Hi,
Subset particles by statistic has an option to split based on per particle scale, but does not work on output from a Class2D run where “Minimize over per particle scale” is activated.
The resulting error suggests that it only recognizes scales refined during 3D refinement, not 2D classification.
It might be helpful to be able to inspect particles with unusually low or high scale factors after 2D classification.
Cheers
Oli
Hi @olibclarke!
You’re right that currently Subset Particles only uses the per-particle scale from 3D jobs, not 2D. Do you have examples of datasets for which the 2D per-particle scale is useful?
As an aside, the scales are calculated per-class in 2D Classification, so a given 2D scale value may or may not be comparable across classes.
Hi Rich - I didn’t actually realize these were refined per class and not per particle! If it was per particle, I was thinking that anomalous refined scale factors might be a useful way to identify junk particles, but if it is per class they won’t be useful for that, so no worries!
