Using downsampling to crop particles requires entering a Fourier crop (bug?)


When using Downsample to only crop particles (no binning), I get the attached error if Fourier crop is set to none.

I can get around it by setting the Fourier crop value to the same size as the box I am cropping to, but this shouldn’t be necessary I think (and I hope it isn’t performing some additional operation when I do this - I just want to crop the RBMC generated particles around a local center).



Traceback (most recent call last):
  File "cryosparc_master/cryosparc_compute/", line 95, in
  File "/home/user/software/cryosparc/cryosparc2_worker/cryosparc_compute/jobs/extract/", line 894, in run_downsample_particles
    assert N_out <= N_crop, "Fourier crop size must be smaller than cropping box size"
AssertionError: Fourier crop size must be smaller than cropping box size
1 Like

I saw the same thing. When entering a Fourier crop value the same as the real space crop it gave me the output I expected so thought nothing of it. I don’t usually downsample particles after extraction so don’t remember how previous CryoSPARC releases did it (I usually Fourier crop during initial extraction, do early processing and re-extract - I rarely decide my box size is too large, usually I decide it’s too small! - and if too small I prefer to re-extract rather than pad out.

Generally I do the same thing - but if I want to refine subparticles after RBMC, just cropping around the local center from the RBMC corrected particle image is convenient

1 Like

Ah, I see. Yes, with RBMC that makes sense. :smiley: Only an issue since RBMC release… :wink:

Hi @olibclarke,

Happy New Year! I’m just replying to clarify that when the Fourier crop parameter is set to the same as the Crop/pad to box size, the effect is to only crop in real-space – this is the workaround mentioned in your original post. We have made a note to correct the behaviour when Fourier crop parameter is not set, so that if just the Crop/pad to box size parameter is set, only real-space cropping will be done.


1 Like

Hi all,

Just a brief update to let you know that as of CryoSPARC v4.5, the behaviour of downsample particles is such that:

  1. If only the realspace crop size is provided, the particles will be cropped to that size (and not Fourier-cropped)
  2. If only the Fourier-crop size is provided, the particles will be directly downsampled to this box in Fourier space.
  3. If both are provided, realspace cropping will proceed before Fourier-cropping.

This should eliminate the need to specify both box sizes unless the behaviour in 3 is desired.

In addition, we have added a new parameter “Desired approximate pixel size”, which can be specified instead of the Fourier-crop size; this will automatically calculate an appropriate box size such that the final pixel size is as specified.