I know what you mean. From my findings, even on medium slow, the AF is pretty quick to jump to a closer subject, so I will give the faster settings a try. I doubt it will dramatically influence the overall AF performance but is worth trying. In the situations you illustrated, my habit is to try to anticipate this and let off the AF button, then reengage AF once the closer obstruction has passed, but isn't perfect.
Perhaps this is where CF III-4 might have an influence (I have not activated it). I don't have the camera or instructions handy at the moment, but I believe one of the settings will shift the AF point to keep focus on the same subject while the other setting will give priority to picking up a new subject, IIRC... Likely not to be a significant factor with single point selection, but with 45 point it might avoid the problem of focusing on the closer subject as it momentarily passes through the frame.... but I haven't tried it. At the moment I'm trying to get more used to using 45 point, or more accurately, trusting it. So far, it's better than I expected.
How have your results been in situations like the training camp? Your examples are backlit during the day, which is where I first had the most serious problems with the III.
My results have been pretty bad when the sun is out no matter what direction the light's shining. I actually think I get better (still poor) results when it's backlit, though, than when it's direct.
I was convinced certain colors of jerseys were making it worse only to have different results with the same colors on different days under similar conditions. I still think it doesn't like dark blue jerseys no matter what color the numbers are. Several days ago I had fairly good results with dark numbers on white jerseys, but this afternoon I had terrible results shooting a similar combination.
I always used center AF point only when I used a 1D and this setup really shined. I remember trying expansion a few times, but it always seemed to make the AF seem sluggish and my keeper rate went down. I don't think 45 point focus is an option for me during football season with the 1DIII because players are usually so close to each other. I did try expansion during baseball season and it didn't do anything to help.
I'm basically waiting to hear from Canon and or Rob Galbraith on the issue.
We've probably traded notes in the past, but I'm curious which lens you primarily use? So far I've had the most trouble with the 400 2.8 IS (which you may have already read in previous posts). But most of my sports work with the III so far has been during evenings/nights. The only day stuff I've shot has all been on Field Turf and I can't get a clear idea if the AF problems here are due to the camera or primarily due to the heat coming off the field... results with the IIN in the same daytime conditions are also poor.
I know what you mean about needing to pick off the right player with the AF point. But 45 point did work better than I expected when I used it at the last game. Compared to the 1D, the III packs a lot more computing power. I can't say the III has ever given the impression it was struggling to keep up with expansion or 45 points active. Yet in the past with older cameras, this definitely could happen with 45 points. If I'm shooting vertical iso-action with the 400, then I prefer to use single point one up from center with CF III 8-2 on. At least for me, I do think the camera has behaved more consistently with AF expansion turned on, but again, I'm somewhat skeptical if it actually makes a difference because so many games have been later in the day. The greatest difference has actually been in using the 600 instead of the 400 whenever possible. Not to suggest that's an acceptable solution.
I'm having problems with the 400 2.8 IS, Sigma 120-300, and 70-200 2.8 IS. Indoors, after dark, and overcast conditions generally yield better results, but I still get that focus slightly behind the subject more than what I consider to be acceptable under those conditions.
Most the stuff I've shot this summer has been on grass, so I don't think the problems I'm having are caused by the kind of heat that radiates from artificial surfaces
I've decided that if Canon can't get my camera to work with center point only at least as well as the 1D, then the whole thing's a bust for me. I won't accept anything less. I'm all ears when it comes to work-arounds until a fix is provided, but without a real fix I'm getting closer each day to feeling like my only recourse is to raise hell with Canon (whatever that means.)
Also, I keep reading that Canon says only a few people have sent their cameras in to be repaired (implying most work correctly). I don't know why I'd send mine in if nobody there knows what to do with it.
One factor to consider is Nikon apparently coming out with new cameras thurs. This may make Canon reluctant to talk about the AF issues.
I expect Canon's position will be that there are new algorithms and new trade offs. They probably feel part of the issue is setting up the camera. But hopefully they will also tweak the firmware. This may take time, and they may not want to talk about it until they really get a handle on the issues.
I have the mkIII and have not found focusing issues. But I haven't tested the camera. I am confident the issues are real. But as long as the focus hardware is as good or better than the IIn, I'm also confident Canon will get it right.
Canon has had some firmware bloopers. The focus algorithms are surely complicated compared to most programming in the camera.
Garylv wrote:
Steve, that's incredible, isn't it. Similar to what Jeff experienced when he posted his thread. He was not seeing any trouble before, and all of a sudden had horrific results. I wonder if that's still a big mystery to Canon. I hope they know what's causing all the trouble. If they don't, the 1DsMarkIII will be the same way. It has the same AF system.
See my first post in this thread. I guess great minds do think alike…
If anything, IMO the problem may be that the AF system is overly sensitive, .........................
After only two days and 6 hours shooting of mostly bird-in-flight, my early feeling is the AF of the Mk3 is overly sensitve and too eager to lock on high contrast subject. I have updated the firmware from 1.0.9 to 1.1.0.
As I have been using my Mk2 for the same type of shooting on the same location (my local bird sanctuary), I have really good idea what works and what does not work. With my Mk3, I have been trying to limit myself to only changing two variables: the tracking sensitivity (to SLOW or one step more sensitive) and the AF expansion (C.Fn III - 1/2). I was testing my MK3 (using the center AF point) with my 500f4 with or w/o 1.4X, on small subjects like sea gulls with relatively busy and high contrast background (trees). So far, the results are still inconclusive but the Mk3 definitely has no problem acquiring focus. The Mk3 AF acqusition is for sure faster than my 1DMk2 (may be too fast ).
523xxx firm:1.1.0 One shot issues. Often, with center focus chosen, area chosen is either out of focus or not the area in focus. More hesitation grabbing focus than either my 5D or my, now sold, mkII.
I can remember always being amazed how much quicker and more % infocus the mkII was than the 5D when "point and mash" focus of a fleeting subject. Now with the III the only way I can get high % of infocus shots is to point and 1/2shutter release, wait for infocus beep/flash then fully depress. This itself looses half the opportunities, AND often, by then the object moves ever-so-slightly off center and the camera focuses on the background.
This may be a 'getting used to a more sensitive focus mechanism' issue but in 4700 pics so far I haven't figured out how to come close to the quickness of the 'less sensitive' focus of the EOS3, EOS1V, 1D, and the 1DmkII bodies I have always been happy with.
My first 1D Mark 3 was returned, terrible AI Servo tracking, loads of out of focus issues, shooting at motorsport events as well as sports. It was returned for a refund.
A month ago I bought another Mark 3, this one works superbly at tracking motorsport and is equivalent to the 1D2N however, I shot a rugby match yesterday and got a huge amount of out of focus frames.
Now I am concerned whether the problem is indeed fixed. When I shoot motorsport, primarily I am at an angle to the car, in some cases it is coming head on, also the car is moving along a predictable path. However when shooting a sports person, say rugby, people do not always move in a predictable path, ie even though they are moving forward, their upper body is generally moving backwards and forwards.
Question, for those people who have had AI-Servo issues, have you had issues shooting everything or say been able to shoot (big target) racecars successfully but people runners unsuccessfully?
Alistair - this goes to the root of the problem, which is the inconsistency of the thing.
It's almost impossible to say with any certainty that if you take out your MkIII and shoot with it today that you won't have horrible results, then you could take it out tomorrow under the same conditions and get perfect results.
When firmware 1.1.0 was released I thought for a few days that the problem was fixed on my MkIII, but then I had a few days when, inexplicably, the camera couldn't AF with any kind of consistency. Since then it's been good days and bad days.
I sent back that MkIII for replacement yesterday. Sending back MkIII's seems to be something I'm doing quite regularly these days...
Sadly I pushed Calumet very hard to return the first Mark 3 and they accepted but weren't happy since there was no fault they could prove. I am now out of the 30 day return period of the 2nd Mark 3 and I can't see Calumet refunding me again.
The funny thing is that with this body I can ALWAYS get excellent results in servo mode shooting motorsport, but then again the target is nice a big and moving in a consistent way. Not so with people sports so far. Frankly I am getting rather annoyed.
Looks like the rugby event I will be shooting the week after next will be my trusty 1D2N for primary and 1D3 for backup. Not quite how I intended this to work out!
1. Camera cannot focus through water droplets. Tried shooting through a sprinkler to see what it would do during bright daylight and it would catch the droplets and never focus on the subject behind. Impressive actually but not necessarily helpful
2. In Servo mode the first shot has to be in focus. If the first shot is good the rest are good, if the first is off the rest are off.
3. Getting the center point to focus with 45 pts enabled seems to be about impossible, it does happen but takes alot of effort.
From what I have experienced it seems like a case of Canon upping the sensitivity of the AF sensors to a very high level making it quick to focus and giving it low light ability. (low light AF is unprecendeted) The downside to this is that getting the AF system to grab the first shot in confusing situations in both AI and One Shot is a pain sometimes. Still need to shot more to determine how to achieve consistent results in tricky situations and do it fast.
I wish Canon would tell us exactly how the system decides what to focus on, give us a flow chart of decision making it goes through. (maybe there is one but I haven't found it) My hope is that Canon engineers were smart enough to develop a system that with practice will produce quicker focusing with more consistent results. My last shoot did not go well and I am glad I had a second shooter because my OOF rate was 50+%. Part of this was my fault though because the viewfinder sharp I assumed the picture was even though the focus point was off one from where I wanted, but at F2.2 beyond 4X6 size it is noticable that it is OOF.
Until Canon has something to add I hope that our collective experience and thought can come up with some answers to getting the most out of the Mark III.
Has anyone used Arthur Morris' mkIII user guide and still had problems? I'm still wondering if most of the issues are dealing with increased sensitivity.
Kier wrote:
My first MkIII was a 520xxx serial and had terrible AF inconsistency problems, both in one-shot and ai-servo modes. That camera was sent back.
It was replaced with a 523xxx model, which unfortunately also suffers from AF inconsistency in a bad way.
I sent that one back yesterday and am expecting a replacement to arrive tomorrow.
Keep us posted, Kier. It's logical to assume the one you receive next will have a later serial number again. I tend to think it's not going to be fixed, because they are still investigating the problem (from what we know at this time).
But anything is possible. Let us know how the newest camera performs. If it's a good one, then there is hope they can fix the rest of them for us.
Iberian wrote:
From what I have experienced it seems like a case of Canon upping the sensitivity of the AF sensors to a very high level making it quick to focus and giving it low light ability. (low light AF is unprecendeted)
From the conversation I just had with a technical product director at Canon I would guess this is true and that there are problems maintaining AF servo lock on targets. Based on the conversation I will reset my servo tracking sensitivity to 0 and next week make an experiment of the rugby game using the AF expand options since this seems to be the current recommendation to resolve this. Wish me luck!
1. Camera cannot focus through water droplets. Tried shooting through a sprinkler to see what it would do during bright daylight and it would catch the droplets and never focus on the subject behind. Impressive actually but not necessarily helpful
Mine must have backfocused on this one..
J/King.. But seriously.. At least with this sequence of about 5 shots (this is a crop)..every one was spot on focus wise. I am planning to do a lot more testing though. So far, so good.
I have 1.1.0 and Af set to 1 stop slower than normal. Temp. 101 degrees.
Garylv wrote:
Keep us posted, Kier. It's logical to assume the one you receive next will have a later serial number again. I tend to think it's not going to be fixed, because they are still investigating the problem (from what we know at this time).
But anything is possible. Let us know how the newest camera performs. If it's a good one, then there is hope they can fix the rest of them for us.
I too fully expect the replacement camera to perform in the same flawed manner, as I believe that the problem is inherent to the MkIII rather than being a case of a few dud cameras.
However, the fact that all these cameras are being returned to dealers with customer complaints of unacceptable AF performance must be getting back to Canon, and hopefully will force them to take some action to resolve the matter.
I'll post my results as soon as I've had a chance to test the latest one.
I'm not convinced that the variations by serial number indicated in the data so far represent anything but random chance, e.g., I think the data is consistent with the hypothesis that the probability that an owner believes his MkIII has AF issues is (so far) uncorrelated with serial number.
).
