A Simple Image Synthesis Pipeline for Machine Learning Model Training

This project demonstrates a simple image synthesis pipeline based on Mitsuba3 and visualization tooling using Rerun. The pipeline generates synthetic images for machine learning model training with comprehensive visualization capabilities.

Overview

The pipeline combines the power of Mitsuba3 for physically-based rendering with Rerun's advanced visualization tools to create high-quality synthetic datasets for machine learning applications.

Key Features

  • Physically-based rendering using Mitsuba3
  • Interactive 3D scene exploration with Rerun

For more details about the implementation, visit the GitHub repository.