Supported file types: 3D data

The platform supports the following 3D file types:

• .stl RECOMMENDED
• .vtk/.vtp/.vtu (The Visualization Toolkit)
• .h5/.hdf (Hierarchical Data Format) BETA
• .ply (Polygon File Format) BETA
• .op2 (Nastran) BETA
• .step/.stp/.iges/.igs/.brep (NURBS data) BETA

Some common error messages that indicate meshes/geometries are not supported are:

IDs of 3D files and tabular data should match

IDs of 3D objects are generally the names of the 3D files. Make sure you match the names of 3D data files to identifiers within tabular data (of boundary conditions, simulation results, etc.) which they are associated to (see figure below).

Handling of NURBS data

Monolith’s 3D AI algorithms work using polygon meshes or point clouds. If you upload NURBS CAD data, it will be converted to polygon meshes as it is imported. A simple automesher is used for this. To have full control of mesh quality is is recommended to generate the mesh outside of Monolith with dedicated tools.

The recommended CAD data format is high-resolution STL data.

3D data as tabular input

If your existing tool chain does not provide any possibility to export 3D in one of the recommended/supported formats 3D data can be brought as tabular data and converted to point clouds in the Monolith platform.

The table should have the following structure with one row for each vertex (mesh point):

• Use Tabular to import the data into a notebook
• Use Table to 3D to convert the tabular dataset into a Point Cloud.

Point clouds can be used with Surface Field models but not with Autoencoders or 3D-to-Scalar models.

Reference data structure

Download an example of a recommended 3D data structure here.