mill4570 wrote:
What

See above.

The more I hear about these issues the more I suspect the problems surrounding the MKIII's focusing revolves around an overdependence on particular wavelengths of light.

Alistair,

With respect, I believe you and your CPS friend may be mistaken. If I follow what you are saying, AF sensitivity set to slow, the camera checks the camera to subject distance less frequently, then a subject moving towards me a high rate of speed would never be tracked correctly, only subjects at slow speeds could be tracked. I would have to set the camera to fast to follow a race car and as you know that is not the case. Canon's paper on the MKIII and other ID series cameras clearly state " the tracking caluculation speed is unchanged based on tracking sensitivity. Tracking sensitivity determines how fast the camera will switch targets."

As far as the CF III-4 set to 1 the camera will continue to track the original subject only if an assist point has picked it up. If the subject that crosses in front of you is so large that it blocks the main and assist points, focus will be lost.

Richard K.

I didn't say that above 0 the camera would track a moving object any faster. This is clearly written in the Custom Functions III : Autofocus/Drive 1D3 White Paper. You are right though, I could have explained it more clearly.

Lets put it another way.

Say you are tracking a racecar and for whatever reason AF lock is lost. With AI Servo sensitivity set at 0 it takes, for example, 0.5 seconds for the AF to realise lock is lost and trigger a re-focus. Set at +2 when AF lock is lost, an re-focus is requested in 0.1 seconds. Set at 0 the AF tracks the car smoothly, no lens hunting and lock is maintained, set at +2 the lens hunts continually, minute movements, no smooth AF tracking movement as on the 0 setting and there is nothing between me and the target for the focus to jump onto, just the racecar, the same target.

I don't know how else to explain it than it feels like the camera is continually calculating range data to the subject, I really don't know how else to say it. At +2 my results are rubbish anyway so that is all there is to it.

Now I have confused myself silly so I am off to the pub, I really need a big drink!

Alistair Watson wrote:
I really need a big drink!


As do many MKIII owners.

Bottom line; If the OP is having trouble getting her MkIII to focus as well as a 30D in low light, worrying with AI Servo CF's is not the answer. I believe she has bigger problems with the camera.

Sorry Valerie,
Richard K.

Someone over on Dpreview claims to have 1.1.1 installed
http://forums.dpreview.com/forums/read.asp?forum=1032&message=24862197
Alan.

Has anyone heard if the latest firmware update had a significant impact on the MIII focusing issue? The last I heard was that is was being installed on select cameras at some sporting even in Asia. What was the outcome?



Edited by Jeff on Sep 20, 2007 at 06:38 PM GMT (Reason: Title for clarity)

I was speaking to CPS the other day, all they keep telling me is they are working with the pro photographers and developing a new firmware. Last I heard was that it wouldn't be available durinng September.

When was that? Just last week they were installing 1.1.1 in some cameras at the IAAF World Championships. Canon set up a special temporary service center just for that purpose.

Edited by nathanlake on Sep 19, 2007 at 08:51 AM GMT

A few days ago. Apparently they have been beta testing many of the pro bodies at the Osaka event but what version will eventually be released to us and when is open to guesses it seems.

Alistair Watson wrote:
A few days ago. Apparently they have been beta testing many of the pro bodies at the Osaka event but what version will eventually be released to us and when is open to guesses it seems.

At this point, I don't really care when it is released. I want to know how well it worked.

Well I have 1.1.1 and it has improved AF... I did not have the problems others had. However in AIServo focusing on a still subject it no longer hunts but stays solid.....

The CPS rep did say it addresses the AF issues (his words)

Peter Woods wrote:
I did not have the problems others had. However in AIServo focusing on a still subject it no longer hunts but stays solid.....
quote]

Mine has never hunted in AI servo when focusing on a still subject.

Mine does, You can watch it go in and out of focus slightly, I think one of the bigger problems I see in my camera is when I am tracking lets say a duck in the sky, Very easy subject to lock onto, When the duck moves lower where the background suddenly has trees in the far distance the camera will almost always head over to the trees even though I had been following the duck for some time. I have tried all sorts of CF changes and nothing has helped. I used center point, center point expanded and all 45 points. The camera wants to head into the trees almost every time.

Mine has never hunted in AI servo when focusing on a still subject.

Mine jiggles back and forth constantly when still and pointed at a still object. It is always close to focus, but it regularly goes from sharp focus to less sharp. 1.1.1 sounds like great news.

brainiac wrote:
Mine has never hunted in AI servo when focusing on a still subject.

Mine jiggles back and forth constantly when still and pointed at a still object. It is always close to focus, but it regularly goes from sharp focus to less sharp. 1.1.1 sounds like great news.

But you are supposed to use One-shot AF for still subjects and AI servo for moving subjects. When you do the opposite it will not work as good of course.

Peter Woods wrote:
Well I have 1.1.1 and it has improved AF... I did not have the problems others had. However in AIServo focusing on a still subject it no longer hunts but stays solid.....

The CPS rep did say it addresses the AF issues (his words)

Peter,
Did CPS indicate when the firmware will be available for public release?
most seem to think end of september. Alan.

Where can I find the 1.1.1. How do only some people have it? Is it safe to run the beta? Thanks

Lars Johnsson wrote:
But you are supposed to use One-shot AF for still subjects and AI servo for moving subjects. When you do the opposite it will not work as good of course.

Not all moving objects move at a steady direction at a steady pace, example football players change speed / direction all the time and AI Servo should lock onto them when they temporarily stop and stay stopped. AI Servo should lock on moving objects when they stop and hold still. This "over sensitive guessing" by the MKIII could be why many folks are having trouble in AI servo.

John Ferguson wrote:
Lars Johnsson wrote:
But you are supposed to use One-shot AF for still subjects and AI servo for moving subjects. When you do the opposite it will not work as good of course.

Not all moving objects move at a steady direction at a steady pace, example football players change speed / direction all the time and AI Servo should lock onto them when they temporarily stop and stay stopped. AI Servo should lock on moving objects when they stop and hold still. This "over sensitive guessing" by the MKIII could be why many folks are having trouble in AI servo.

When you talk about shooting still objects like he did, you are not talking about football players.............
With your kind of reasoning we should never be using one-shot because AI servo should be as good with still subject as the one-shot mode

>With your kind of reasoning we should never be using one-shot because AI servo should be as good with still subject as the one-shot mode

It seems to me that one-shot was designed for focus/recompose. I really don't think servo focus is designed to FAIL for stationary objects. That would be disastrous for sports shooters. Federer runs to the net, drops to his knees, and all the cameras go quiet as everyone fiddles with her focus mode button. You would have to change focus mode every 5 seconds. Servo is for tracking moving, intermittently moving, and therefore also stationary objects. It should be designed to hold focus and not jiggle too much. Predictive focus simply means looking at the last 3 focus commands sent to the lens and continuing the trend. It shouldn't require going out of focus in order to measure the trend.

brainiac wrote:
>With your kind of reasoning we should never be using one-shot because AI servo should be as good with still subject as the one-shot mode

It seems to me that one-shot was designed for focus/recompose. I really don't think servo focus is designed to FAIL for stationary objects. That would be disastrous for sports shooters. Federer runs to the net, drops to his knees, and all the cameras go quiet as everyone fiddles with her focus mode button. You would have to change focus mode every 5 seconds. Servo is for tracking moving, intermittently moving, and therefore also stationary objects. It should be designed to hold focus and not jiggle too much.

I agree otherwise it is pretty useless. Also, in many cases people are locking on to a subject before it starts to move and tracking / shooting from that point.

And now back to the OP's original topic...

The silence on firmware 1.1.1 is deafening

= Ed =

Peter Woods wrote:
Well I have 1.1.1 and it has improved AF... I did not have the problems others had. However in AIServo focusing on a still subject it no longer hunts but stays solid.....


The AF hunting on a still subject was the first thing I noticed about my camera. While some of the AF issue are complex, Canon had to know about that issue prior to release.

Why does a pro camera even need one shot?

Peter Woods wrote:
Well I have 1.1.1 and it has improved AF... I did not have the problems others had. However in AIServo focusing on a still subject it no longer hunts but stays solid.....

The CPS rep did say it addresses the AF issues (his words)

How did you get it? Was it installed when sent in for service?