Crowds in Houdini

Taught by Charles Chorein

Course Number:
Software Version:
Original Run Date:
June 2024 
2 hours 9 minutes 
This course, taught by VFX supervisor Charles Chorein, provides a comprehensive guide to creating and animating a crowd scene using motion capture data and various assets from our VFX308 course. Starting with the acquisition of models and motion capture from the Mixamo database, you will create and enhance assets in Maya, then import and set up these assets in Houdini. The course covers the application of motion capture data, adding props like swords and shields, and managing all elements within a single agent.

You will learn to incorporate the terrain, perform rigging adjustments, and populate your scene with varied agents. The course also delves into simulating agent collisions and interactions using DOPnet and introduces techniques to add variation in animations based on different triggers, ensuring a dynamic and realistic crowd scene.

Charles Chorein is a CG supervisor in London with 15 years of experience. He has been working in London since 2009 and has achieved various movies such as The Hobbit, Prometheus, The Hunger Games 2, Superman: Man of Steel, Harry Potter 7, Pirates of the Caribbean 4 and DespicableMe. He has more recently been working on TV shows and recently finished Dark Crystal for Netflix.
Crowds in Houdini
Watch our overview of the course

Class Listing

Class 1: Initial setup

The Mixamo database assists us in obtaining a model and motion capture library as a starting point. We create our assets in Maya and then use Houdini to initiate our crowd setup using the Agent node to import our assets and apply the motion capture. Finally, we enhance our model in Maya by adding shields and swords before transferring the asset to Houdini.

Class 2: Motion capture and run cycle

We apply our initial motion capture data and introduce additional variations in the run cycle. Next, we incorporate the swords and shields into our running agent. Finally, we explore how to manage all assets within a single agent.

Class 3

Now it's time to to incorporate the ground from the VFX308 layout to assess our agent under realistic conditions. Additionally, we perform rigging and make terrain adjustments tailored to our specific model.

Class 4: Populating the scene

Next, we populate our asset with hundreds of agents, introducing variation and randomness in animation, props, agent size, and speed.

Class 5: Collision and interaction

We use a DOPnet to simulate and generate collisions among agents, allowing interaction with elements like diverse terrain. The setup necessitates a few adjustments, so we need to evaluate our previous configuration.

Class 6: Adding more variation

Incorporating various "dying" motion capture data to introduce variation based on factors such as Time, VEXpression, or Particle Speed. We adjust properties to enable the triggering of different motion capture animations based on specific action triggers.