In analyzing the photo, it appears that due to the angle of looking upward, the focus plane cuts through the icicle and the brick wall, as would be expected. Images do not have a focused spot or in-focus region, generally, especially a rectangular one. The image plane, the lens plane and the focus plane are all involved, mathematically modelled as flat planes of infinite extent. Almost all SLR lenses have lens planes parallel to the image plane, resulting in focus planes that are parallel to those two. Note also that focus is not a binary issue, but is a question of degree, though a criterion can be applied to transform it into a binary decision.
The image at the size as shown looks reasonably in focus. Without a closer larger crop, such as the fearsome 100 % crop, it is hard to tell how much the icicle is in or out of focus. Also not given is one of the most crucial pieces of information, which focus mode was used, crucial because the image contains a moving drop that follow focus could have been tracking. Note also that f1.4 is wide open and will have the softest performance for that lens and the thinnest depth of field.
The applied focus points (red/black squares) do not always get applied exactly at the point selected.
My personal experience with 20D is that all focusing issues were due to my own operator errors.
I guess if I had used say f/2.8, the problem would've been avoided. I was being greedy and trying to get as fast a shutter w/o compromising significantly on the ISO.
data1ore wrote:
I guess if I had used say f/2.8, the problem would've been avoided. I was being greedy and trying to get as fast a shutter w/o compromising significantly on the ISO. Oh well, a good lesson learned!
There are few pure black and white (binary) issues in photography. f/2.8 would have helped, but how much and if it was enough is hard to say. You were not being greedy, you were trying to make the best compromise you could among several factors. Experience will help, but will not always be definitive, which is one reason why professionals will take many nearly identical shots for one image, bracketing different issues in an option matrix.
One of the secrets of a being a "good" photographer, or at least perceived as one, is to not show bad images. (Showing this image here was good for diagnostic purposes.) If a pro takes 100 shots to make a single great image, that image is the one they show, not the 99 other defective ones. Since digital film is cheap, take the 100 images if necessary. That is the right kind of greediness here!
Random focusing errors can also be one of the reasons. I have once tried shooting a page of magazine taped to the wall with my 50 f1.4 and 135 f2L at maximum aperture. Many shots were taken and about 20% were found to be OOF but the rest were spot-on. The camera was a 20D mounted on a Gitzo tripod weighted down with a heavy lathe chuck. Mirror lock-up + self timer were used to minimize vibration blur.
The icicle and wall appear in the same plane. Remember that you are not square to the wall. It's just that the angle you're at makes that part of the wall equidistant to the icicle.
Also people forget that the plane of focus is not flat but rather a curved hemisphere/saucer especially at close distances this becomes a problem. That's why macro lenses cost so much - they are corrected for flatness of field.
If you want to execute shots like this then a macro lens would be more suitable.
I'm inclined to the theory that the focus point used wasn't sensitive in the direction of the edge you were looking at. Remember that only the centre point has a cross sensor.
