Questions on scan rotation

74 views
Skip to first unread message

Larry Aagesen

unread,
Jul 10, 2019, 4:42:28 PM7/10/19
to dream3d-users
Hello,

We are working on using Dream3D to import an EBSD scan of rolled pure copper into the MOOSE phase-field application to simulate recrystallization of the deformed grains. We have a few questions related to the process. We have created a pipeline based on the prebuilt example "INL Export". Right now we are not doing serial sectioning so we are just looking at a single layer EBSD scan. We had a couple of questions related to the sample and Euler reference frame rotations. The EBSD system is an EDAX system and we are importing the .ang files that it generates, so we are using the rotations as suggested in the documentation for Import Orientation file to H5EBSD :

Rotate Sample Reference Frame filter: 180 @ <010>
Rotate Euler Reference Frame filter: 90 @ <001>

However we noticed there is a difference between the original image file produced by the EBSD system and the image written out by the ImageWriter filter at the end of the pipeline. The original .bmp file output from the EBSD system is attached as EDAX output.bmp (prior to cleanup so it's relatively noisy). The results after cleanup and running the pipeline are attached as Dream3D output.jpg. As you can see the images are mirrored relative to each other. Is this expected/desired? Is this related to the issue of putting the (x,y) = (0,0) position at the lower left corner for Dream3D's coordinate system?

Also, I was wondering if you can recommend a review paper/textbook/manual that would give some background information about the rotations? I'm struggling a bit to understand the difference between the sample reference rotation and Euler rotation.

Thanks,
Larry Aagesen, Ph.D.
Computational Scientist
Fuel Modeling and Simulation Department
Idaho National Laboratory
EDAX output.bmp
Dream3D output.jpg

Michael Jackson

unread,
Jul 11, 2019, 1:15:07 PM7/11/19
to Larry Aagesen, dream3d-users

Generally the sample and Euler reference frame corrections are correct but if the EBSD instrument is setup differently than those we have encountered in the past then the rotations can be incorrect as performed by DREAM.3D. There is also the issue of the different output image formats. I noticed that JPG was used, can you try TIFF and PNG to see if they give any different results. If those results are all the same then my best guess is that the EBSD instrument is setup slightly differently and the initial sample reference frame filter that is used may not be needed. You can temporarily disable that filter by hovering over the filter with the mouse and 2 icons will show up. One to disable the filter. One to Delete the filter. Disable the filter and run the pipeline and see if the output matches the EDAX output.

 

https://www.material.ntnu.no/ebsd/EBSD/

 

Quickly: The sample reference frame relates to the morphological orientation of the sample. The Euler reference frame relates to the crystallographic orientation of the grains within the sample. In the link above there are enough slides and papers in there where a deeper explanation should be available.

 

--

Michael Jackson | Owner, President

      BlueQuartz Software

[e] mike.j...@bluequartz.net

[w] www.bluequartz.net

--
You received this message because you are subscribed to the Google Groups "dream3d-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dream3d-user...@googlegroups.com.
To post to this group, send email to dream3...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dream3d-users/6cab7faf-1886-4a7d-aea2-e30998f24ce0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Goutham Sreekar A. R.

unread,
Jul 12, 2019, 12:33:53 PM7/12/19
to dream3d-users
Hello Michael,

If the sample reference frame rotation is disabled, does it automatically mean that we have to disable Euler reference frame rotation filter as well? 

Thanks

Michael Jackson

unread,
Jul 15, 2019, 1:12:59 PM7/15/19
to Goutham Sreekar A. R., dream3d-users

For consistency you probably should disable the Euler rotation until you understand exactly what it is that you want to do or what transformations need to be applied to your data to bring the sample and Euler frames into the same reference frame.

 

 

 

--

Mike Jackson

 

From: <dream3...@googlegroups.com> on behalf of "Goutham Sreekar A. R." <sreekar.g...@gmail.com>


Date: Friday, July 12, 2019 at 12:33 PM
To: dream3d-users <dream3...@googlegroups.com>

--

You received this message because you are subscribed to the Google Groups "dream3d-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dream3d-user...@googlegroups.com.
To post to this group, send email to dream3...@googlegroups.com.

Goutham Sreekar A. R.

unread,
Jul 17, 2019, 3:59:54 PM7/17/19
to dream3d-users
Hi Michael,

Sorry to disturb you again. 

When the Euler frame is rotated by 90 degrees about <001>, is it being rotated in the clockwise or anti-clockwise?

Thank you

Michael Jackson

unread,
Jul 19, 2019, 8:30:52 AM7/19/19
to Goutham Sreekar A. R., dream3d-users, Michael Groeber

I believe it is right-hand rule so counter-clockwise.

 

--

Mike Jackson

--

You received this message because you are subscribed to the Google Groups "dream3d-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dream3d-user...@googlegroups.com.
To post to this group, send email to dream3...@googlegroups.com.

Goutham Sreekar A. R.

unread,
Jul 19, 2019, 11:18:55 AM7/19/19
to dream3d-users
Hi Michael,

If it is counter-clockwise, I have one more question to ask.

Attached image shows the standard settings of reference frames in EBSD. So, by following the rotations, mentioned in the filters, we would definitely bring the sample reference frame coincide with the Dread 3d's reference frame (which is bottom-left), but we would not be able to align the Euler frame with the sample frame by performing 90 degree counter-clockwise rotation. A 90 degree clockwise rotation would be required to do the job. Kindly, correct me if I am wrong.

L17_Dasher_BASSEM_EBSD-analysis-of-EBSD-data-L17-26Oct09-contains animations.png

Thank you for your time.

On Wednesday, 10 July 2019 14:42:28 UTC-6, Larry Aagesen wrote:

Michael Jackson

unread,
Jul 19, 2019, 5:25:32 PM7/19/19
to Goutham Sreekar A. R., dream3d-users

After consulting with Mike Groeber on this I may have had it incorrect. We do a clockwise rotation but use -90 degrees which is the same as +270 which would allow the crystal and sample reference frames to line up properly. This also assumes that the EBSD instrument is setup with default settings.

 

--

Mike Jackson

 

From: <dream3...@googlegroups.com> on behalf of "Goutham Sreekar A. R." <sreekar.g...@gmail.com>
Date: Friday, July 19, 2019 at 11:18 AM
To: dream3d-users <dream3...@googlegroups.com>
Subject: Re: Questions on scan rotation

 

Hi Michael,

 

If it is counter-clockwise, I have one more question to ask.

 

Attached image shows the standard settings of reference frames in EBSD. So, by following the rotations, mentioned in the filters, we would definitely bring the sample reference frame coincide with the Dread 3d's reference frame (which is bottom-left), but we would not be able to align the Euler frame with the sample frame by performing 90 degree counter-clockwise rotation. A 90 degree clockwise rotation would be required to do the job. Kindly, correct me if I am wrong.

 

Thank you for your time.


On Wednesday, 10 July 2019 14:42:28 UTC-6, Larry Aagesen wrote:

Hello,

 

We are working on using Dream3D to import an EBSD scan of rolled pure copper into the MOOSE phase-field application to simulate recrystallization of the deformed grains. We have a few questions related to the process. We have created a pipeline based on the prebuilt example "INL Export". Right now we are not doing serial sectioning so we are just looking at a single layer EBSD scan. We had a couple of questions related to the sample and Euler reference frame rotations. The EBSD system is an EDAX system and we are importing the .ang files that it generates, so we are using the rotations as suggested in the documentation for Import Orientation file to H5EBSD :

 

Rotate Sample Reference Frame filter: 180 @ <010>

Rotate Euler Reference Frame filter: 90 @ <001>



However we noticed there is a difference between the original image file produced by the EBSD system and the image written out by the ImageWriter filter at the end of the pipeline. The original .bmp file output from the EBSD system is attached as EDAX output.bmp (prior to cleanup so it's relatively noisy). The results after cleanup and running the pipeline are attached as Dream3D output.jpg. As you can see the images are mirrored relative to each other. Is this expected/desired? Is this related to the issue of putting the (x,y) = (0,0) position at the lower left corner for Dream3D's coordinate system?



Also, I was wondering if you can recommend a review paper/textbook/manual that would give some background information about the rotations? I'm struggling a bit to understand the difference between the sample reference rotation and Euler rotation.



Thanks,

Larry Aagesen, Ph.D.

Computational Scientist

Fuel Modeling and Simulation Department

Idaho National Laboratory

--

You received this message because you are subscribed to the Google Groups "dream3d-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dream3d-user...@googlegroups.com.

Goutham Sreekar A. R.

unread,
Jul 19, 2019, 5:31:54 PM7/19/19
to dream3d-users
Hi Michael,

But, if the Euler is rotated clockwise, then the default values given for both these filters (Euler and Sample rotation filters) will align the frames (Euler and sample). I mean, using +90 in Euler rotation filter would align the frames, not -90.

Also, I sincerely request you to please update the documentation accordingly.

Thank you for your time

On Wednesday, 10 July 2019 14:42:28 UTC-6, Larry Aagesen wrote:
Reply all
Reply to author
Forward
0 new messages