Issue with sharp texture

[stephane...@sintef.no]

Hello,

I have an issue with textures generated by DREAM3D. I have created by own texture which is very sharp (it's basically only the copper component with 10% random - see attached file Copper_random10.txt). So I use this as definition of the ODF in the StatsGenerator. Then I create a synthetic volume, pack the phases, match the crystallography ... as it should be. Then I extract the orientations of each feature and export the texture (Euler angles) obtained. When I compare the input and output texture, there are very different. Is that because my initial texture is too sharp?

Another question, I would like to have a random MDF. How do I do that? Or is it already the default; the distribution in the StatsGenerator is not flat as it should for random, shouldn't it?

Thanks.

Regards

Stéphane

[Michael Jackson]

Stephane,
After looking at the files I have determined that the input texture
data needs to have a weight of 10.0 instead of 1. The weighting tells
the match crystallography filter how much preference to give to those
angles. By setting them to 1.0 it basically tells the filter to give no
preference at all. I replaced the weights with 10.0 and ran the pipeline
(Attached). I also had DREAM.3D generate a Pole Figure after the
synthetic was created and compared that to the input pole figure and
they match very well (Images included).

I also wrote out a ODF file and those weights are set to 1.0 along with
the sigma=1.0 (which is noted in the documentation for the filter).
Reviewing the workflow it might be better to allow the user to enter a
value for Weight/Sigma in that filter so that the user will get the
expected results.

Let me know if this does not fully answer your question.
--
Michael A. Jackson 400 S. Pioneer Blvd
Owner, President Springboro, Ohio 45066
BlueQuartz Software, LLC EMail: mike.j...@bluequartz.net
Voice: 937-790-1601 Web: http://www.bluequartz.net
Fax: 937-746-0783

> --
> 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
> <mailto:dream3d-user...@googlegroups.com>.
> To post to this group, send email to dream3...@googlegroups.com
> <mailto:dream3...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/dream3d-users/377188c8-f85b-4d19-a376-35d897000ab3%40googlegroups.com
> <https://groups.google.com/d/msgid/dream3d-users/377188c8-f85b-4d19-a376-35d897000ab3%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Stephane Dumoulin]

Hello Michael,

Thanks very much for your quick answer. I see that the help has improved very much lately. Thanks for these efforts from the DREAM3D team.

I have a question regarding the meaning of the Weight and Sigma. Do you any reference that I could read?
Is Sigma similar to a scatter width (see attached article - I used that approach to generate my texture file).

I think also that it would be a good idea to give the user the possibility to choose the weight in the Export ODF filter.

Regards
Stéphane

-----Original Message-----
From: Michael Jackson [mailto:mike.j...@bluequartz.net]
Sent: 14. juni 2017 18:43
To: Stephane Dumoulin <Stephane...@sintef.no>
Cc: dream3d-users <dream3...@googlegroups.com>
Subject: Re: Issue with sharp texture

Stephane,
After looking at the files I have determined that the input texture data needs to have a weight of 10.0 instead of 1. The weighting tells the match crystallography filter how much preference to give to those angles. By setting them to 1.0 it basically tells the filter to give no preference at all. I replaced the weights with 10.0 and ran the pipeline (Attached). I also had DREAM.3D generate a Pole Figure after the synthetic was created and compared that to the input pole figure and they match very well (Images included).

I also wrote out a ODF file and those weights are set to 1.0 along with the sigma=1.0 (which is noted in the documentation for the filter).
Reviewing the workflow it might be better to allow the user to enter a value for Weight/Sigma in that filter so that the user will get the expected results.

Let me know if this does not fully answer your question.

--
Michael A. Jackson 400 S. Pioneer Blvd
Owner, President Springboro, Ohio 45066
BlueQuartz Software, LLC EMail: mike.j...@bluequartz.net
Voice: 937-790-1601 Web: http://www.bluequartz.net
Fax: 937-746-0783


stephane...@sintef.no wrote:

[Michael Jackson]

Dear Stephane,
I'm not the original author of the Texture codes so I am going by
memory but I think the basic idea is that the "Angle Space" is broken
into a number of bins based on the Laue Group (CubicOps.getODFSize())
(Only Cubic_High and Hexagonal_High are allowed) and each bin is given
the same chance of selection when asking for an angle. By adjusting the
weight to a larger value that bin will end up having a higher chance of
being selected (or something close to that). Dr. Groeber's thesis may
better explain the algorithm. One can also look in the
Texture::CalculateCubicODFData function located in the
OrientationLib/Texture/Texture.hpp file.

The "Sigma" value will spread the chance over the sigma value of bins.
For example, if you use Sigma = 1 (must be an integer) then the only
chance will be from that particular angle bin itself. If you use Sigma =
2 then the 2 neighboring bins will have a higher chance of being
selected, so in effect using a larger value for sigma will spread out
the "hot spots" in the pole figure. I hope all of this makes some sense.

I have added the features to allow the user to set the Weight and Sigma
values when writing the ODF file. If you are keeping up with the develop
branches on all the repos you should be good to use it or you can try
out a nightly build from the usual spot
(http://dream3d.bluequartz.net/binaries/experimental)


--
Michael A. Jackson
BlueQuartz Software, LLC
[e]: mike.j...@bluequartz.net