Hello,
I’m using Patch motion correction (multi) on my movies. The rigid motion plots show some trajectories; however, the plots showing patch motion show only dots which makes me think there’s not patch correction happening. I would appreciate if someone has any suggestions or can comment on it. Please see sample pictures attached.
Best,
Christian
Hi Christian,
Could you let us know some more information about your movie data? Camera type, magnification (i.e. raw pixel size), super res or not, etc?
If the data was collected at high magnification, the algorithm may be auto-tuning itself to not allow very much anisotropic deformation within the extent of the field of view. It is also possible that the algorithm is detecting very minor anisotropic deformation that does not show up in the plot.
Hi,
Thank you for your response. The movies were collected on a K3 direct electron detector at magnification of 60k and raw pixel size 0.76 Å. Hope that information helps.
Christian
Hi @Xian,
This is pretty typical magnification and data so nothing suspicious there. It may be possible that there isn’t any substantial local motion in the data. You could try running “local motion correction” on a few micrograph with particles after picking to see if you see any substantial trajectories there? Let us know what you find. (Note that the multi-GPU version of local motion does not create any plots)
Aha - these plots help. Clearly there is local motion, but the exposure curves indicate that you must have set the exposure dose rate to a very low value. Patch-motion autotunes it’s temporal smoothness priors based on the total dose of the movie.
What is the total exposure for the movies? Perhaps at import time you put in the dose per frame rather than the total dose? The total dose should typically be 30-100 e/A^2
Yes, that’s what happened, I put in the dose per frame. I will find the right value and run again. Thank you for your help,
Christian
