Leginon

Could you post a screenshot of the image with the defocus, pixel size, image size of the preset used? This calibration needs preset similar to manual focusing because it looks at the power spectrum.

The other thing would be helpful to me is a screen shot of a tableau from "Beam Tilt Image" node using the same beam tilt and preset you used in this calibration.

Thanks.

Calibration will indeed fail with any CTF estimation failed. That part is not surprising.

I got it to work on the F20 with a Gatan camera. The problem seemed to be the image configuration. Switching to 512 4X binning worked. Next I'll try it on the F30 with a TVIPS camera. Thanks!

You are right it is very picky. It does not work for the same settings but 1024 4X binning. It seems to need at least 3-4 rings to get the calibration.

Beam tilt coma calibration failed with our F30 and TVIPS415. I used a 512 4X binned with a pixel size of 6 Angstroms and FC preset defocus is -3.28 microns. A snapshot of the image and error are attached. There was no tableau. I am also attaching the same configuration showing the manual focus FFT. There is notthing in the tableau.

Bob,

Sorry for not responding. This is a hard one to debug since I don't have a simulator for this function and scope time is rare at the moment. It did help either that I didn't do good enough error reporting in leginon log. I have create a bug issue for it #1560

If you don't mind helping me, please update that issue and attach a pair mrc image of one such failed images with pixel size, and HT value so that I can use for debugging. Thanks.

Anchi

Anchi,
Where can I find the MRC images that are used for the calibration? I do not have access to the scope until later this week so want to look for them in the database. Thanks
Bob

Unfortunately we didn't make a check box in Leginon for saving images acquired through calibration nodes. All I can ask from you is an mrc image saved of the preset that is used in the calibration. Image saved by any of the acquisition node, say Exposure, with the preset you use in calibration set in the settings is good enough. If you are reaquiring it, you can use your camera software to do that as well. I just need the image. I used the same ctf estimation function yesterday here and it didn't give me the kind of error you had. It indicates that it might have to do with the range it needs to fit.

Anchi

Sorry for taking so long.

I've updated issue #1560 about error reporting. Still can not reproduce failure of ctf estimation on the test image you sent.

The behavior that no tableau is produced when the ctf estimation fails is actually intentional. The calibration is only meaningful if ctf estimation of all the beam tilt images are successful. There is no point in producing the tableau if any ctf can not be estimated by the algorithm. If tableau is what is desired, "Beam Tilt Image" node should be used where it will display the tableau without estimated value shown if estimation failed.

On the other hand, if you still want to do this calibration, you probably should use smaller pixel size and lower defocus. The coma effect on defocus and astigmatism are not defocus dependent. Therefore, for, say 0.005 radian as used as default in the "Beam Tilt" and Cs of 2 mm, defocus change relative to the untilted one is always only 0.1 um (-2*Cs*tilt^2) regardless of the defocus appled to the scope. Between the two intentional 0.002 misalignment, the defocus difference is going to be -2*2000* (0.005+0.002)^2+2*2000*(0.005-0.002)^2 = 0.16 um. Since you (and the algorithm) will be able to see the absolute change better if your appled defocus is closer to zero, in the same way that we detect astigmatism better when we are close to zero defocus, the alignment and calibration will perform better at closer to zero defocus. For 200 kV instrument, I typically align at 0.3-0.8 um defocus with binned pixel size of 1.6 A/pixel (512 binned by 2).