[3dem] Gatan camera images with odd sizes

[Martin Schorb]

Dear David,

> 
> A user and I just realized that many, many of the images recorded by our Gatan OneView camera do not have the proper dimensions.  The images examined should have been 4096 x 4096 pixels (full sensor, not binned) but are not.  Some of the images are the correct size, but some are 4087 x 4093, some are 4095 x 4091, some are 4086 x 4089, etc.  The strange file sizes seem to be random, including the fact that images taken within minutes of each other can have totally different sizes, no sequential images seem to have the same odd size and some of the images in the sequence might be 4096 x 4096.
> 
> The only thing that occurs to me that might cause this is that the images were acquired using the built-in drift correction.  It's unclear what that does exactly, but since I don't know, it's definitely a candidate for doing odd things.
> 

You are on the right track. For the drift correction, the "image" that GMS produces is a sum of the movie frames recorded during "exposure time". When using drift correct, the single frames are aligned. Imagine you acquire 10 frames (0.4s exposure). From frame 5 there is a shift by 3 pixels to the right. Now there are two options for image output.
1) you use the entire native camera area which means that there will be a 3 pixel wide strip  that receives data only from 5 frames and hence appears darker than the rest.
2)  the software automatically crops the image to the size that only contains data from all frames. This is what yo observe.

I think there is an option selector in GMS where you can set the intended behaviour.

I only use SerialEM to acquire, so I cannot comment on the following points. What we do there to avoid strips at the image edges is to only use a subset of the image (4064x4064 or  sometimes less, depending on specimen drift). These are then all the same size.

Best,
Martin

> Also, I occasionally find that the pixel size included in the dm3/dm4 header is incorrect.  In other words, during a data collection session recording images at the same magnification, I will see a single image with what Gatan says are 1.11836 nm pixels while the other dozen-plus image have 1.11808 nm pixels.  All these decimal places are most likely bogus, but I totally do not understand how or why they would change and then change back within minutes.
> 
> Finally, I never save images in MRC format directly from the Gatan software, but my user does.  We noticed that the pixel size in the MRC headers is odd.  It is often 1.000 (which is what I thought the default was) but can also be 1.9xy, where again, the xy values change from one minute to the next.
> 
> To say the least, this is all very strange...
> 
> Any thoughts would be appreciated.
> 
> -- 
>      politics is more difficult than physics.
>                                               A. Einstein
> 
>              David Gene Morgan
>          Electron Microscopy Center
>               047E Simon Hall
>               IU Bloomington
>            812 856 1457 (office)
>            812 856 3221 (3200)
> https://urldefense.com/v3/__http://iubemcenter.indiana.edu__;!!Mih3wA!XG2Hz1kfdXK4OV-DMzIFIV2vw8NlDpCjL8qD0sfB_yenmJquj-_bp3rvKFkSlxP4uQ$  <https://urldefense.com/v3/__http://iubemcenter.indiana.edu__;!!Mih3wA!VkQVx2b4HYSjWuBWEKnH444173FFqLouf411VPnMYKkAsSsyFK4aDsQbx_o2Pa-m8w$>
> 
> _______________________________________________
> 3dem mailing list
> 3dem at ncmir.ucsd.edu
> https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem
>