Issue with the importation of the subset coordinates (not taken into account)

[GuillaumeThomasDL]

Hi,

I want to compute the displacement field around a crack. The subset coordinates are written in a subsets.txt

subsets.txt

such as :

BEGIN SUBSET_COORDINATES 310 717 311 717 312 717 . . . 599 1317 600 1317 601 1317 END SUBSET_COORDINATES

The subsets coordinates are extracted from a Mask as I unfortunately did not manage to import the mask directly into DICe.

I then import the .txt file using the 'import subset locations' icon :

and click on the 'Show subsets' icon to display the valid subsets :

As you can see on this last screenshot the subsets are not displayed. And if I run the DIC analysis, the displacement field is calculated on the whole image and not on the region of interest as I wanted. I tried both versions Version 3.0-beta.4 and Version 3.0-beta.3 and the same issue arose. I then downloaded DICe on a different computer, but still the problem persisted. Both computer used the Windows version of DICe.

It might perhaps have something to do with the fact that my GUI seems a little incomplete or even buggy. As you can see it does not seem to display all the features that I can see on the Youtuble tutorials video examples (mainly the one related to the ROI - no matter what file selection method I am in).

I leave you 3 images in case you want to try the .txt file :

Thank you in advance !

[dicengine]

Sorry you had trouble with this. There was a limit set on the number of custom subsets which was only 500. Now it has been in creased to 2M in this commit: https://github.com/dicengine/dicegui/commit/c1d88b74879fb1d57140d8ae91281290b51733b3

I tried out your example problem and it worked for me with the increase of the limit. I did notice that the subset locations seemed to be rotated by 90 degrees from the orientation of the crack. X and Y might be swapped in your subset locations file.

The commit above will be included in the next release of DICe (v3.0-beta.5) which should be out in the next couple weeks. If you want to use the update sooner, you will need to build from source.

[GuillaumeThomasDL]

Hi !

Thank you for your quick and precise answer ! If it's just a matter of a couple of weeks it should be okay then.

Although, as I explained in my last message I am still not able to directly interact with the image to draw a ROI manually (I am missing the 'Regions of Interest' tab). I am for exemple unable to replicate this example 'https://www.youtube.com/watch?v=TD7HDwPxx1Q&t=7s'

Is it a feature that is just missing on the Windows GUI ?

Thank you again !

[dicengine]

The video above is from an older version of DICe. In v3 and beyond, the icons for drawing an ROI (using a rectangle or polygon) are shown in the image below .

[GuillaumeThomasDL]

Thank you for your answer.

I tried 2 different things:

1/ I tried to draw a common rectangle on the surface using the tool shown in the picture you sent me, unfortunately the rectangle was ignored during the correlation process and the whole image was again considered.

2/ I created 2 new .txt files:

subsets3.txt subsets3_rev.txt

Each of them containing the coordinates of 484 subsets, the only difference being the position of the x and y axis.

Like during my first attempt, loading these .txt files change nothing and clicking on the 'show subset' icon either does nothing or make all the pictures disappear and the GUI display a white image (which persists even after closing and re-opening the software).

If the pictures did not disappear, running the correlation process will give me this result:

I then closed DICe, waited 5 mins re-opened it again and tried the same thing, with the same parameters and this time I obtained the following result:

with the following error message

'Error, not enough features matched for feature matching initializer./nIf the images have low contrast, consider using the feature matching initializer with thresholding by sepcifying a threshold block size > 0.'

Which is strange considering I ran DIC analysis on this set of images with other DIC softwares and obtained reasonable results.

Do you have an idea what might be causing that ?

Thank you

[dicengine]

Can you supply the input.xml file and params.xml file that were generated by the GUI? That will help me figure out what is going on.

[dicengine]

Also, which operating system are you using? There must be something strange going on. I can run your test images fine, the feature matching works as well as setting an ROI or importing the subsets .