Originally posted by jeffshaddix Thanks for the tidbit. A 20MP sensor would be good. Do remember that in one of the many interviews, Pentax said they were looking into other forms of sensor technology to enhance detail beyond removing the AA filter (and no, I don't think they were referring to increasing MP or sensor size). Foveon is a possibility, as well as other non-bayer sensors. It'll be interesting to see where Pentax goes...
I do remember reading that Pentax alluded to this. Personally, I do not think that Foveon has all the qualities that Pentax really needs, however it does have a number of positive qualities. I think that these positive qualities can be replicated by Pentax fairly easily, but would be somewhat difficult for either Canon or Nikon. Pentax has all the pieces available to create the same type of ultra high resolution shifted images as Hasselblad.
They can shift the sensor, even rotate the sensor, taking multiple images with a 1.5 pixel shift, and then combine them together (in camera HDR processing). Thus an APS-C sensor based camera becomes the ultimate high resolution landscape camera with exceptional color accuracy. Using this approach, the K5/K5II/K5IIs/K30 could produce 64MP images in this mode, using the current 16MP sensor.
This method can also provide one additional attribute. By shifting the sensor just by a bit more than a pixel, the Bayer pattern of the photo receptors (the pixels) can align (or come close to aligning), in a fashion similar to the Foveon sensor. In this way, Pentax can maintain their Bayer based sensor with all of the positive attributes (wide ISO range, etc.), while also potentially matching the positive and desirable Foveon qualities. The downside would be the need for a tripod and the taking of 4 to 6 successive frames, back to back (in an automated fashion). For landscapers this would not be a problem at all. Of course there would be uses that this would not be possible - i.e., sports and action images. Essentially anything that moves - however this mode would not need to be active across the board.
Actually if Pentax applied this in a similar fashion as bracketing (a special case of bracketing), just take and store the 4 or 6 images (as RAW or JPG) and let post processing perform the stacking, would certainly provide a much higher quality result. Just tag the EXIF with the amount of pixel shift and direction (within each frame), and the stacking would be easy and quick.
I can see folks then wanting to do both the shifting and the bracketing so as to capture the additional dynamic range.
The only downside is that the Forum is pretty fixated on a FF sensor and a higher resolution sensor to match or exceed Nikon, and thus would probably perceive this as lagging behind, loosing even more ground to N&C.