pixel size relevance

Hi Tim,
I'm considering an unblur wrapper inside relion-2.0. I was thinking of only using your frame alignment method, and leave the dose-weighting up to relion's polishing. Also, I was thinking of avoiding any of the advanced settings. In that case, does the pixel size (in Angstroms) parameter actually do anything else than providing the correct units for the plots? Is there any internal weighting that would be affected by it? If not, I was thinking of setting it always to 1.0, and that way make plots of shifts in pixels.
Thanks for your time,

PS: is there any advanced setting that you regularly change? If so: which one and in which cases do you change it?

Hi Sjors,

Sounds good! Apart from the exposure filtering, the pixel size is basically used to convert the other inputs. B-factor (which is input as A^2), the min/max shift limits and termination threshold which are all given in Angstroms. So you could just set it to 1, and consider everything in pixels, although for the B-factor that could become a bit of a complication for weird pixel sizes. The default is a 1500 A^2 low-pass filter, basically you usually need to low pass filter the images pretty heavily.

The defaults tend to work quite well, so i rarely set the expert options, but for problem cases I tend to play around with the Outer radius shift limit and B-factor from the expert options.