIII were extremely in focus speed at ISO1600, 1/6, 2.8.

This is interesting to me, so I'd like to hear more experiences.

 

[Kernuak]

The article I was thinking of was actually for the 1D X, not the 5D MkIII and isn't specific, but it states that you shouldn't use single point spot for low light (which is irrelevant to this thread and obvious to most of us anyway) and that point expansion is to improve tracking. Reading between the lines, it suggests that point expansion shouldn't be used for slow moving/stationary targets. So while it isn't clear, it's suggestive that single point is the best option for low light. Of course, whether that makes it it any quicker isn't guaranteed.

 

[epsiloneri]

The 5dII and 5dIII were extremely in focus speed at ISO1600, 1/6, 2.8.

Hm, a critical word is missing. Is it "similar"?

 

[shizam1]

The article I was thinking of was actually for the 1D X, not the 5D MkIII and isn't specific, but it states that you shouldn't use single point spot for low light (which is irrelevant to this thread and obvious to most of us anyway) and that point expansion is to improve tracking. Reading between the lines, it suggests that point expansion shouldn't be used for slow moving/stationary targets. So while it isn't clear, it's suggestive that single point is the best option for low light. Of course, whether that makes it it any quicker isn't guaranteed.

I think I read that as well, but my take was that the extra points wouldn't hurt, and would indeed help in case you landed on a low contrast area with the center, and then it would look with the extra points around the point you chose.

 

[Bosman]

I shot some surfing this weekend, just trying to get the hang of the 5D3. Servo mode is amazing with the expanded points. Because the motion is so erratic, I usually get about 40-50% in very good focus. With the 5D3, that jumped to 90%. It was hard to find shots that weren't in focus.

Right after, I swapped to one-shot to take a picture of the wife, and like some others here the focus confirmation seemed slow. There was plenty of light, it just took an extra second for the confirmation to flash. I was expecting pretty much immediate focus so I accidentally took a shot before confirmation, then took another right after to make sure I had a good image. What was interesting is the focus on the two images was practically identical, I couldn't tell them apart.

I did have point expansion on and used the points on the left hand side of the frame for both images. I'm wondering if maybe the shot does focus quickly, but the confirmation is lagging behind. Or perhaps the central point you choose focuses quickly, but the AF system takes a little time to check the surrounding points to make sure everything is fine and dandy before giving confirmation. I'll try with expansion off later and see if anything comes of it.

I think that is something to look at, the lag thing notifying of in focus is sometimes not on when it shoots and i have it set to not take it til focus lock is achieved. this may be a bug in fact, since i have noticed that as well but didn't find anything out of focus so i didn't worry about it. If the camera takes the shot before notifying me and its in focus i guess if you are going to have problems then this isn't a bad one.

 

[Bosman]

The article I was thinking of was actually for the 1D X, not the 5D MkIII and isn't specific, but it states that you shouldn't use single point spot for low light (which is irrelevant to this thread and obvious to most of us anyway) and that point expansion is to improve tracking. Reading between the lines, it suggests that point expansion shouldn't be used for slow moving/stationary targets. So while it isn't clear, it's suggestive that single point is the best option for low light. Of course, whether that makes it it any quicker isn't guaranteed.

I think I read that as well, but my take was that the extra points wouldn't hurt, and would indeed help in case you landed on a low contrast area with the center, and then it would look with the extra points around the point you chose.

Which by all means shouldn't be as fast as one point doing the measurements i would think. Chuck Westfall says the 5dm3 won't focus as fast as the 1d because the voltage power of the battery system is no where near what a 1d battery is.

 

[eeek]

Is this the one you are talking about:

http://learn.usa.canon.com/resources/articles/2011/1dx_af_pts_article.shtml

edit: this is for the 1DX, but it is helpful as well:

http://www.learn.usa.canon.com/resources/articles/2012/1dx_guidebook.shtml?categoryId=12

 

[Bosman]

The article I was thinking of was actually for the 1D X, not the 5D MkIII and isn't specific, but it states that you shouldn't use single point spot for low light (which is irrelevant to this thread and obvious to most of us anyway) and that point expansion is to improve tracking. Reading between the lines, it suggests that point expansion shouldn't be used for slow moving/stationary targets. So while it isn't clear, it's suggestive that single point is the best option for low light. Of course, whether that makes it it any quicker isn't guaranteed.

That is helpful information. If you find the link back that would be sweet.
This link has good info at the Canon learning center.
http://www.learn.usa.canon.com/dlc/search/search.spr?keyword=Rudy%20Winston&filterBy=Article

 

[Tim Larsen]

The 5dII and 5dIII were extremely in focus speed at ISO1600, 1/6, 2.8.

Hm, a critical word is missing. Is it "similar"?

yes, similar!

 

[Loswr]

edit: this is for the 1DX, but it is helpful as well:

http://www.learn.usa.canon.com/resources/articles/2012/1dx_guidebook.shtml?categoryId=12

Thanks - that's a very helpful document!

 

[SeanNY]

Strange... I was shooting a concert where I had to use around 12800 - 25600 iso and had it straight on center point AF and it was focusing amazingly (in that kind of situation my 5d2 wouldn't even try). In what situations did you try the expanded? Not trying to doubt you but it does focus much differently than the 5d2 when you put it on the expanded mode and if you're not used to it then it might have caused you problems.

If you hadn't already written that, I would have written the exact same thing ... except that I didn't even set center point AF yet.

I shot a concert on Friday night with my just-out-of-the-box Mark III with my favorite 85mm F1.2L lens. I have done the exact same thing many times before with my Mark II and I can say for certain the focusing was much quicker. I would go so far as to describe it as snappy.

I was mostly shooting at 1/160th - 1/400th of a second, aperature 2.8 - 3.4 and ISO up to 12,800.

In the interest of full disclosure: The only setting I changed on the camera before the concert was that I selected only cross type AF points because I thought that would mean only cross type AF points would be automatically used (I thought this might make focusing faster). But reading the manual later, I think that just means only cross type AF points will be manually selectable. Since I didn't do any manual selection, therefore, I think this setting was irrelevant for me.

Here is a typical shot from that concert (at 12,800 ISO!!):

http://www.flickr.com/photos/edwinland/6886762262/#in/photostream

It may look like the subject was posing for me but he was actually moving forward from the back of the stage while playing his guitar, and I managed to catch him the one half second he glanced directly at my camera.

 

[shizam1]

I just did some shots in my basement, lighting was dark, ISO 1600 f/2.8 and 3 second shutter speed.

It would take 2-3 seconds to acquire focus, and changing between expansion mode and not didn't do anything.

So at least for my camera, in really dark scenes, changing that setting won't give better AF performance.

 

[Bosman]

I just did some shots in my basement, lighting was dark, ISO 1600 f/2.8 and 3 second shutter speed.

It would take 2-3 seconds to acquire focus, and changing between expansion mode and not didn't do anything.

So at least for my camera, in really dark scenes, changing that setting won't give better AF performance.

now try it with a 70-200 instead of an 85mm F1.2 or whatever you used.

 

[shizam1]

I just did some shots in my basement, lighting was dark, ISO 1600 f/2.8 and 3 second shutter speed.

It would take 2-3 seconds to acquire focus, and changing between expansion mode and not didn't do anything.

So at least for my camera, in really dark scenes, changing that setting won't give better AF performance.

now try it with a 70-200 instead of an 85mm F1.2 or whatever you used.

I actually used my 70-200 f/2.8L II

 

[Bosman]

I just did some shots in my basement, lighting was dark, ISO 1600 f/2.8 and 3 second shutter speed.

It would take 2-3 seconds to acquire focus, and changing between expansion mode and not didn't do anything.

So at least for my camera, in really dark scenes, changing that setting won't give better AF performance.

now try it with a 70-200 instead of an 85mm F1.2 or whatever you used.

I actually used my 70-200 f/2.8L II

DOH! That sucks man, what are you going to do? Do you have a 1d to test against it?
Wait...3 sec shutter speed? I wonder if that has a bearing on anything...just hinking im going to try some tests myself now.

 

[altenae]

How dark is 2.8 @1600 ISO with 3 seconds exposure ??
Just a question, but what do we expect from the AF with this kind of darkness ??