search

EMsoft-org / EMsoft

Public EMsoft repository
Other
63 stars 96 forks source link

Zero Index rate #165

Closed Diego8901 closed 2 years ago

Diego8901 commented 2 years ago

Dear Developers,

I am new to EMsoft. I tried to run dictionary indexing on the Ni example data following the tutorial paper (Integrating Materials and Manufacturing Innovation (2019) 8:226–246). I didn't change any of the parameters but surprisingly got a 0% index rate. It looks like the program failed to read the patterns. Can anybody help me to solve this issue?

I am using the the nightly compiled distributions EMsoft-5.0.20221118.-Win64.zip on Windows 10.

The screenshots are attached.

Thanks a lot.

Jie

1 2 3 4 EMsoft-5.0.20221118.-Win64.zip

marcdegraef commented 2 years ago

Can you post the namelist file ?

Thanks, Marc.

Diego8901 commented 2 years ago

Dear Prof. Graef,

Thanks for the quick reply.

Here are the namelist files I used. EMEBSDDI-Ni1_test.txt EMgetADP-Ni1_test.txt EMEBSDDIpreview-Ni1_test.txt EMEBSDmaster-Ni1_test.txt

Jie

marcdegraef commented 2 years ago

Can you post the output image of the EMgetADP program ? Does it look correct?

Diego8901 commented 2 years ago

Here it is. It looks correct to me.

Seems Github do not support .tiff, so I convert it into a .jpg.

NiScan1_test_ADP

marcdegraef commented 2 years ago

can you try and set numsx and numsy to 60 and binning to 1 in the EMEBSDDI namelist file ?

Diego8901 commented 2 years ago

Thanks, Prof. Graef,

I guess it is working!

Here are some screenshots. Are they correct? 5 6 7

Does this mean that I should always input the pixel dimensions of the pattern (instead of that of the camera) and set the binning to 1?

Jie

Diego8901 commented 2 years ago

It does not look good again. The index rate is ~14%

8

marcdegraef commented 2 years ago

yes, you should always enter the actual size of the patterns in the experimental file and set binning to 1.

It is possible that you need to change the sign of the pcx pattern center coordinate... there were some changes in the code after the tutorial paper was published, and I think this is one of them...

Diego8901 commented 2 years ago

I just changed the xpc from 3.4848 to -3.4848 as you suggested. The index rate increased from ~14% to ~15%.

Here is what the IPF looks like. On the top is the .ang output by EMsoft and the other one is the original data.

9

marcdegraef commented 2 years ago

ok, the change in the pattern size and binning parameters also requires that you multiply the detector pixel size delta by 8, and that you divide xpc and ypc by 8 ... I think those parameters were valid for the full size patterns and they must be scaled for smaller patterns.

Diego8901 commented 2 years ago

I didn't realize the connections between these parameters. Now the problem is solved. Thank you very much for your detailed explanation!

Jie

marcdegraef commented 2 years ago

No problem. Glad to help.

I will close this issue then.