MACHINE LEARNING FOR PYRO FX

TOTAL DURATION: 2h 08m 19s

Learn how to build an efficient smoke and fire workflow in Houdini using machine learning, then bring the results into Unreal Engine for real-time playback. This series covers the complete process, from understanding Pyro fundamentals and creating low-resolution simulations to training custom upscaling models that add high-resolution detail at a fraction of the simulation cost.

You’ll also explore how to prepare loopable volumetric effects, compress them with Zibra VDB, import them into Unreal Engine, refine their look in real time, and evaluate performance, storage, and memory usage. This workflow is valuable for artists who want to create detailed volumetric effects while improving iteration speed and making high-quality smoke and fire more practical for real-time production.