implicit neural networks have emerged as a crucial technology in 3D surface
reconstruction. To reconstruct continuous surfaces from discrete point clouds,
encoding the input points into regular grid features (plane or volume) has been
commonly employed in existing approaches. However,