Suggested alteration to Class3D defaults

3D Classification
1

Hi,

By default, Class3D recalculates/optimizes scale factors, even if they have been refined in the preceding refinement.

This is usually what we want (if coming from e.g. NU-refine where scales may or may-not have been optimized), but can potentially cause issues if one is classifying after local refinement (where refinement of scale factors may be unstable/inaccurate, and where refined scale factors may already have been inherited from an upstream NU-refine job).

Would it make more sense to have a conditional default for this parameter? E.g. if all scales are 1, refine them, but if input scales are already refined, default to using the input values?

(Also applies to 3D-VA, which does the same thing)

Cheers
Oli

2

Thanks for the suggestion @olibclarke? Did you consider workflows and/or job blueprints to apply context-sensitive Per-particle scale defaults?

3

Hi @wtempel - the issue is that either of those solutions requires (I think) the user to recognize that they ought to alter the parameter (in which case they might as well just do that each time).

Most users will just use the defaults, so I would try to make the defaults as sensible as possible given the input data

Cheers
Oli

4

Thanks @olibclarke . Noted.

1 Like