Leginon

That does seem low. The mean count you get in leginon should be similar to what you get on GMS under the same imaging condition as you stated. Which node did you use to get this reading ? Also check that you get double the count if you double the exposure time in case it is not being set the way you expect it to do.

What is possible, though, is that my calculation did not account for binning as well as I could as it was original written for bin=1 while we use 2x binning to do K2 counting-mode like acquisition.

Try to acquire as super-res (Leginon K3 bin=1). It should avoid this problem. The camera sensitivity of K3 in super-res (GMS called x0.5 bin) should be 1/4 of reported on GMS as x1 bin.

In addition, K3 has faster frame time than K2 so you can aim for higher eps to get the same DQE. There are many advices from various publications. Some prefer highest DQE, others prefer quick data collection. I think you can find one that you like and fit your scope/detector combination.

I am in the Navigation node and I sent my en preset to do the reading (it is 8eps in GMS at GMS bin1). The mean counts do double when I double the exposure time so that is working.

If I override preset so that my exposure is 1 second:
GMS bin0.5 = 2eps
GMS bin1 = 8eps
Leginon mean bin1 = 0.06
Leginon mean bin2 = 0.24

But my sensitivity is still 0.03 in either case?

My eps is low because I usually image in CDS

I will have to do some test on the real instrument. We have not used CDS regularly at NYSBC, just testing. I do know that switching CDS and non-CDS is not quick, so it is important to stay in one mode in all presets used.

OK thank you! If it helps, I am having the same sensitivity/ratio issue with CDS and without.

That is even more wierd. I can see sensitivity issue if you attempted to do an actual calibration in this case instead of entering the sensitivity (counts/electron) manually because the count is unusually low. Not clear about what you mean by ratio issue. Ratio between Leginon mean and GMS mean ?

Sorry, yes I mean the ratio between leginon counts/GMS electrons. It's very low when I use either CDS or when I use nonCDS mode.
If I can test anything to give you more info, please let me know.
Thank you!

I am puzzled by what you get so I'd better confirm is. I won't have scope access until Sunday, so please relex for the moment.

I checked GMS against Leginon. It looks normal. However, I want to make sure that you and I are talking about the same thing.

Everything I measured and talked about is not about camera sensitivity, but the count mean value you get in Leginon gui vs the mean value you get in GMS through Analysis->Mean & Stddev.

When you make this measurement, Navigation and or any Calibration node you use should have the setting of overwrite-preset=off. This way PresetsManager controls the camera behavior.

When a preset is sent to the scope, it means the spot size and beam intensity is fixed, as well as camera bin and exposure. Here is how I did my test. The example is non-CDS, but it would be the same
with CDS activated.

1. I go to PresetsManager, select en preset (2 s exposure, bin=2, CDS unchecked), sent that to scope.
2. Navigation->Acquire.
3. The stats box shows mean count about 30.
4. GMS->Camera Settings: Unprocessed, CDS unchecked
5. GMS->Camera bin=1, Acquire 1.99 s exposure, 40 frames.
6. At the status bar below, it shows 20 e/p/s for our imaging condition
7. GMS->Analysis->Statistics->Mean & Standard Deviation mean about 30 as well.

I did notice that step 4 is eccential for matching numbers. Gain-Normalized mode on GMS does not retain the original mean, it seems, but instead scale it to 1000.

For camera sensitivity calibration, I would first scale my mean count to 1 second : 30 /2 = 15 counts/s and then take ratio of that to GMS dose rate = 15/20 = 0.75 counts/e and that is what I would enter at the last line in the Dose node Calibrate Tool, and then click Save.

GMS version 3.32.2403.0

One concern I have is that if your mean count is in 0.24 range, the image would be very noisy. If you have feature in your image, are they clearly visible ?

Also, please check your dmsem.cfg regarding k3->DM_Processing. It should be set to "dark subtracted". Is it possible that it got set to "gain-normalized" there ? Leginon takes in dark-subtracted images and movie so that the movie can be compressed. The gain reference is generated and saved in Leginon separately.

Yes great! Thank you so much, Anchi. The image looked very dark before. Now the sensitivity is 0.969 just like it should be. I had the "gain-normalized" selected in the dmsem.cfg file because I thought "dark subtracted" was inaccessible and wouldn't work unless GMS was in service mode.

Good. For your information, Leginon access this mode through back-end.