Much discussion has been poured on DPReview’s opinion that the K-1 II’s “baked in” noise reduction is a problem relative to the K-1, in which RAW output is free of manipulation. One contention might be that to may be difficult to use desktop programs to reduce noise from the K-1 II’s files, as noise reduction has already been applied and the noise “smooshed” to the extent that a pixel-based NR approach is no longer effective.
Let’s take a look at applying an arguably state of the art noise reduction program to the K-1 II’s RAW files, and compare that to the K-1.
On the right, we see that the K-1 file has markedly more detail surviving. Noise reduction for the K-1 II is smoother and more thorough. It’s possible to say it comes down to a matter of preference, but I would much prefer the image on the right.
As usual, there is a caveat to this. The difference between the two images is much less apparent if they are processed with “HQ” denoising instead of PRIME:
One shouldn’t expect anything else, since the desktop noise reduction has to catch up to the in-camera one – apparently this isn’t clearly achieved until you get to something as intense as PRIME denoising. However, PRIME has gotten much faster since its introduction in DxO Optics Pro 9, its only lasting downside being that you don’t get full screen preview so you have no idea what your final image will look like until you export. The workaround for this is to export the RAW to 16 bit TIFF with PRIME enabled and any other basic necessary changes that affect the subsequent versatility of the file (exposure, contrast, highlights, shadows), and then continue to work with that exported image in DxO. That way, you can have a what-you-see-is-what-you-get (WYSIWYG) experience.
Now, if you have an older version of DxO, or none at all, or a slow machine, and you don’t want to upgrade any of those things, the K-1 II with its pre-smoothed files and supposedly better autofocus might still make the most sense to you. Equally, if you never shoot above ISO 400 or so, or you upload your images at a small resolution and don’t ever crop, the noise reduction arguments aren’t particularly relevant to you, and the K-1 II will be an excellent camera for you. Since the autofocus in the K-1 classic actually works just fine, I would probably recommend that camera in all other scenarios.
Update 2019/1/1: In response to comments, I’m now also showing the following result with PRIME denoise strength set to 10/100 rather than the default “auto” value of 40/100. I tried to match the residual grain in the images edited by BigMackCam in this thread.
Other similar tests I’ve done: