Leginon

Yes, NYSBC use this option all the times on Krios/Glacios. The quality is good enough for low 2.x resolution ApoF without making group-divided aberration correction in Relion 3.1. Others can comment on the specific tests more than I do. Depending on the setup of the scope and grid, up to +/- 5-7 um offset is used. You will squeeze out a bit more if you group them and refine further since the calibration is not perfect.

Thanks, Anchi.
When you use +/- 7um, that probably means 5x5=25 holes in a hole target image for R1.2/1.3 grids. How many images do you typically get every 24 hours with that setting?

Depending on which camera, scope configuration and how much pause time you use for the settings.

For K2 counting mode, (40 x 150 ms frames) 25 holes scheme can get to about 170 images per hour without changing any default pause time. The record is 70 holes per stage move to 15 um, and that ends up to be 4250 images per day (no energy filter, special SA mag at < 600x)

For K3 counting mode (50 x 39 ms frames), and the same is about 200 images per hour.

It turns out that as the detector gets faster, other factors are limiting the speed at per image level. When this option is used, we start to make a few changes in the wait settings:

1. Scopes like Krios/Glacios are best used using the factory lens normalization. These takes a long time to do but it means once the preset change is complete, it is stable right away. So I have made our operation staff to reduce the wait time in PresetsManager to as low as 0.2 s

2. When the aberration correction option is used, it makes lots of small adjustment locally on the lens and deflectors after preset change where we normally pause before acquiring image. Such small change settles quickly, too. The Exposure Node wait time before acquiring can be reduced to 0.2 s

3. Leginon transfer the sum image array from camera to linux workstation and do gain/dark correction there. That can take 6 seconds on K3 super-res images, about 3 seconds on counted image. I added an option to let go the immediate feedback in favor of speed by reducing the dimension of the array transferred to minimal but still carries the statistic characteristics. This cut the time down, too. This option is available in the feature complete but not fully documented myami-3.5. I have to finish the document.

4. A few low-level calls were modified to speed them up.

Right now on GIF+K3, the more typical is averaged at 10-11 s per beam-image shift images (40 x 50 ms frames). 240 images per hr including overhead of autofocus on most scopes. Some of them were faster than others if we can get an undistorted hl mag images at lower magnifications. With faster computer, it also cuts down per image time.
One such case got 30 en images/stage shifts and 6-7 s per beam-image shift move (38 x 39 ms frames), 400 images per hr including overhead of autofocus/liquid nitrogen fill.

These do not include grid screening and picking your favorite targets. Queuing at high magnification slows down data collection because it has to return and adjust to make sure it gets to the right place. These numbers come from queuing at square level and auto-target at higher mag.

That's a lot of info'. Very much appreciated.

Hi Anchi,

If I could ask a related question - we have a Krios G4i with "aberration-free imaging". Can leginon use these look-up tables provided on the Krios PC so that beam-image shift data collection? Or do we need to set this up within Leginon? I know that this can be done within Leginon but wanted to know if we can take advantage of the settings already put on the instrument.

Thank you,
Mike