EFS 17-55 USM corner blurriness: experts' opinion needed !

70d IF wrote:

So my question is: are Canon testing criteria too low or my expectations too high ?

Personally, if that sample is representative, I wouldn't consider your expectations too high. The main issue with decentering is that while your wealth is amputated by the same amount regardless of the copy you get, the value you get in return can wildly vary.

However, for a zoom lens, with their complex and difficult to adjust optical formulas, I wouldn't evaluate them from the extreme corner, which is likely to be irrelevant to photography except for specific applications. I'd rather focus my attention on 85% or so of the frame wide open. As an example, my second Olympus 12-40 is decentered, but this only shows in the extreme corners wide open at the widest focal lengths. As soon as I start to look into that 85% frame coverage, it's fine. The first copy was utter garbage though.

Regarding your lens, it seems to me that it stays decentered quite a while when moving towards to frame centre, but I wouldn't be able to be sure of it.

Please give me your opinion.

As previous posters have suggested, the "brick wall" + 180° upside down test is a good start to analyse a lens for a number of decentering issues, especially if it's at least moderately rigorous (example : same exposure).

Note though, that, especially for zoom lenses, performance may vary when the zoom is held in the upside down position (barrels may move inside). Not that, so far, I've seen this phenomenon render the test above totally inaccurate.