I've been experiencing this problem since the first picture taken with my K20D many weeks ago.
I wanted to give it a thorough workout, testing as many variables as possible before sending it in to Pentax. This is what I've found:
1) image samples here:
Flickr: My K20D's Photostream i wonder if this is similar to what you've seen?
2) There are *no* truly guaranteed consistent factors that I can find. I've tried and kept track of so many variables...high quality memory cards, cheap cards, battery grip, no grip, high ISO, low ISO, cold camera, warm camera, fresh battery, almost dead battery, bracketed exposure, all drive modes...literally 10 hours or more of dedicated trials. The best I can do is state that the *possibility* of an occurrence
is maximized by the following on my camera.
3) High or Low continuous shooting Drive Mode + ISO 400 or higher + Shake Reduction. about 90 percent of the time these settings will trigger the freeze. about 90 percent of the freeze-ups will result in the white image with pink band (or some other extreme banding result).
4) almost *every* instance of a lockup is accompanied by the sound of the Shake Reduction motor being active *even if Shake Reduction is not turned on*.
5) These are only the most consistent factors. The freeze can happen with none of these factors in place, and rarely all of these factors can be in place and a freeze will *not* occur.
I've spent an insane amount of time trying to correlate all of this, and I'd be pleased if someone could at least take a few steps following me down the path to crazyland to verify some of the points.