Although the core functionalities of PSDR-CUDA are written in C++17 and CUDA, to enable easy integration into Python-based inference and learning pipelines, running PSDR-CUDA requires using its Python bindings.
This section focuses on the use of the bindings for forward rendering. We will show how to perform differentiable and inverse rendering in the next section.
To import PSDR-CUDA into Python, you only need to import its central module named psdr_cuda
:
import psdr_cuda
PSDR-CUDA relies heavily on the Enoki library for elementary arithmetic types and mathematical operations. Commonly used data types include Float32
, Vector3f
, and Matrix4f
from both enoki.cuda
, which can be imported as follows:
from enoki.cuda import Float32 as FloatC, Vector3f as Vector3fC, Matrix4f as Matrix4fC
For more details on Enoki’s data types and their Python bindings, please refer to its official documentation.