As for the why of the ISO 64 on the D810 being helpful, I am not a technician, but I do have eyes. IMO, the blacks and shadows at ISO 64 on the D810 are “better.” I felt the same way when the D3x first appeared. I find the “LO” settings on cameras, which I imagine are some kind of electronic baseline, are not helpful in my work. Show me how they are, please.
The Pentax K1 on Pixel-Shift is VERY touchy. In fact, it is a joke to take a landscape shot in that mode in my experience. You don’t just get some vague movement artifact, you get really ugly artifacts, which I am sure some one of you can explain why. I will try it again, just in case I missed the boat on this.
The K1 has a mode for pixel-shift that attempts to compensate for motion, but it is NOT supported by Adobe Camera Raw, and the Silkpix software shipped with the K1 is perhaps the worst software I have ever attempted to use, like out of the 1980s or something. As a systems programmer, I know a little about software design, and this is atrocious. However, some who have learned it (not me) say that it does do something useful with the Pixel-Shift Mode that includes movement.
My biggest gripe with the K1 is how difficult they make it to use LiveView and see (in real-time) changes (via dials) you make with the shutter and aperture. It kind of can be done, sort-of, but obviously no thought was put into making ease of use for fine focusing a priority. Their approach is itself a workaround for the fact that they did not consider this aspect of photography.
I also find Pentax camera supporters fierce and unrelenting, so just comments like these can upset them. This has been my experience. I kind of see what they are about, because the K1 is an interesting camera, with lots of thought put into the features. And they are well built and water resistant. I watched a video of someone holding at K1 under the waters of a rushing stream, etc. The Pentax support forums are anxious to help and knowledgeable. It is almost as if they hearken from a bygone age on the Internet.