Gabriel Neville FX Demoreel
During my studies at ArtFx, I had the chance to work on several technical challenges such as water, smoke simulations, rigid bodies destructions but also procedural modeling, crowd and pipeline. I hope you will find it interesting !
One of the most complex FX of my graduation year at ArtFx was the car crash.
I had to find a way to bend the body of the car while having the correct interactions with the metallic pieces inside the car.
Vellum wasn't realeased back then, so I used rigid bodies simulations with custom constraints and post process deformation.
In order to manage the huge amount of geometry of the environment and create really quick set dressing for our 50 differents shots, I developed a Procedural Stadium that creates the Road, the tribune, the commercials, the lights, all with proper UVs from boxes and a simple curve.
I then created my custom crowd system in VEX code that would populate, loop, offset and randomise arnold instances from the stadium input.
We imagined that in the future, we would drive cars with holograms.
As director of photography on set, I had to anticipate the position of the lights coming from the holograms on the actor.
I then decided to do the holograms fully in Houdini with procedural animation and particle simulations.
Personnal shot i did few years back for practise at ArtFx. I thought it would be funny to imagine a small robot in charge of destroying the cigarette butts.
I used a mini skate and a Canon 700D to shoot the original plate. Then i did the tracking with synthEyes and the animation / lighting in Maya.
Finally, I made the Fx in Houdini and composited the shot in Nuke.
To create this wave that floods Paris' poor neighborhood I worked with several layers of FX such as FLIP, RBDS, Pyro, White water, wet map and particles.
I also had to find a reliable way to upres the flip simulation. I could then wedge different parameters in Low Resolution and getting the same behaviour for the final high resolution simulation.
The main challenge was the huge amount of geometry to fracture as the camera in going up from the bottom to the top of the pagoda.
I decided to fracture only one piece of each type and then re-instance them back inside the simulation. In order to compute faster, I have deactivated the pieces that went outside of the camera’s frustum.
The close shot on the sparks had to be really detailed, that’s why I decided to work on multiple simulations of sparks in order to add variation easily.
Instead of developing a really complex rig, I decided to use a procedural growth algorithm that drives and animate the geometry directly in VEX Code with matrix and point Attributes.