Facet csv file: possible bug in dip direction

[Dave86]

Hi,
first of all, lots of thanks to Daniel. Amazing software. Really!
With the version 2.6.2 I am facing a strange issue with the facet data exporting. I am not sure if it is a bug or just me not completely understanding how the command works.
Having a group of facets, with various orientations, the resulting .csv file, generated with the proper command, will display correct dip/dip direction values for those facet with dip direction ranging between 0 and 180° (as shown in the DB tree). Facets with dip direction 180-360°, on the other hand, will display dip direction values "flipped" of 180° (with respect to the values shown in the DB tree).

An example:
a facet with dip/dip direction of 046/29 will be show the same (right) data,
a facet with dip/dip direction of 264/68 will display values of 084/68 in the exported .csv file.

Am I missing something?

Thanks,
Dave

[daniel]

Have you tried with the 2.6.3.beta version? I believe Luca has fixed a bug related to the dip/dip direction computation.

At least on my side I've tried to classify the facets then export a family with dip > 180 and I get a csv file with the right dip values. Or did I miss something?

[Dave86]

Hi Daniel,
sorry for the late reply.

You are right, the dip/dip direction values are correctly exported in the 2.6.3 beta version.

Thanks a lot,
Davide

[daniel]

Good to hear!

And thanks for the feedback.

[Dave86]

Daniel,
I will use this post again because it is related. As you said, the problem occurring in exporting the .csv file of the facets has been solved in the 2.6.3 beta version. On the other hand, I noticed that, when a facet is fitted to a plane, the dip/dip direction of plane and facet not always correspond, and it is possible to see that the two surfaces diverge. If the facet is fitted directly through the points, however, it will be always similar to the plane.

This does not occur using the 2.6.2, with which the orientation of all the facets correspond exactly with the dip/dip direction of the planes that originate them. I cannot recognize a pattern (e.g. if there is a direction for which the divergence is higher), but I saw differences up to 15 degrees in dip direction and up to 10 degrees in dip values.

For now I am using the 2.6.2 for creating the .bin file, and then use the 2.6.3 for exporting the .csv file, but it is not very quick;).

Thanks a lot,
Dave

[daniel]

What do you mean by 'when a facet is fitted to a plane' compared to 'the facet is fitted directly through the points'?

I'm not sure to understand the difference (I'm not used to play with qFacets). Also, if you can provide me with snapshots it would be clearer for me.

[nmenegon]

Hello Daniel,

Using versions 2.6.3, 2.7 and 2.8 I have the same problem with the Fit Plane Tool.
When I use the 2.6.2 version the plane fitted is correct, but when I used the newer versions I achieve a plane perpendicular to that fitted with the version 2.6.2.
In general this error happens when the point cloud is circa pseudo horizontal.

I attach some snapshot as example.

Thank you very much
Bye

Niccolo'

[daniel]

We have changed the plane fitting algorithm a while ago (it is in fact only an Eigen vectors/values determination). This was due to the fact that it was behaving poorly on some clouds...

We opted for the very common "Eigen" library (instead of old code inspired from the old "Numerical Recipes"). However it's possible that other degenerate cases exist for this new version of the algorithm.

Can you send me your cloud example? (cloudcompare [at] danielgm.net). Not sure that I'll find a solution but who knows ;).

[nmenegon]

Hi Daniel!I don't have the same example of the post but i can send you now another example.

[daniel]

Meanwhile I have fall-back to the old plane fitting algorithm. The new one was clearly doing even stranger things.

Can you test again with the latest 2.8 beta version?