User Forum of Software BASEMENT

basemind

User

Registered: 2015-10-28

Posts: 15

BASEmesh shp-conversion fail

Hi,

I finally got triangle to run without problems (they were mostly related to me not defining the breaklines nice enough), I now got problems getting those triangle results to be converted into shapefiles, however.
So triangle finishes its work after 2 seconds; it creates the following:
~160k vertices
~320k triangles
~490k mesh edges

The basemesh-GUI then freezes and shows no more progress. The qgis-bin process still shows some activity though (RAM-eating and CPU-heating).

I am using Windows 7 64bit with an Intel i7 920 processor and 12GB of RAM. I also tried running QGIS as administrator, to no avail.
The conversion to shp stops latest at ~5%.

Greetings,

base mind

Last edited by basemind (2015-11-04 12:43:38)

maarnold

Advanced_User

Registered: 2014-09-13

Posts: 28

Re: BASEmesh shp-conversion fail

Hi,

the conversion from triangle to shape file can take some time and the gui freezes. However, the conversion is still working! You've got to be patient, especially in the interpolation of elevation data.
Make sure that you have activated the snapping function in QGIS. Vertices which are not snapped properly lead to a lot of very small elements. This leads to a very slow conversion to shape file and ultimatly to a very slow simulation. Also, breaklines which are very close to each other can lead to very small elements.

The elevation interpolation from vector data (elevation mesh) is very slow. See: http://people.ee.ethz.ch/~basement/foru … php?id=314
The interpolation base on a DEM is much faster.

Hope this helps,

mat

basemind

User

Registered: 2015-10-28

Posts: 15

Re: BASEmesh shp-conversion fail

Hi,

I did not yet run the elevation interpolation, since I do not yet have my quality mesh.
I am still curious why the conversion to shape takes such a long time? The elements and nodes are generated by triangle within seconds without any errors.

maarnold

Advanced_User

Registered: 2014-09-13

Posts: 28

Re: BASEmesh shp-conversion fail

Hi,

320k elements is quite a big mesh. I checked one of my meshes with 250k elements yesterday. It took quite some time and also the gui freezes. But a freezing gui is no indication for a failing conversion. You just have to be patient.

basemind

User

Registered: 2015-10-28

Posts: 15

Re: BASEmesh shp-conversion fail

EDIT: You were right from the beginning, I should have been patient instead of aborting the process! It finally converted to shapes after ~3h (at least that was the timespan I didn't dare look at my pc). I exported it to 2dm - I already learned to be patient, so I waited for that for ~3,5h to finish.

The function write2dm in meshConversion.py tells me it is just writing all elements and nodes to the 2dm file, I don't get why this is taking so incredibly long.

Last edited by basemind (2015-11-11 11:58:01)

maarnold

Advanced_User

Registered: 2014-09-13

Posts: 28

Re: BASEmesh shp-conversion fail

Hi,

Glad to hear it worked out.
I ran a test:
1. Imported my ~250k elements mesh with BASEMesh into Qgis. The Import took about 1h (2dm to shape)
2. Exported the mesh to 2dm format with BASEMesh as well. This conversion took about 1h 15 minutes (shape to 2dm)

Did you interpolate elevation data to your quality mesh? Did you use a raster file or the elevation mesh?
A faster conversion tool would be very interesting!

Sam

Developer

From: VAW - ETH Zurich

Registered: 2014-09-04

Posts: 46

Website

Re: BASEmesh shp-conversion fail

dear BASEmesh users,

thx a lot for revealing the application limits of BASEmesh. As maarnold mentioned, a mesh consisting of more than 100k elements is already quite a big mesh. The developing team knows about the mentioned performance issues when dealing with large meshes. I'd like to clarify:

The main goal of BASEmesh is to have a meshing tool that provides (1) graphical user interface (using QGIS) and (2) maximum flexibility for the user (importing, generation, interpolation, manual changes, reviewing, exporting etc.). BUT, there is no free lunch! We have to pay some performance due to these features and stability of the tool. I agree, Improvement is needed and there are ways to make things better.

Nevertheless, for large meshes you are always better of with simple script-based tools that solve one specific problem, e.g. reading triangle result files and directly write a .2dm file! BASEmesh will always be slower than this solution.

cheers, sam