Also, I forgot to mention that you were shooting handheld, which means there is some movement in the camera, and with a macro lens shooting with such thin DOF, any slight movement may force the AI Servo computers to readjust for focus. Try this on a tripod, and see if it still hunts. I'm curious.
Call me crazy, but I'd have to say that AI Servo isn't for potted plants. FWIW, I think the AI Servo performance on my 40D is nothing short of amazing, and much better than my 30D. I've already posted this link, but here's what I did to get these impromtu shots (15 minutes after my 40D arrived vis FedEx). AI Focus mode, all AF points active, ISO 200, handheld, 300mm f/4L IS at f/4. These are all the shots, from first to last, at 6.5 fps:
Those are 50% crops, with the crop box not being moved between the shots, so the 'copter was caught by several differing AF points as I was tracking it. Generally, I'm not very good at these sorts of shots. FYI, that's a medical helicopter from one of our local ER's, and it was hauling @ss, not hovering. Look at how rapidly the cloud background changes frame to frame, even at 6.5 fps.
Alan321 wrote:
If the problems with my camera are normal then anyone in need of better AF for action / sports / wildlife needs to consider a second hand 1D2(N) rather than a new 40D.
- Alan
Hi Alan. I don't think your problems are normal for the 40D - at least to that degree. I just ran a few tests with mine. I did 3 series - 16 shots, 11 shots and 12 shots. I used 70-200 f4 IS handheld at 70mm, f5, ISO 800, 1/100. I didn't have any issue with the AF. I went through each series to look for any big differences like yours and they are just not there. If you think any of my settings should be different, please let me know and I will test again. I also tried to use a subject with print similar to yours - if that makes a difference.
I have examples of the biggest change in AF, during a series, if you are interested. What little difference there was, I would say was caused by my poor handholding skills - which is really better on the 40D than the MK IIN becasue of the weight difference in bodies
rachp wrote:
Hi Alan. I don't think your problems are normal for the 40D - at least to that degree. I just ran a few tests with mine. I did 3 series - 16 shots, 11 shots and 12 shots. I used 70-200 f4 IS handheld at 70mm, f5, ISO 800, 1/100. I didn't have any issue with the AF. I went through each series to look for any big differences like yours and they are just not there. If you think any of my settings should be different, please let me know and I will test again. I also tried to use a subject with print similar to yours - if that makes a difference. ...Show more →
Try shooting at F4, not F5?, and at 180mm, not 70. DOF will be significantly reduced at those settings, and any focus drift would be amplified. Also, his subject distance is probably rather short, considering he is shooting with a 180 macro. Not sure what distance you are shooting at. All these things could make a difference.
Dan, I can't do more testing now - Canon has the camera. Thanks for the other info.
You wrote The AI-Servo is designed to track subjects that are moving in predictable patterns. Once the pattern is disrupted, it has to regain it's focus points and predictions again
Quite true. However, when it loses track it should freeze with the last known focus position wherever it was. It's not supposed to wander all over the place. In this way DOF may well cover any errors if there is enough light to allow small enough apertures, but the way my camera was behaving it could not.
Will, yes I've read that the 30D behaved the same way, as apparently did the 5D. I just expected better from the first camera in the series with an actual upgraded AF system since the 20D was introduced (the 20D AF hardware is shared by the 30D and 400D too). Not to be, this time.
I'd like to stick with pro cameras and pro AF but they can't do them right at the moment either.
Alan321 wrote:
However, when it loses track it should freeze with the last known focus position wherever it was. It's not supposed to wander all over the place. In this way DOF may well cover any errors if there is enough light to allow small enough apertures, but the way my camera was behaving it could not.
That's exactly what it is supposed to do. It can't just "freeze," otherwise it will fail to find the new focus point and pattern. The camera can't find focus without the focusing mechanisms actually moving. That is how focus is detected. If it "freezes" it is no longer capable of finding focus again. So once it loses it, it will hunt until it finds it again. And if you are shooting a still object in AI-Servo, the camera thinks it is moving. If the object doesn't actually move, the focus will still keep anticipating the movement that isn't happening (or in handheld, may very well be moving because of your body's subtle motion), and therefore begin to drift.
Then focus got back onto the couple for another 4 frames until the sequence ended.
These subjects are not large in the frame but are certainly bigger than the central AF sensor. There remains enough contrast on the subjects for the focus not to drift off to the far distance another 100m away.
I questioned your initial conclusions about ai servo performace based on your shots of static objects. I will also say that a macro lens -handheld - would not have been my choice for testing tracking performance as the focus is slower with most macro lenses, and slight movement of your body and camera will impact ai servo focus. Are you sure the problem is not lens related?
I have a tamron 180 macro which has had to go back for service twice now as there was an issue getting it to lock focus properly.
Look at the position of the circle, and then look at the whitepaper. Not sure why you think the position and sizes of the sensors is immaterial, other than that you are accustomed to 1 series AF. Again, it appears to me to have intentionally locked AF on the far background because it is very much in focus. In order for this to happen, an AF sensor had to be registering the background. Maybe you are right and the camera "accidentally" decided to use the upper left sensor for a moment (note the pole), but that seems pretty farfetched. Anyway, that much of a difference cannot simply be "drift", it had to be intentionally done by the camera.
Haibert B. wrote:
I hope they fix the problem and it doesn't come back without repairs as "up to spec."
::rolleyes::
I haven't seen anything to tell me it is broken. You can't use AI servo on a busy static subject, and you need a longer lens to take the shot of that couple.
timbop, I agree the camera did it deliberately. But the sensors are not long enough for that to reasonably be the cause. It did take three frames (half a second) to get to that focus point, so it didn't just suddenly jump from the people to the far background as if another sensor kicked in. The lens can respond faster than that too - it managed fine for the first 6 frames of shooting and a couple of seconds tracking before I started shooting.
Never did I think the size and position of the AF sensors immaterial. I just know that with previous cameras they have been only up to about three times the length of the AF position indicator box, and only one pixel wide. And the system was biased to staying in focus if there was still enough contrast to focus on, rather than go searching for a new subject. These subjects are big enough to be covered by the central AF cross-type sensor given the aiming point. Had I drifted off target then of course the sensor would and should latch onto something else, but it should do so instantly and not over several frames.
I had similar experience later on with my Canon 70-200 f/2.8 IS on the 40D, so the problem is probably not the fault of the Tamron lens, which is just the one I had been using before these sample shots were taken.
Wade, that didn't sit well with me In the past nine months three of my cameras (1D2, 1Ds2 (new) and 40D (new)) and a 100-400 L lens have had to be serviced by Canon. Two had confirmed focus problems. One is not yet back from service and so I don't know what they found. Another is now just on its way to service. I don't go looking for the "big flaws" - they find me while I am out there trying to shoot. Then I get peeved because it takes a lot of time testing and analysing to find or verify a fault that should not exist and then it takes a further two to three months to get the gear fixed and returned. There's no one-week turnaround here.
Having one problem is unlucky. Two is a nuisance. Four is very annoying. It's even worse when I have two faults at a time. I have a long history of Canon gear problems over the past 8 or 9 years but I was hooked on the EF lens system back in the days when only Canon had IS.
I guess in a few weeks we'll find out what if anything they find wrong with my 40D and then I'll be sure to let everyone know. Either what I am complaining about is normal for all 40Ds and readers will now be aware, or else mine is worse than normal and it'll be fixed. Whichever it is, you'll get to find out the pain-free way
Having one problem is unlucky. Two is a nuisance. Four is very annoying. It's even worse when I have two faults at a time. I have a long history of Canon gear problems over the past 8 or 9 years but I was hooked on the EF lens system back in the days when only Canon had IS.
Alan, my sympathies re: all your problems with Canon gear. I know a guy that can't
wear watches. Maybe it's his magnetic personality, dunno but any watch he puts on fails!
Just wondering if you guys may have gone to the same school together?
Canon cameras cannot be used with AI Servo on a static subject, the AF starts to hunt. This was detailed on the old CPS FAQ site which I don't think still exists? This has been long known.
AF with Macro is marginal at best, using it with AI Servo and a target such as that without a clear deliniation of contrast is just asking for trouble IMO.
With moving subjects you can expect a less than 100% focus accuracy between shots. The same FAQ detailed that the camera will fire the first shot of a series in AI Servo regardless of whether focus had been aquired and confirmed. My experience backs this up. On a test done with a 1D mkII when it was new a magazine shot a moving car till the limit of the buffer, 30 out of 40 were sharp, the rest the focus had hunted. They seemed to think that it was pretty good, I remember the D2X also being tested and coming out only very slightly better. Your ratio was 2 out of 3 not the 3 out of 4 of the pro bodies but to be honest I would expect that relative to the non pro AF.
Regardless of all this and I'm not justifying it in the slightest, I hope you can get some luck out of canon service. It's amazing the things which people can discount as impossible on this board, I had a problem with f2.8 or faster lenses focusing on the background in fluorescent light only with my 5D's - few believed me but when it came back from CPS it was perfect!
In the past nine months three of my cameras (1D2, 1Ds2 (new) and 40D (new)) and a 100-400 L lens have had to be serviced by Canon. Two had confirmed focus problems. One is not yet back from service and so I don't know what they found. Another is now just on its way to service. I don't go looking for the "big flaws" - they find me while I am out there trying to shoot. Then I get peeved because it takes a lot of time testing and analysing to find or verify a fault that should not exist and then it takes a further two to three months to get the gear fixed and returned. There's no one-week turnaround here.