Gert-Jan Maas wrote:
[...]
In other words, focus center point with AI-servo, move out of center (camera tracks to outer point), take a shot.
[...]
IIRC if you have all focus point active (ring of fire) then the AF will begin searching with the centre point, if you lose AF lock with the centre point the AF will then activate the outer points as required - in theory.
I never shot with like this, I always select my point manually and enable AF expand as I need it. For birds in flight ring of fire can work well but you need to ensure your DoF is large enough to ensure focus placement/selection errors.
Alistair is right about the theory, but I find it does not really work tha way in practice. One of the AF issues that has been documented with the MIII is its tendency to unexpectedly move off the desired subject. I notice this quite often. I am not sure if this is a product defect, or the AF is working in some way that I don't fully understand.
It seems I have to work a lot harder to keep the subject over the center point than with prior 1D models.
If you use center point only, the camera will attempt to focus on the bg or fg when the af point leaves the subject. This will occur after a delay. The length of the delay depends on the tracking sensitivity setting. Tracking sensitivity may be set on pro bodies, but non pro bodies have a fixed setting.
When RG wrote his second article on the MIII (the testing he did with Canon), he said Canon recommended setting the micro adjustment to -1 for all lenses. I had this set for a while, but changed it back to zero recently. With it set to zero, I think my MIII is more prone to front focusing in AI Servo when the subject is moving toward me. After my shoot this past weekend, I moved it back to -1.
The maddening thing about my example. There were 3 frames shot.. All standing still, no M/A and in ONE-SHOT (not servo). Each time, I recomposed, focused and shot (ie, didn't burst). All look very similar to this. Some things to note (not that it matters)... I shot with a 580EX-II and it was the first shots with the 16-35 on that body and the first 3 shots of the day. A few minutes later...inside...it worked fine with the same combo.
apdieb wrote:
The maddening thing about my example. There were 3 frames shot.. All standing still, no M/A and in ONE-SHOT (not servo). Each time, I recomposed, focused and shot (ie, didn't burst). All look very similar to this. Some things to note (not that it matters)... I shot with a 580EX-II and it was the first shots with the 16-35 on that body and the first 3 shots of the day. A few minutes later...inside...it worked fine with the same combo.
Where did you place the AF sensor? On top of the face or at the edges? If you put it over the face, did the AF point completely cover the face? How far away were you? Is this a cropped image? What AF point did you use?
Top Center AF point Squarely on the face. Beeped/solid dot focus confirmation. The first image is no crop, the others are crops from that same image. AF point covered his face (mostly, if I remember correctly). There was no focus and recompose either. Shot at 23mm... I would say I was 6 or so feet away. Shot with a 16-35L on MK3.
For the record, I understand that using the edges seems to help... But if I cannot point an AF point at a person's face and take a picture, there is something grossly wrong. Luckily, it's not all the time that this happens.. But enough to really Pi$$ me off.
Never EVER had this problem with any of my (2) 20D's, (2) 5D's, or Mark IIN...
Your OOF shots are the typical shots when the mkIII AF got fooled and doesn't know what to do. Just like a kind of default mode (severe back or front focus or nothing in focus at all). Possible explanations:
1) Rather busy background - could have fooled AF
2) Difficult lighting - could have fooled AF
3) Low contrast on the face (the AF point covered the entire face) - could have fooled AF
4) The mkIII AF (or any AF for that matter) can't handle field curvature that well. Meaning less accurate outer AF point performance with wide angle focal lengths and large(r) apertures - could have fooled AF
Aiming for the side of his head probably would have given an in-focus shot. Using the center AF point under these conditions probably would have given a better result also.
I know... all the other DSLR's behaved differently and probably wouldn't have had any trouble nailing this shot. But that's like comparing apples to oranges and not very productive. The point is, the mkIII AF requires a different approach under certain conditions. The sooner you are aware of that, the better results you are going to get
Sure.. But I think asking someone not to be able to take a snapshot from 6ft away of their kid when the AF confirmation beeps and it APPEARED in focus through the viewfinder is a bit far fetched. I use all of my bodies professionally and this isn't an acceptable "Difference". It is a problem plain and simple. If the AF gets fooled, then it certainly shouldn't confirm that it is in focus when actually nothing in the frame is focused. The extreme bottom right of the frame has some marginally sharp blades of grass, but certainly not what I'd consider in focus.
Also. The fact that I shot 3 separate frames, probably 15 or 20 seconds apart, completely reframing and focusing shouldn't all be this far off.
What is also disheartening is the fact the the plane of focus seems off as well. The right side of the frame is much sharper (albeit, still soft) compared to the left side.
I have been using my MKIII since July of 2007. I did not have any problems at all until AFTER the submirror "FIX". It has been back since too. Like I said, it's not all the time, but I certainly have used it enough "to be aware of different approaches" to know when my camera is malfunctioning (and this is in ONE-SHOT not even SERVO).
I choose not to candy coat it by saying it requires different approaches. I choose to say that it grossly fouls up what should be a normal shooting condition on occasion. Does that make me hate the M3? No. But it frustrates the heck out of me when it spoils a key moment.
I don't think it's incorrect of me to expect that my $4500 camera with $1400 lens should be able to take a shot outside at 8am from 6ft away and get at least something in the frame in focus.
apdieb wrote:
Sure.. But I think asking someone not to be able to take a snapshot from 6ft away of their kid when the AF confirmation beeps and it APPEARED in focus through the viewfinder is a bit far fetched. I use all of my bodies professionally and this isn't an acceptable "Difference". It is a problem plain and simple. If the AF gets fooled, then it certainly shouldn't confirm that it is in focus when actually nothing in the frame is focused. The extreme bottom right of the frame has some marginally sharp blades of grass, but certainly not what I'd consider in focus.
Also. The fact that I shot 3 separate frames, probably 15 or 20 seconds apart, completely reframing and focusing shouldn't all be this far off.
What is also disheartening is the fact the the plane of focus seems off as well. The right side of the frame is much sharper (albeit, still soft) compared to the left side.
I have been using my MKIII since July of 2007. I did not have any problems at all until AFTER the submirror "FIX". It has been back since too. Like I said, it's not all the time, but I certainly have used it enough "to be aware of different approaches" to know when my camera is malfunctioning (and this is in ONE-SHOT not even SERVO).
I choose not to candy coat it by saying it requires different approaches. I choose to say that it grossly fouls up what should be a normal shooting condition on occasion. Does that make me hate the M3? No. But it frustrates the heck out of me when it spoils a key moment.
I don't think it's incorrect of me to expect that my $4500 camera with $1400 lens should be able to take a shot outside at 8am from 6ft away and get at least something in the frame in focus....Show more →
Of course you are free to expect all that you want. Obviously, your mkIII isn't going to give it to you like that (under these conditions). I simply gave some possible explanations for the OOF shots based on my own findings after extensive use of a few 1Ds3's. You can ignore them of course. OTOH I have found my keeper rate went up considerably after adapting to these "characteristics". YMMV
BTW I am referring to proper working mkIII's. There is of course always the possibility that your cam is broken or got messed up during the "fix".
I have followed your posts here and completely agree about getting the most out of the AF by modifying the approach. However, I am pointing out that something rather simple like a snapshot can obviously fool the flagship Canon sports camera.
It's not all the time where you have the ability to play games with the camera and try and outsmart it. Sometimes, like this particular time... (I was in a hurry to get my kid to his first day of school). Where I just wanted to say, smile point the camera, focus and click. There is not one camera I have used in many years that couldn't get this shot at least 1 in 3 tries. All 3 of my "attempts" resulted in a completely smeared/blurry pic. It's just not what I consider acceptable from any camera, much less the flagship series.
I do definitely appreciate you taking the time to respond. As you may have guessed, I am venting a bit. I just hate worrying after every time I shoot with it, whether my shots are going to be in focus or not...worrying if I have mastered the art of fooling the camera's natural tendencies to misfocus, etc...
Andrew, I understand your frustration... It is no fun if you feel you can't trust your cam's AF. I also agree that the confusion only gets bigger when the AF confirmation turns up green and the pics are OOF. I do hope you come to terms with the mkIII's quircks. Maybe when the opp arises, you can try to similate the shoot with your son and see if a different approach will lead to better results. I would be interested to hear
Not bragging with this post just trying to help people. I also thank Daan for his work and attitude that this is what it is lets learn to use it well.
I can tell when I will have an OOF shot just by what I see through the VF. I don't know if it is because I started with EC focus from the film days so I almost always use the VF and keep my eyes there even after the shot is over or what but it is something I can tell.
The benefit though is knowing that the shot was missed and taking it again. Even in burst shots you can tell if it focus shifts. Doesn't help recapture a moment generally but at least you know.
What I do find interesting is that on static subjects I cannot remember it missing a shot. This is mostly in one shot center focus but sometimes on AI if I was shooting some movement prior. I don't find that I need to change from AI unless I know they will mostly be statics and I don't miss many shots to oof.
What is nuts for me.. It looks in focus in the viewfinder. It's almost like something like the sensor cleaning mechanism or whatever is causing this between the time it focuses and the time it records the image data. I know that sounds crazy, but it is what it seems like to me (or something similar). I have seen the same thing at 1/1000th + shutter speed too.. So it isn't camera shake I wouldn't think.
Does it look in focus as the shot is actually being taken or just before you release the shutter?
It looks in focus when I press the shutter but as it is firing I notice it move. This only happens in AI mode and mostly when I am tracking a subject and shooting multiple shots. I also get the best results when I let the center AF point lock but even then I don't get 100% keeper rate. It can actually get pretty low depending on the movement/background. Low like 20% for 5-10 shots. Wish I knew how to mitigate that.
Interesting...I suppose it could be shifting as the shutter release is being activated... That would explain a lot. Not why it's happening, but definitely the results.
