p.28 #1 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Jeff wrote:
Finally spoke with the actual manager at Customer Relations, and after several people telling me that the submirror work was repeated during this second repair trip, he acknowledged that it was not, and that the camera had tested to factory specs (whatever those are!).
So I wonder just how effective this "test" is to determine whether or not they perform the sub-mirror fix, or just paint a white dot inside the battery compartment and send the camera back. Some of these cameras are coming back awfully fast, including mine.
This "test" is a big unknown factor to everyone, including RG, I would imagine.
When the warm days of Summer begin, there may be a lot of people experiencing the same inconsistent AI-Servo focus.
p.28 #2 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Rob-Bob wrote:
"...Makes me wish we could ship ourselves along with our cameras, so that we can actually have some interaction with the individuals, some relationship...
...Maybe you can get the manager at Customer Relations email so you can send him a link to this page, and the other countless sites / posts that are trying pretty much in vane, to get some attention directed towards these "imaginary" issues. Maybe just tell him over the phone the web adresses, in hopes that he can read up, then maybe experiment some himself, and as a best case senerio, take it up with the "Powers that be"..." ...Show more →
Yesterday I sent him a detailed e-mail with link to 3 different websites, example images with discussion, etc. If today they still say they cannot replicate the problem, I'm going to ask him if there's someone in the office with tennis shoes on, have him find a 70-200/2.8L and have the person run at him in the parking lot. I don't know what kind of test apparatus they use to 'qualify' these cameras, but clearly it does not work for these kinds of AI Servo problems.
mill4570 wrote:
Did you discuss with him the "ghosting" or secondary image problem? If I remember, there are two problems with your camera; "ghosting", and a "focus plane" issue. The reason I ask, I shot some soccer this weekend, and experienced the ghosting at f5.0 (300f2.8). The dof seems to be on the player, if you look at the ground, and the shutter speed is Ok at 1/3200. This may happen on two frames out of a five frame burst, maybe one out of every twenty burst. This doesn't look like the camera missed the focus, just parts of the frame have this ghost image that makes it look soft. ...Show more →
I discussed both issues with him, and made sure that he knew that there was an image on the DVD that showed the 'focal plane' issue. That's distressing to hear that your camera shows this secondary image issue, as well, but based upon RG's results, I kind of assumed all MkIIIs would tend to exhibit this. I'm hoping that the 'secondary image' issue is unique to my camera.
Remind me Richard, is yours pre-fix, 'fixed', or 'Blue Dot'? It's becoming harder to remember who has what.
p.28 #3 · ['NEW Fix' UPDATE!] - MkIII AF still broken
after the fix, canon probably (at best) reverifies the critical dimensions of the camera, a per the mechanical schematics. now, i'm not sure if they're verifying just the sub-mirror assembly, or the sub-mirror assembly relative to the camera or what. or maybe they don't even bother doing that after they use their CMM'd (verified & certified) jig to reinstall the sub-mirror assembly. pure speculation based on the reports i've been reading about.
judging by Jeff's encounters with canon, i'd hazard the guess that canon does not perform functional tests on their cameras, nor do they seem to have standardized functional acceptance criteria, specifically around the ai-servo function. i don't know for sure- obviously. but as long as that practice continues, canon will be susceptible to serious errors like we've been seeing in the 1d3. especially when they make significant technological strides which require structured and rigorous testing, prototyping, etc.
p.28 #4 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Simon, 'submirror assemblies' are apparently not routinely replaced, though sometimes are. It's the 'stopper, submirror' that is being replaced on all cameras, and I'd hazard a guess that it's easier and cheaper to do, thus the very quick turnaround times for many.
p.28 #5 · ['NEW Fix' UPDATE!] - MkIII AF still broken
hm. Jeff, then i've technically misunderstood the problem, which is not difficult to do with the patchy information we have . what does the "stopper" do? does the sub-mirror float? i thought it just sat there and focused light onto the AF sensor. if the stopper is simply a datum position for the sub-mirror assembly to rest up against (to meet a critical dimension), then you are effectively "replacing" the sub-mirror assembly because you are repositioning it. and what if your sub-mirror assembly is no good to begin with...?
anyway, the problem still persists- yours, mine and others'. whether you mechanically modify a tool, or whether it has just come off the assembly line. you perform your system functional tests in some controlled manner. that way, you know what you're aiming for and so does your customer. and if your tool doesn't meet or exceed the acceptance criteria, the customer can hold the manufacturer responsible beyond a reasonable doubt. obviously, nothing of the sort exists here and thus we have much unwanted subjectivity, bickering, etc.
p.28 #6 · ['NEW Fix' UPDATE!] - MkIII AF still broken
As I understand it, the stopper is involved with how the AF submirror comes to rest in between exposures, at the critical time when AF is being read/calculated. Any misalignment of the submirror at this point results in focus errors, and it would appear that sometimes the AF submirror was coming to rest properly, other times not. It's anybody's guess as to the effect of shooting at 10 FPS on this issue (i.e. does it make the errors more frequent?). The presumption has been in relation to the rubber compound used in the stopper being subject to temperature fluctuations, which in turn affects how the submirror comes to rest in between frames; hotter presumably equals more expansion of the rubber compound, making the submirror assembly rest in a different place than it should, creating errors in the accuracy of the focusing. [As an aside, remember from the White Paper how the mirror (in addition to the AF assembly) was completely redesigned to mitigate mirror slap, etc. Lots of new stuff going on in that general area.]
As I understand it, anyway. Canon has given so little concrete information, so much of the above is speculation to some degree, and perhaps someone else can add additional technical info, or perhaps shoot some holes in what is above...
p.28 #7 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Jeff wrote:
Simon, 'submirror assemblies' are apparently not routinely replaced, though sometimes are. It's the 'stopper, submirror' that is being replaced on all cameras, and I'd hazard a guess that it's easier and cheaper to do, thus the very quick turnaround times for many.
Thanks for mentioning that, Jeff. I'd forgotten they were replacing the stopper on all cameras. It was beginning to sound like they test the cameras in some way first, and if they meet the factory specs, nothing was really done.
Also as I remembered back, Canon said all cameras within the noted serial number ranges could be affected, but not necessarily.
p.28 #8 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Jeff,
To me, the major issue with the MKIII is simply no two really behave alike. Mine has not been repaired, but is within the SN range for "possible fix". I have had no trouble with One-Shot operation with this camera (weddings). Servo works better than my MKIIN in low light (HS football with flash), and almost as good in bright sun. The only problem I am having is the AF will sometimes jump two to three feet in front of a target that is moving directly at the camera. The AF will correct it after one or two frames. Then there is the "ghost" thing. They are not happening enough to cost me money, but it is a pain and I believe the camera was designed to out perform the MKIIN not be "as good as".
RG did not mention the "ghost" issue in his article, I believe you spotted that one first, and I'm not sure Canon is aware of, or working on t. That is why I'm curious about your problems.
p.28 #9 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Richard, I'm also surprised that RG didn't note the 'ghosting' issue, but perhaps he's simply being silent until something else happens. He certainly has no interest in returning my (rather detailed) e-mails, that's for sure!
CITS absolutely knows of the 'ghosting' now (as of yesterday), though to what level I do not know. They very well may consider my camera an aberration, but I think that there is enough data out there to suggest that the 'ghosting' issue is somewhat widespread, based upon what I've seen around various websites. The 'focal plane' issues very well may be limited to my camera, or only a very few. If it were widespread, I'd think that other people may have noted it, but it's really only obvious for landscape-type shots. For vertically-oriented 'runner AI Servo' sequences, it is very, very difficult to discretely identify, because of the lack of subjects in the focal plane (shooting at f/2.8). However, it could very well be the reason that many have reported 'not finding anything in focus' in what should be the plane of focus, for example, near a runner's foot you can find the focal plane, but the top half of the runner shows little or nothing in focus, or possibly something severely front- or back-focused. It's just not a simple one to nail down.
I personally believe the 'ghosting' issue to be the cause of many of the 'slightly soft' images that people often report in AI Servo, but many of the images can be saved (with over-sharpening) in post. So, like you, not enough to be a fatal flaw, but occasionally noticeable if you look. The problem becomes that it cannot likely be caused by the submirror, nor to AF at all (that I can surmise).
Of note, I went back through Rob's 'post-fix' images the other night to refresh my memory on what a 'Blue Dot' MkIII's AI Servo AF performance should be like. I was going through them thinking, "Wow, it really did help the AF, even though it did nothing for mine!"
Then I realized that I had accidentally opened the MkIIn's 'fast runner' folder of images...
p.28 #10 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Jeff,
I do most of my sports shooting in vertical mode, and the best way to describe what I am calling "ghosting"is this.......a player's foot or leg plus the ground are clearly in focus. As you move up the player's body, the shirt and face are much softer. If you check the image at 100% or 200%, the face and body look as if there is motion blur, if that makes sense. The uniforms the kids were wearing Saturday had the Nike logo on them and you could clearly see the logo was "smeared" for lack of a better word. SS was 1/3200 by the way. Maybe I'm not doing a very good job of describing this but it is difficult to explain. I believe I deleted all the images, but if I find another one how about checking it for me? Your keen eyes have developed a much deserved reputation I'm not sure now if I'm describing ghosting or the "fp" issue.
p.28 #11 · ['NEW Fix' UPDATE!] - MkIII AF still broken
That sounds very much like how you'd expect the 'focal plane' issue to manifest itself. What I have not been able to sort out is if the 'ghosting' and 'focal plane' issues are one and the same, since to see the ghosting, you need a high-contrast edge to clearly see it. The lack of focus on the right side of the sensor (top, in portrait) is pretty easy to see, when it manifests itself severely.
In RG's images where the 'ghosting' can be seen, I think it's fair to say that most of the time you cannot find anything in focus in the upper half of the image, when (based upon the focal plane at the feet) you'd expect to. I suggested this very thing to Rob in two e-mails. So, unfortunately, your description is perfectly clear to me, and I must say that I'm sorry to hear it. Perhaps my camera isn't so unique, after all...
p.28 #14 · ['NEW Fix' UPDATE!] - MkIII AF still broken
I also just missed a call late this afternoon from someone entirely unknown to me (within Canon) with info about my camera. Have to wait until tomorrow, but at least it sounds like they have a word now.
p.28 #17 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Wickedfn4u wrote:
Lets hope the word is discovered. As in we discovered another major design flaw that will resolve all issues!
I know I know.. some of you have perfect ones. Enough said you don't need to fix them but mine is not as described yet. Quality matters.
I got the impression from another thread that you were happy with the 1D3. What issues do you see? Is it the same as in this thread or something different?
p.28 #18 · ['NEW Fix' UPDATE!] - MkIII AF still broken
I am happier than with my early M3's but still feel it is not a leap forward from my M2's as far as AF/IQ is concerned. I love the layout, size, menus, ISO and physical feel of it. I just feel with the early M3 AF was horrid the fix is much better but I don't feel it is an improvement over the M2's. I have been more scared to shoot at 2.8-3.5 than I was with my old bodies. At 4.0 and up it is killer with a capital K.
p.28 #19 · ['NEW Fix' UPDATE!] - MkIII AF still broken
Jeff wrote:
I also just missed a call late this afternoon from someone entirely unknown to me (within Canon) with info about my camera. Have to wait until tomorrow, but at least it sounds like they have a word now.
Any news?
I sent Canon a follow-up email about the lack of scheduling mine today... Hopefully, they'll get on the ball here soon. Not that I "need" the fix...it's just that I have a perfect layoff for next 4 weeks (after tomorrow) to have it serviced. Hope they call me soon.
Very interested to find out what they have to say about your camera Jeff...
. what does the "stopper" do? does the sub-mirror float? i thought it just sat there and focused light onto the AF sensor. if the stopper is simply a datum position for the sub-mirror assembly to rest up against (to meet a critical dimension), then you are effectively "replacing" the sub-mirror assembly because you are repositioning it. and what if your sub-mirror assembly is no good to begin with...? 
