By default, 2D classification rearranges the display of classes in terms of population at each iteration.
This has the advantage that the most populated classes are near the top, but the disadvantage that it makes it hard to monitor the evolution of particular classes over time.
Would it be possible to add an option to keep the classes in a fixed, numerical arrangement? so that class 1, for example, is always displayed in the same position at each iteration?
It would also be helpful if one could preselect classes for select 2D while a 2D classification is ongoing. Often, it is already apparent which classes are good a few iterations in, and the last iteration of an initial classification can take quite a while (several hours). It would be nice to be able to preselect which classes to take from the classification into the next step in the pipeline (e.g. another round of 2D, or ab initio), and then have that next job run after 2D completes without further user intervention.