Originally posted by normhead He didn't address Pentax specifically, but he does say somewhere that all brands suffer from out of the box QC issues. The last actual comparison I saw, was years ago, and at that time during the height of the DA 16-50 SDM fiasco, Pentax still had the best QC and endurance by several percentage points. But that's so long ago, I can't remember the source.
As for the brick wall, it's not what I look for in images. 3D rendering is always more important for any lens than flat plane rendering. If you want great flat plane rendering, use a macro. The big problem with this discussion is two different people trying to replicate each others work. In our case, our judgments on decentering are based on a lenses ability to keep both sides of a landscape in focus at ƒ5.6, because for us that's a working set up. We have a ridge near our house on our morning dog walk that makes a convenient test subject. How a camera does at ƒ2.8 on a flat wall, is irrelevant to our shooing.
The problem with so many test sites is, most only test one copy. No one knows where on the curve that copy is, and for all companies "the lens is within spec" seems to be a pretty low bar. the reason I like lens rentals is because he can give you the curve based on multiple lenses, so you can see what your odds are of getting something acceptable. One lens tests without an understanding of where that lens is on the sample variation curve are pretty much useless.
I've read lots of those where I thought, "you should have tested my lens, not the lens you tested" Not to mention that most reviewers are not aware of their own biases. Read Klaus' review of the 18-135. By his own testing it has 17 excellent data points, yet he rates it at 1.5 out of 5, Some lenses rate a whole point more have 3 or less excellent data points. Simply stated for a lot of my shooting my 18-135 kills some of the lenses he rates better, some of which I've owned for direct comparison. Klaus heavily weights lenses on corner sharpness, as if corner sharpness is important for every image. For many images it isn't. (and that was the basis of Pentax's "lenses for the way people shoot pictures, not for the test charts" philosophy. For some of us, excellent centre sharpness is more valuable than edge to edge consistency with a weaker centre. So, based on previous experience I pretty much discount such comparisons.
When I need corner to corner consistency and sharpness, I use a macro, and my own tests suggest my FA 50 macro is equal to or better than the DFA 50 1.4 for absolute corner sharpness and edge to edge consistency, although out of focus areas and transitions may suffer. You have to be sure what you are testing for is relevant to your shooting style. Flat plane corner to corner consistency is not always the most meaningful attribute.
Sometimes absolute centre sharpness is.
Sometimes the best out of focus areas and transitions are what you're looking for.
I agree with all your points.
A brick wall is photo is not the ultimate test of image quality. Its just the easiest way to get an indication of a problem with a lens by looking for asymmetries.
For my shooting, the ultimate test of a lens' performance is to shoot it wide open at a clear/dark night sky. This will highlight in painful detail just about every lens aberration there is. Obviously this would not constitute a relevant lens test for everyone!
If Klaus is the author of OpticalLimits, then I'm sure his review of the 18-135 ruffled a lot of Pentax fans feathers when it was released, and he is not very popular on these forums haha
I have a copy of the 18-135mm (it came with a camera used camera body I purchased). Looking at the QC first, unfortunately my copy is soft on the left side from around 60mm and narrower. In its defense, if you have the light to stop it down to F8, it performs very well from around 21-45mm. More than enough performance for my needs.
In this "sweet spot" it performed better than either of the 16-85mm copies I tested (both decentred however), it also matched the first 16-50mm F2.8 I tested (not perfectly centred), and beat the more decentred other copy of the 16-50mm I also tested. For context though, In this limited range (21-45mm stepped down a lot), basically any OEM kit 18-55mm would be more than sufficient for my needs.
I think OpticalLimit's main issue is that after your get past 50mm (even with a well centred lens), the perform drops off so significantly. On one hand, that's the price you would expect to pay for the convenience of the zoom range. On the other hand, the competitors lens, the Nikon DX 18-140mm, and the Canon EFS 18-135mm STM both maintain a significantly better performance at the long end. I have borrowed a friends copy of the Nikon 18-140mm, and its very impressive optically for its range. It edges the Pentax 18-135 even in its sweet spot, and absolutely blitzes in at the long end (even on the Pentax's "good side" haha).
Ultimately, I still get good use out of my Pentax 18-135. I use it as the default lens on my K10D. I find the lower pixel density of this camera a good means of masking the lens' QC issues. The vast majority of the images I have posted on this forum have been taken with that combination if you want to look them up.
Nice duck photo too! Looking at the exif data, is that a native 860mm lens? I'm still learning about Pentax's lens range. Or is that through a teleconverter?