lev29 wrote:
So now I wonder, is there a significant difference between the design and implementation of pixel-shift technology by Sony as opposed to that by Pentax and other camera manufacturers? If so, then any response to my original challenge that involves cameras other than Sony is moot.
Another possible question (forgive me for not reviewing and remembering all the hype about the a7R IV,) is how does this compare with any previous Sony camera's use of pixel-shift technology?
If we compare the Pentax method with the Sony A7Riii, they both do four images. Pentax can process in camera and can do this without a tripod, if you desire. But since it is four images, there is no detail improvement. But there is color refinement. To produce an image with correct colors using a single image, there are a number of interpolation calculations that are performed on the pixels to determine what the real color of each pixel is supposed to be. But these interpolations can be off a bit. The idea behind 4 images is to allow each spot in the image to be captured with a red, a blue and two green pixels on the sensor. This changes the interpolation equations making it a more sure thing that the color is assigned correctly - if the camera is held steady.
In the case of doing this handheld on the Pentax, the user is going to jiggle a bit. Now it may be possible image stabilization holds the sensor steady enough that it doesn't matter. But I am guessing that this is dreaming. Anyway, remember the goal is to improve color accuracy. One of the four images will be used as the main image, and the 3 others will be used to add correction confidence to the color computed for each pixel. I am not privy to the actual algorithms, but keep in mind, correct color is what is being pursued.
Olympus does 8 images. This also provides color correction like Pentax, and the Sony A7Riii. But it also doubles resolution in one axis with the half pixel stepping.
And now the Sony A7Riv uses 16 images. Still provides the color correction. But it doubles the resolution in two axis.
Now, whether hand held pixel shift is as good as tripod assisted pixel shift. I have read on other forums where users have compared hand held to a regular image and found not appreciable difference and have compared tripod held to a regular image and found a significant difference. So perhaps hand held pixel shifting is a marketing gimmick to make more sales. And it looks like using a tripod (with IS turned off) is a better way to do this.
And what about processing in-camera? I suspect that post processing is going to be superior. In-camera is certainly optimized for speed as that want that result as fast as possible. And post processing is certainly not fast as it really does a lot of computing to come up with that result.
It would be interesting to see how many users of pixel shifting actually use the in-camera processing versus doing it in post processing. Of course, not all cameras can do it in-camera.