Local refine NEW - job runs but all results remain empty


I was just trying the new local refine job on v 3.3.1 and all plots stayed empty, although I got no error message.

The resulting map was just a cube.

I just started a legacy local refine job and it seems to work correctly so far with the exact same inputs (particles and volume from a homogeneous refinement, and a mask I used before for local refinements).

Any suggestions?

Kind regards,


@cryo-lg I wonder what happened there. Please can you post:

  • screenshot of the Overview tab showing an empty plot and some surrounding log context
  • screenshot of the bottom of the Overview tab
  • screenshot of the expanded Inputs
  • screenshot of the outputs tab
  • text from the file job.log inside the problematic Local Refinement (New) job’s directory

Hi there,

uh… in the heat of the moment I got mildly annoyed after the n-th attempt and deleted all jobs from my workspace. :slight_smile:

I am starting one again now, with the exact same inputs as before. We’ll see how it goes.

I’ll update in a bit.

Kind regards,


EDIT: seems like it’s doing it again, as the alignment maps are empty. See attached.
Let the “run” finish and I’ll upload everything you asked for.

EDIT2: in the meantime some more context

EDIT3: just to reiterate, the legacy local refine ran to completion with the same inputs and great results.

Hi @cryo-lg,

It looks like you are using a binary mask. The local refinement job requires softly-padded masks, as detailed in our guide page for Local Refinement. Please let me know if this issue persists after using a padded mask; also feel free to refer to an older thread on this topic here.


Hi @mmclean ,

the mask should be soft padded (I’m currently not connected, but from memory I’d say 5 dilation and 15-25 padding - downsampling limits data res to 4.6 Å).

As I said, this is only a problem with the new method. Legacy works fine with the same inputs (see screenshot).

I will try making a new mask from scratch, and will update.

Kind regards,


Edit: @mmclean I see what you mean, my previous screenshot shows the mask without padding… ??? I’ll log in and check.

Edit2: @mmclean , ok I see a possible problem I caused: I padded the original mask (512 box size), and then downsampled it. I will now re-pad it and see if it changes anything.

Any idea why legacy doesn’t seem to mind?

Thank you @mmclean !
yup, it’s running with the re-padded mask.
My bad for not wondering about that first.

Kind regards,


1 Like