Table Of Contents

Core features

Python interface

Inverse rendering

Plugin references

  1. Docs
  2. Introduction

Introduction

We have showed how to perform forward rendering using PSDR-CUDA in the previous section. This section focuses on the use of PSDR-CUDA’s Python bindings for differentiable and inverse rendering applications.

Automatic-differentiation types

A key ingredient to differentiable rendering is automatic differentiation. PSDR-CUDA relies on Enoki’s to perform transparent reverse-mode automatic differentiation. To this end, commonly used data types include Float32, Vector3f, and Matrix4f from enoki.cuda_autodiff, which can be imported as follows:

from enoki.cuda_autodiff import Float32 as FloatD, Vector3f as Vector3fD, Matrix4f as Matrix4fD

For more details on how Enoki’s automatic differentiation works, please refer to its official documentation.

Rendering a scene
Differentiable rendering