With the GFX100, Fuji did a lot of good things, but their implementation of pixel shift to me is far from the best. And what suprised me on this issue that Fuji was the latest to bring a solution to market and thus they plenty of time to study other companies solutions (Panasonic, Leica, Sony, Olympus, and Pentax).
Fuji went for 16 shots, and 400MP, (why not 8 shots and 200MP? also). 16 shots is a bit too many to capture in the field, as will never get 16 shots clean without some motion. Fuji's software can't accommodate any motion at all without artifacts. Similar to the a Phase One Frame Average capture albeit the Phase shot will have more blur than artifacts. I can't believe that Fuji could not allow for 8 shots, as most other cameras doing this type of shift have 8, and 16, some only 8.
Fuji has no ability to mark the files, place the in a separate folder, etc. so you have to keep track of them manually. This may have changed with the 100s but the few tests I ran with the 100 I did not see any clean way to tell where a pixel shift group started or ended. So I made a point of shooting black frames before and after.
Fuji can't process in camera, so photographer is forced to use the Fuji App, which also is less than stellar. Panasonic and Leica both allow for the processing to be done in camera (taking less than 16 images) and both cameras do a excellent job on creation of their 187mp image. Subject motion can also be problematic, but there is an option to correct for motion artifacts an in at least 70% of the shots I have taken with these cameras, slight motion can be accounted for.
The more shots required also means taking an inordinate amount of time, so if you are required to shoot a 5 second shutter speed for each shot, you are waiting a total of 80 seconds, well over a minute. Again the ability to take 8 would be an advantage. The need for 400MP in output printing has yet to be a factor for me.
Resolution increasing software solutions are gradually getting better, Toapz Gigapixel AI and now Adobe's Super Resolution can get some very good results. Both are very depending on the original image that is being considered for up scaling.
Fuji now has an option to capture "real color" by using fewer images and defeating color aliasing. This came out with the latest version of the Fuji combiner software, (I have not personally tried it and or read anything thing about it). However it appears that it still requires 16 total captures, thus all of the previous issues still are present. From reading about the process, it seems that the combiner software only uses 4 images of the 16.
Detailed reviews have also shown that the current Fuji pixel shift images have a "combing" artifact issue which does lead to both blur and detail aliasing around certain subject matter. Apparently the Panasonic multi shot solution also had a similar issue, which was fixed in camera firmware.
Fuji had over 1.5 years to work on this, and so far, it's not shown me that for field work there is anything to gain by it's use. Sure for Art Repo or interiors where the conditions can be carefully controlled, it makes sense, but that is a very limited use.
Paul
