Denoiser crash edge case

Features and Functionality
1

Hi CryoSPARC team,

I ran into a crash with the micrograph denoiser which I think should be recorded but equally I think is not a priority issue:

If “Output denoiser training data” is disabled during motion correction, then the Patch Motion Correction job is cleared and “Output denoiser training data” is re-enabled and Patch Motion Correction re-run, Denoising will fail reporting that there is no training data.

Obviously this is not a scenario which will happen usually, but it’s curious that the re-run of Patch Motion does not appear to be honouring the change in input parameters.

I’d also like to support @olibclarke’s request (which I missed originally I think) for a specific output slot for the training micrographs.

2

Hi @rbs_sci ,

Sorry for the delay. We’ll look into this and let you know what we discover.

–Harris

3

Hi @rbs_sci, re-reading your report, I’m guessing that after the PMC re-run you didn’t clear and reconnect it’s output to the denoiser job? Unfortunately there’s a known issue right now where pre-existing connections don’t pick up new output groups, even if the upstream job generated them correctly. Try reconnecting the PMC output and let me know if that fixes it. It’s on our radar to fix this.

–Harris

4

You’re absolutely correct.

Thanks for explaining, @hsnyder!

1 Like