|
"Addressing this problem, our shape representation is based on deforming a template mesh."
Please don't take quotes out of context, the full quote is included here if you don't have access to the article
"However, it is difficult to discretize volumetric fields consistently
with such a grid-less point-based discretization.
Addressing this problem, our shape representation is based on
deforming a template mesh. While Umetani [2017] deforms a cube
surface to parameterize a 3D shape, we further extend this approach
by deforming a PolyCube. The PolyCube approach allows us to parameterize a wider variation of shapes"
I don't see how in any way that is relevant to this debate.
I will provide a brief article summary below
The authors state that instead of deforming a cube as a starting template for parameterisation of a 3D model, they could use a polycube, essentially many cubes and deforming those using a machine learning algorithm and Gaussian processes to create a 3D model for CFD analysis.
Please only quote things you understand, otherwise things can get taken out of context or misinformation occurs.
|