1/125 is sufficient as a "reasonable expection of sharpness" for hand held only if your subject is perfectly still. I think the girl in the second shot might be moving enough to cause motion blur. I also think the 40D is softer than the 30D (mine was). My opinion after owning a 1DMKIII and 40D is that Canon DSLRs with the built in dust shaker take softer shots on occasion than those without the dust shaker ... possibly due to the sensor not being mounted as "solid" as those without the dust shaker. Doesn't always result in soft shots, but happened often enough to me without other explanation (mirror shake?) that I sold both and am getting sharper shots and am much happier after going back to a 5D. The 40D does a a better focus system tho ...
"Actually shot #2 was taken @ 125/sec w/ a Canon 85mm prime f1.8. This is not camera shake it is simply OOF. This is exactly what I have been troubled with. I have the camera set to give the audible signal focus when locked, used the center focus point and took the shot."
Yeah, I kind of figured that 1/125 would not yield camera shake. It's OOF no matter how much you guys try to ignore it or blame it on other variables.
"Well, there you go. 85mm * 1.6 = 136, so you shouldn't use any shutter speed under 1/136 second to expect handholdability. 1/125 is pushing it, and in this case pushed it too far."
Ok, treebeard. I'll break it down more. The rule of thumb is: Don't use a shutter speed below 1/focal length if you want to handhold. For an 85mm lens on a 1.6 camera, your slowest handholdable shutter speed should be 1/[85*1.6]=1/136 second. 1/136 doesn't exist, of course, but the idea is you would then pick the next shutter speed up from there to be safe. So, the moral of the story is, with your 85mm lens on a 1.6 camera, stay in the 1/160th or up range. Your shutter speed was too slow.
This is only a rule of thumb. I've taken acceptable sharp images at an effective 320mm focal length using 1/125th second, but that is a fluck, the exception to the rule.
And going back to the image, there are hints as to what happened. If you had simply missed the focus, there would have been some part of the image that is in focus. If the camera had focused very closely, say 2 feet away, the image would be blurrier that it is. The fact that the image is uniformly soft, tells us that the focus was pretty close, but camera shake caused the blur.
Several other posters after me understand what I'm talking about. Since several of us here seem to get it, maybe you should consider the idea that it is you who doesn't understand. The fact that you are not familiar with the handholding shutter speed rule of thumb tells me you have a limited photographic knowledge. This is right up there with the sunny 16 rule, concepts that I consider common knowledge.
I haven't had any problems with the 40D's single point system which weren't operator errors.
About the only criticism I have of the 40D is that the sensor doesn't look as nice as my old 10D. Oh, and the battery grip is plasticky and poorly integrated.
Ditto on unit variation. I'm on my second 40D and it is certainly front focusing a bit compared to my previous 40D (stolen in Amsterdam, shame on the thieves) which was generally spot on. Mounted a severely front-focusing 35L (2nd copy again) and the results were horrible even on the LCD w/o zooming in. (This compared to a near perfect first-batch 40D which was bought the day it became available, and a very good 35L copy as well)
Buying Canon is such a hassle now - some of my friends are willing to pay 5-10% more to mom&pop camera stores to pre-test equipment for us and deal with the frequent returns.
Recommendation to all new Canon buyers - test AF for total lemons before you buy! Canon service in Hong Kong is still pretty good, but there are just so many people there...sigh.
About the AntiAliasing filter.... The 1D2 is ragarded as having a very strong AA filter because the jpegs that it produces are quite soft until sharpening is applied (especially when compared with the 1D files). However, the raw files are not affected so much, and that implies that the jpeg softness is due to in-camera processing rather than a hardware filter. My guess is that the 40D will be much the same. I no longer have my 40D but if anyone wants to compare in-camera jpegs with converted raw files then you'll soon know whether or not the softness is due to an AA filter.
So many shots from mine missed the focus but usually that was obvious because something else was in focus. When it got the focus right the image was hardly soft (I shot raw) and so it seems to me that the AA filter was not a problem. Sometimes I got generally soft images in which it was impossible to see anything in focus and yet the image was softer than previously obtained. To me that is a camera defect rather than an AA filter characteristic. Maybe the sensor started cleaning itself or something silly like that. I know it isn't supposed to but then it also isn't supposed to produce soft images after confirming successful autofocus.
When a camera produces a generally soft image in AI Servo mode you may need to check the preceeding and subsequent photos (if shot as a series) in order to see if there was a trend as the camera may have been changing focus from one subject to another, for whatever reason (mine did that often, for no satisfactory reason). If shooting one shot at a time then you aren't able to do that and you may have to put up with occasional soft shots.
In one-shot mode the camera usually won't fire until focus is achieved and then a totally soft image should generally not happen unless you have moved too far and/or too fast after focusing. Nevertheless I did get what I regard as false focus confirmations on a number of occasions, and my G9 does that too. I haven't noticed why it happens - there seemed to be no obvious pattern or highlight or anything else to mislead the AF system - but avoiding it required that I keep an eye on the focus rather than trust the camera and that kind of defeats the purpose of having an AF camera.
Hello everybody,
This is my first post here so I'm even more sorry I have to dig this thread out at that occasion.
I've had my 40D for 2 months now and I find it a great improvement over 400D in terms of AF accuracy but at the same time I'm not so happy about its reliability.
In other words, most of the time it focuses really well with my primes (85 1.8, 50 1.4) but on some occasions it makes weird errors where I would never suspect it could have a problem.
An example is given below. A parked garbage track focused using One Shot and the central AF point. See how much the AF mistakes in the second shot.
My question - is this a normal experience with 40D? This happens every once in a while and I don't know if I should send the body to Canon or just get used to it.
I've found that my 40D occassionally goofs on the AF, but it doesn't happen nearly enough for me to consider it a problem.
Was the truck moving in the second shot? Perhaps the movement of the truck caused the front focusing, although it certainly could be attributed to other factors.
My question - is this a normal experience with 40D? This happens every once in a while and I don't know if I should send the body to Canon or just get used to it.
Best regards,
Tom
I would guess the truck is moving in the 2nd shot because all the guys are in the cab, therefore One Shot AF will not work with this.
It would be better to test against a completely static subject in One Shot AF, or a moving subject in AI-Servo AF. Years ago when I had a 20D, AI-Focus was rubbish, so certainly on that basis I would either use one focus mode or the other, no the AI one.
The truck wasn't moving yet, the shot was taken just after the men had got in. And it's hard for me to point what other element in the picture could attract the AF sensor more then this rather contrasty truck front.
With my body, I'd say approx. 10% of pictures are this way, depending on the distance (gets worse with far targets). The 85mm is spot on when conducting the focus chart test.
Maybe one of the falling leaves fooled the AF system? Just a thought...I've experienced similar results on rare occasions with my 20D when shooting wide open with fast primes but not as much when I'm stopped down slightly (ie f2). Not quite sure why.
It looks like the same effect I get with my 1Ds3 once in a while when I focus on (small) distant targets with fast lenses (near to) wide open... The focus actually lies way in front of the subject... As if the AF was fooled or didn't know what to do and defaults to this position. I know it sounds weird (because a lens focuses wide open) but stopping down and/or aiming for edges usually does the trick. I suspect that by stopping down you will give the AF more room to coreect itself. But this is just a wild guess of course...
Thanks for the replies. It feels reassuring to know that I'm not alone with that.
I've also noticed that stopping the lens down helps. I mean, not only the object gets within DOF due to the smaller aperture, but the DOF itself is better placed around the target. I don't know why it's so, perhaps there is a focus shift with the distance/aperture combo. And since one rarely use a wide open portrait prime on a distance, it seems quite reasonable that the focus is optimized for shorter ranges. On the other hand, it doesn't explain why the misfocus happens only occassionaly, so I really don't know if I'm even close to the truth.
Has anybody else noticed the same effect with EF 85mm 1.8 or other fast primes?
And one more question just to verify my expectations... From the AF's point of view, would you find the truck in the examples being a (too) small target?
tomdz wrote:
[...]
depending on the distance (gets worse with far targets). The 85mm is spot on when conducting the focus chart test.
Then it sounds like you have identified one of your problems, the focussing getting worse with far targets. Focussing on a small object in the distance, barely covered by a single focus point is always tough, the only thing you can do is 'more mm's!!'.
On general note, when I am up on the hills shooting low level military aircraft, alot of the people I shoot with have 40Ds, primarily with the 100-400 or 300/2.8 IS lens, and even when it is raining and cloudy, sure focus performance isn't great, but they walk away with a couple of superb shots, at least as good as the results from my 1D3 and 300/2.8 IS.
tomdz wrote:
Has anybody else noticed the same effect with EF 85mm 1.8 or other fast primes?
And one more question just to verify my expectations... From the AF's point of view, would you find the truck in the examples being a (too) small target?
I see the effect with my 35L, 85L II and 135L.
As far as the truck being a too small subject... You can read earlier on in this thread that some folks suspect that the 40D center AF point may cover a very large area, and that because of that focusing on small (distant -?-) subjects may not be accurate.
Looks like it's more like a "feature" rather than a bug, then... That means my body should be just fine, right?
Regarding the AF point size, what bothers me is why there are virtually no problems getting an accurate focus on an eye that usually covers much less area than the exemplar truck. Is that due to the fact that in the eye case there is less varying distance information in the frame?
?? 
