Machine Learning In Nuke – Creating Mattes Using Copycat Node


Our CG industry grabbing every new technology to make work faster, Foundry has launched Nuke 13 which supports AI and machine learning. Their many new improvements for performance optimizations and new tools added like CopyCat Node. This tutorial helps you to learn the concept of machine learning and how you may already use it in your day-to-day life.

We already know that we can Rotoscoping using machine learning. learn how this AI and machine learning impact the VFX industry, and how you can introduce it to your compositing workflows in Nuke.

Use CopyCat Node in NukeX to Save a Lot of Production Time for Rotoscoping 

This tutorial helps you how to learn how the CopyCat node works in NukeX. You will also learn how to create an effective data set of images to train the network. Before learning how you can use it to create a garbage matte. Using CopyCat you can train a neural network to replicate various tasks, dramatically speeding up your workflows.

This is the 2nd part of the series about machine learning. How training and monitoring the neural network and training using copycat to creating mattes.

Creating Mattes Using CopyCat in Nuke

Applying and Improving the Results of Rotoscoping the footage using inference node in nuke. The Inference node reads a .cat file data created by the CopyCat node and applies the effect that was learned from the data set to the entire sequence.

Leave a Reply