I think your examples look sharp, but it seems that the Sharpness (+2) parameter doesn't change much?
Could it be a bug in the software? Phil over at dpreview also says that using the sharpness doesn't make the JPGs sharper.
I don't miss any sharpness in your examples, but if you look the dpreview JPG examples, you'll see that the K10D example isn't as sharp as eg. the D80 example.
K10D:
D80:
Comparing RAW image from the K10D with RAW iamges from the rivals, the Pentax comes out best in (almost?) all the examples I've seen, which is why this is surprising me:
Pentax K10D Review: 24. Compared to...: Digital Photography Review
This is a comparison of resolution between four cameras, but are based based on JPGs, in spite of the fact that the reviewer knows that JPGs conversion is K10D's weak point in terms of sharpness. It would be interesting to see a resolution test based on RAW images (which are processed only once, not twice), because that would say
more about resolution and the cameras abilities to capture images, and
less about the algorithms used to convert the images to JPGs.
"Not that surprising to see the K10D under performing in our resolution chart shot, this due primarily to the already discussed lack of sharpness in edge detail (be this due to poor demosaicing or unusual sharpening)."
At the bottom he adds that "A RAW converted resolution chart exhibits about 2200 LPH in the horizontal direction and 2000 LPH in the vertical direction.", but doesn't say anything about how the competition performs in terms of resolution/RAW.
The camera that has the lowest solution of the four (Canon 30D), gets this comment in it's own test: 'Excellent resolution' (the 30D JPG resolution is 1850 x 1650, the K10D JPG resolution is 2000 x 1800).
To write that the image processor is 'unable to deliver crisp sharp edges' isn't correct, even when followed with 'better to shoot RAW'. The image processor may or may not be able to deliver crisp sharp edges, but the current firmware are delivering
JPGs with less sharp edges than eg Nikon d80.
RAW pictures are also processed. (Look
here: "Also, RAW does not mean a real unprocessed RAW. It is the data created as an image and each maker's image creation philosophy is built in"). Maybe they are not processed by the image processor (?), but it's quite clear that the K10D is able to deliver crisp sharp edges. Pentax may loose sales on people who spend less time or reading reviews thoroughly, especially due to dpreviews 7.5 (out of 10) for K10D's image quality (the image quality of the Canon d30 is described as excellent). I hope dpreview will consider editing their article because the conclusion gives the image quality isn't good.