Differentiable Rendering

Differentiable rendering can be used to optimize the underlying 3D properties, like geometry and lighting, by backpropagating gradients from the loss in the image space. We provide an end-to-end tutorial for using the kaolin.render.mesh API in a Jupyter notebook:

examples/tutorial/dibr_tutorial.ipynb

In addition to the rendering API, the tutorial uses Omniverse Kaolin App Data Generator to create training data, kaolin.visualize.Timelapse to write checkpoints, and Omniverse Kaolin App Training Visualizer to visualize them.