Originally posted by jaad75 Yeah, it's obvious, but it looks like you don't listen what I say. Having the same noise level as the K-5IIs on the web size picture (or let's say A4 print) is not an improvement for me. It is a stagnation. Of course, slightly more pixels (and hopefully detail, cause it's not as obvious as one might think) gives slightly more room for post-processing (including NR), so there might be a little improvement in the end, but it's still no improvement on the hardware level. And moreover, the higher noise floor means less dynamic range - I mean the real dynamic range, not the one the DxO shows you after their 8MPix "normalization". I would be happy with the same per pixel noise as the old sensor - that would mean that the technology was finally ready for that jump in resolution.
If the S/N ratio stays the same but you increase the number of "pixels" by 50% you are improving IQ. I don't really care about the hardware level. most of the improvement we have seen in the last 10 years have come RAW converters in image processors, not sensors. What I do care about is image quality. Images are sharper, have more detail and the noise that I am seeing (up to 6400) is very fine grained and easy to handle in LR4.4. The A3+ test prints look great and the B&W prints are excellent.
It doesn't look like the K-3 has the processing latitude of the K-5, so people had better get their exposure right. When exposed correctly it appears the K-3 will produce higher quality images than the K-5. I really don't care if that is due to hardware, software, or magic elves.