I am re-running some analysis on a dataset that classifies and refines well to sub-3A resolution in order to test the new cryosparc functionalities in V.2.8. However I got a weird 2D classification error that I have not seen before - the 2D classes are masked completely out with circular white or grey masks (depending on the iteration) even if the outputs from the template picker and local motion correction appear normal and no error message were encountered before. In terms of settings I am using settings from a previous job that ran fine in V.2.5. Any idea what this could be due to?
Did you happen to run patch based motion correction, patch based CTF estimation, particle picking, and then local motion correction before 2D classification? I had the same issue when I ran per particle-based local motion correction of particles. My understanding is that local motion correction is no longer needed.
@PVK could you report whether you had the “Circular Mask” option enabled in 2D classification, and if so, what was the circular mask diameter (and the box size and pixel size of the particles?) We think this error may be connected to circular masking doing the wrong thing if the circular mask diameter is larger than the box for some reason
