Appion

I haven't tracked it down in the code, but I'm sure you're right because I see the same behavior where before doing tilt pairs, the stack is centered.

Centered tilted particles

https://emportal.scripps.edu/myamiweb/processing/stackhierarchy.php?expId=6614

Hi Scott, just for clarification are the zero degree particles not centered or are the tilted particles not centered. In my case, the zero particles were fine, but the tilted ones were off center.

Anyway, have not had much time to track it down.

I finished teaching and got my grant submitted and had a chance to look at this. I repeated what my postdoc was doing but with just one image. I am still getting the systematic offset, but I think I know what is going on. Neil, to answer your question, it is the untilted particles that are offset. I believe that which particles are offset depends on the order in which the images were collected. In my case, the tilted images were collected first followed by the untilted. The offset is happening during the tilt pair picking after the autorefinement step. I haven't confirmed this, but I think what is happening is that the model for the tilt geometry is causing the offset. I believe that the code is doing exactly what it should be doing, but this grid might be ever so slightly bent. Thus the the tilted images are fixed and the zero degree images are being adjusted to them based on the geometry model. Does this sound plausible? If so, I will close this issue.

Thanks for the feedback. There is an X and Y shift it must not be getting applied correctly during the commit for the second image. I have not had time to look at this, but hopefully soon.

Did some testing today and tried to reproduce problem on 2 image set unsuccessfully. I still believe it is related to the shift in the tilt alignment parameters. Introducing a large shift in align and edit tilt picker did not produce problem.
Need to find a way to introduce a large shift into auto-tilt-aligner OR review the code.

Scott did you use auto-tilt-aligner OR align-and-edit tilt picker?

I believe this original problem was related to the specimen instead of the program. The poor centering happened with 3 different samples, then we got one sample where they matched very closely. I believe the original problem was happening because the sample was tilted so the geometry calculation wasn't working out accurately. After preparing a better sample, the problem went away.