18-135mm DA WR

You compare the result with the correction switched off and then with the correction switched on, using the same lens of course.

IMHO you are over-thinking the situation Stefan.

Well over-thinking it is good, it keeps you sharp

The problem lies that the corrected photo and the non corrected photo holds the same information because it's the same photo, only the corrected one is a bit stretched in certain ways.

In a simpler explained way you're suggesting to compare a 10mp photo and an upscaled version of the same photo to 20mp to compare the difference the amount of pixels on the sensor made.
You see that it doesn't work?
You need to compare a native 10mp with a native 20mp photo to see the difference.

So to see the difference the digital correction make you need to compare it with an optical correct image since you can't use the same source as the standard to compare it with.

The digital correction will not add or lower the details in a photo but it will have less details than a photo taken with an distortion free lens.
The same way as a upscaled 20mp photo doesn't have the same detial as a native 20mp photo.

Sorry Stefan to disagree, but the uncorrected and corrected images are exactly the same file size.

Because the amount of pixels are the same.
An upscaled photo of 10mp to 20mp would also be the same size as a 20mp photo but that doesn't mean the upscaled photo has more details.

Well I don't want to go round in circles for ever with this. I think you are misinterpreting what you think the K-5 is doing with these images.

It warps out the lens distortion, simliar to what lens correction software does on the computer, DxO for example link link

Too look at an extreme example, look at the second fisheye photo in the last link of the photo in side a camera store.
The glass cabinet on the right side gets 2 times as long, so it has increased in the amount of pixels but not in quality/detail and nor does it lose any quality, the quality (original pixel) simply get stretch out.
If you would have shot the same photo with a normal lens the cabinet would be the same size as in the corrected photo but with out the pixels stretch out so the details would be 2 times as good.

I know the fisheye is an extreme example but it's the same lens correction does with a normal lens but far less.

Not sure about in-camera lens correction but I use the PTlens software from time to time and it's execellent, I certainly don't see any image degradation

Simon

The in-camera correction basically does the same as PTLens, so results should be similar.