Improve the resolution of the export mesh

[DaichenWang]

Hi! Thanks for sharing this amazing project. I wonder if there is a way to improve the quality or the resolution of the final mesh?

[factoryofthesun]

Hi! There's a number of ways you can go about increasing the resolution of the output mesh.

1. You can run the add_vertices() function to subdivide every triangle by a random barycentric point, then update the mesh object in run_branched() using the output vertices, faces, and normals.
2. The trained NeuralStyleField() is fairly resolution-robust, so after training the NSF on a given resolution, you should be able to take the coarse mesh and upsample it using another program (e.g. Blender or Meshlab), then pass it back through the trained NeuralStyleField() network and get something reasonable.
3. The most naive solution would be to simply train the NeuralStyleField() on a higher resolution version of your input mesh (for instance using any of the programs listed above).