Quote Originally Posted by alex


I think the thing I was most surprised about was the inconsistency in the different shots. I didn't think that was possible.


So you would both say that these are pretty definitive results?


I would not say these are definitive results, IMO. The problem is the set up of the test. You may know that you are intending to focus at the number 4 (and with Live View you certainly can), but how does your camera know that? The actual AF point is larger than the little box in the viewfinder - so, the image element that's being used by the AF system is not the, "intersection of the horizontal and vertical lines right next to the number 4," but rather it's the two vertical edges of the measuring tape against the coffee table, passing through the entire AF point. The fact that those edges are at an acute angle to the camera, and the fact that there are actually two parallel edges, is a challenge for the AF system and that's what accounts for the inconsistent focus you're getting. Most AF testing methods get around this issue in one of two ways - the free 'paper' target which Denise has used recently (focustestchart.com) has a focusing target that is a single line with sufficient blank space around it, and that line is oriented horizontally on the page, so the angle of the target relative to the camera doesn't matter. Other testing systems (e.g. the LensAlign that I use) have a focus target with starbursts that is oriented parallel to the sensor - so, the ruler is at an angle but the focus target is not (an early version of the chart from focustestchart.com actually has a cut-out part that you taped to the current flat paper to provide that parallel-to-the-sensor target). Also the starburst targets (like concentric circles targets) work with AF sensors regardless of sensor orientation.


Denise - thinking about this issue brings a potential confound to mind with your testing as well. Like Alex, with the 135L you're using a lens faster than f/2.8, meaning that you're able to use the high-precision capability of the center AF point. However, I don't think you're actually using it with that chart. The focus target line is horizontal, meaning it's going to optimally activate only the horizontal line-sensitive part of the AF point. Your 7D's f/2.8 sensor is a pair of diagonal lines, meaning a thin black-on-white line oriented horizontally is not going to activate that sensor. Likewise, Alex, if you were to use that chart, despite your XSi also having a high-precision f/2.8 center AF point, you wouldn't be using it either - for your camera, the f/2.8 sensor is a vertical-line sensor that ignores the horizontal line on the chart. So in all cases, that horizontal line is only going to activate the f/5.6 horizontal line-sensitive part of the cross-type sensor in the center AF point. You may be wondering - why did the guy make the chart like that? I would guess it's because he designed it to test AF on Nikon cameras, specifically the D70. Nikon bodies don't have the high-precision AF point capability - their cross-type sensors are f/5.6-sensitive in both orientations.


Alex,bottom line is, if you have a problem, the test you did isn't the best way to show it. You could probably easily put together a modified setup on your coffee table, leaving the ruler but adding a target. Try printing something like this image, then propping it up so it's approximately parallel to your sensor (taped to a building block like my daughter plays with, for example). Just make sure you know which line on the ruler the target is centered on (you might need to prop up the ruler so it's height is aligned to the target as well). Or, you could use the older version of the test chart with the cut-and-fold template, and tape that starburst image to the focus target on the folded piece.


If you do have an AF issue (which seems likely based on your statements about elss than desirable sharpness with your images), Sean correctly ID'd the possible steps - your best bet is to send body and lens into Canon.


Hope that helps...


--John