Animator David Barlow-Krelina on Getting 3D Particle Effects To Do What You Want Them To

This is a guest post by David Barlow-Krelina, one of 6 animators participating in this year’s Hothouse program, the NFB’s 12-week paid apprenticeship for emerging filmmakers.

It’s been an intense few weeks here at Hothouse. We just reached picture lock this weekend, and in the few weeks prior to this deadline, all of us have been scrambling to create as much material as possible for our editor and sound designer.

The idea of picture lock means that all time-related events in the picture must be locked “to the frame”. My film combines 2D hand-drawn animation with 3D particle effects, so I had to time out all of my character’s actions (footsteps, impacts, voice, etc.) as well as determine the movement of the particles. I really wanted the particles to create a certain kind of sound, so in addition to the determining the particle’s sync points, I also had to figure out its “quality” of motion.

The particles can move in any number of ways, but each would invoke it’s own particular sound texture. Between now and the time that the sounds are added, my challenge has been to tweak the particle’s visual aesthetic to match my intended sound aesthetic (which at this point, is still in my head).

Controlling the movement of the particles has been difficult. It’s one thing to visualize how you want them to move, but actually doing it is much harder, especially when you’re restricted to dialing in numerical values in the 3D software.

I’ve been able to create interesting motion using a combination of techniques: force fields (wind, vortexes, turbulence), emitters (point emitters, volume emitters), friction (the rate at which the particles lose their momentum), motion paths and editing techniques (time-stretching, time-remapping).

It seems like the more I try to force the particles to go where I want them to go, the less natural the particles appear. It’s all about finding a balance between having control, and surrendering control to the software. I did a few tests where I tried pushing the particles along a motion path, but the resulting animation looked too stiff; all of the particles moved at the same rate and on the exact same trajectory.

This technique might have worked if I wanted to create a flowing river on a stationary surface, but not for something like a swarm of bees moving through the air. For natural movement, I achieved the best results using a more minimal approach. I was able to push the particles in one way or another using a single turbulence field and experimenting with different values for the phase and frequency. It’s a lot of trial and error… not to mention the fact that 3D particles seem to enjoy making my computer crash!