Class3D output with highpass filter - query

Hi,

I sometimes find it useful to run 3D classification with the highpass filter set to e.g 15Å, in order to separate subtle conformational differences.

I notice that in such cases, the difference map slices in the output look pretty uniformly negative/blue, even when class3D is clearly separating genuine differences as determined from the volumes:

Presumably these are generated by subtracting the output class from the consensus volume. However, it seems like the output classes are highpass filtered, but the consensus volume is not - could that lead to the difference map slices looking a bit odd/uninformative?

Cheers
Oli