I shoot with a very stable microscope focus block, that allows me to make even steps of 0.3 mkm (and less, but so far 0.3 mkm was the least I ever needed, when shooting with NA 0.75). The camera is Lumix g9.
First reason.
Once I received an advice to try turning off alignments. That was on photomacrography forum, given by Rick Littlefield, the developer of Zerene Stacker.
To my surprise, this significantly improved my stacking results on extreme magnifications. (at NA 0.75 significantly, at NA 0.42 noticeably).
At low mags the improvement was hardly noticeable, but it also was there.
And the rules I figured out for myself are: if your images do not rotate (almost every tabletop setup) - turn off rotation alignment, the result will be better.
If your images do not scale (usually when you are shooting with rails and not with AF) - turn off rotation and scale, leaving only XY translation - the result will be even better.
If your setup is very stable - turn off all alignments, and the result will be the best.
Some people who shoot using AF have only scale change in their setup, and stable in XY translation and rotation.
Those can benefit from leaving only "scale", or in some cases of turning off all the alignments as well.
I myself sometimes shoot "fast stacks" like this with AF. Though my main setup is rail (microscope focus block) based. AF-based stacking doesn't afford to use pixel shift - another feature that I like - that allows to get real ~50Mpix of real resolution (or 80Mpix nominal) from the 20Mpix sensor.
Second reason.
First reason was important for everyone, this one is important just for few stereo enthusiasts.
I (and a few other guys) like making "synthetic stereopairs". To stack those, I need to make the alignment externally, and feed the already aligned stack to the software with additional alignment turned off.
Unfortunately I don't have my photo of this type online, but if you are interested what synthetic stereopairs look like, check e.g. this photo on Mindat
by Tony Peterson: https://mindat.org/photo-1022913.html
Or his stereo gallery: https://mindat.org/gallery-4263.html?frm_id=pager&cform_is_valid=1&u=4263&showtype=4&photoclass=0&phototype=0&orderxby=1&submit_pager=Filter+Search
Zerene has a built-in feature of creating these stereopairs from a single stack.
From what I currently saw in the results that Affinity produce, I have an idea (I may be wrong) of how it works.
It most probably uses both of the 2 most popular algorythms for stacking - "depth map" as the main one, and "laplacian pyramids" for "difficult" areas and combines the final result automatically from the 2 methods' results.
None of the other professional stackers can auto-combine the two methods at the moment. There you can only get pyramids or depth map and combine them manually, which is very time-consuming in some cases.
so Affinity has really made a step forward! I think it can even compete professional stackers, if the devs put some effort in adding extensive settings.
As a result the halo artifacts in Affinity are the least of all stackers. It's so cool. I think Affinity should consider giving a bonus to the developers who implemented the above idea of combining the 2 methods in their focus stacking.