Leginon

The first test I would suggest is to take references out of the question. Take the same image in Correction Node and in Raw so that it does not apply the references.

Second is to change the mode in dmsem.cfg

if you were using

DM_PROCESSING = gain normalized

the gain normalization comes from DM.

if you were using

DM_PROCESSING = dark subtracted

switch to the other one and see what it does.

The observation was only on sampled area which is the most strange part because the pattern you described is most like asynchronization which should show up in bright images at empty area as well.

Hi Anchi,

Thanks.
Before I used "dark subtracted", and I changed to "gain normalized" this time in dmsem.cfg
On an empty area, I re-did the dark/bright reference for super-res, and the bright reference shows strip feature, normal.
Then I re-took a "corrected" image, no strip feature, gain correction working(see attachment "empty-area-20190814.png"), normal.
Right after this, I JUST moved the stage to a cross-grating area, and also took a "corrected" image, the strip feature appeared (see "cross-grating-area-20190814.png")
And I went back to take a image in DM, again normal.
Yes, it's weird to have empty area gain corrected correctly, but failed for sample area.
Somehow the gain correction is not working right on sample area, but I have not any clue.

Best,
Guang

I meant to have you taking Raw image, not corrected image in Correction node. That said, with dmsem.cfg set to "gain normalized", the gain correction in Leginon is skipped, regardless.

By the way, I am a bit confused. If you see the problem only with cross grating in the beam, where is the cross-grating pattern in the first attachment ?

With what you have tested so far, I have to guess that there is an electromagnetic interference that is not directly leginon related. Does this go away with another grid ? Something like a carbon grid ?

One possiblity is that the image shape that is returned from the camera is not what is requested.

Could you also check if the python terminal to see if it print out the warning that image is reshaped.

Hi Anchi,

The empty area and cross-grating area I mentioned, actually are both using cross-grating grid; one is in a broken square, and the other is in a square with latex beads.
The "empty-area-20190814" image was taken as a corrected image after dark/bright correction in Leginon, which was in a broken square.
Right after taking a corrected image in a broken square, I moved the stage to an intact square, and took a corrected image, that's the "cross-grating-area-20190814.png"
-----------I took corrected images, even though you wanted me to raw images.
Since the mag is to give about 1 ang/pixel pixel-size, it's supposed to not see the cross-grating feature (that 0.463 um spacing), but latex beads feature, which I never saw.
The feature in the first attachment "cross-grating-xxx.png" is not feature of cross-grating, but the feature of gain-reference pattern.
When I took raw image in Correction node, it still has gain-reference pattern but no latex-beads feature.
Does this mean that's could be something wrong in our configuration files, or it's because of the wrong-returning image?

Back to the point of "the image returned is not what's requested", I more lean on this one.
Since the location is in latex beads, but I don't see any feature of beads, but some feature like gain-reference pattern.
For the python terminal, I did see some warning about image reshaping (see attachment).
Any suggestion what to check?
Thanks.

Best,
Guang

I don't see latex bead pattern in cross-grating-area.png, either.

Do you see the cross-grating pattern at lower scope mag with the same camera configuration ? If you don't, we are looking at a scrambled image from reshaping.

Reshape should not be needed frequently. It is there only because there is sometimes a glitch from the GMS side.
If you see it reshape all the time, there is a problem either in the configuration or SEMCCD.dll

I need more history of this. When did this start happening ? After an upgrade in what hardware, what software ?

Please do the following:

1. Show me the instruments.cfg from your K3 PC.

2. Compare that with what it gives out in GMS software. We need to make sure that these are identical.

3. Show me the GMS camera configuration regarding the dimension, rotation, flip, shutter configuration. There may be some clue there.

4. Find out your SEMCCD.dll version and let me know.

Hi Anchi,

You're right, it's because of the width/height swapping.
See attachment for the difference.
I swapped them, now it's working.
Thanks a lot!

Best,
Guang