The team and I went out and performed reshoots this weekend again. We are trying to get cleaner plates and refine the movement of the camera more. In addition to helping with the filming, this weekend has been dedicated to refining the rocket engine effects for the flying Ducati, and researching/developing the skid mark and smoke of shot two.
Rocket FX
My goal this week with the rocket engine was to smooth out the waviness of the rocket and try to enhance the feeling of force and power in the simulation. I am still working through some issues with integration with the motorcycle’s animation as the flame is becoming more smoke like as it is whipped around by the bike, but the static flame is closer to what I am looking for:
The rocket engine uses a circle as its source and drives the velocity from the normals. The velocity is extremely high to try and make the flame fast moving, and I am using a remap and negative flame expansion value to push the flame into a cone. Viscosity of the pyro smoothed out the flame from my previous tests. A rand expression with a low value is added to the temperature sourcing to create color streaks and variation, but this effect I think can be pushed much further.
For this next week I will be working on adding more variation into the flame so it does not feel as static and integrating it with the animation. Over the weekend, I found more references for the rocket effects, the two main references I am working from are:
Additional Reference:
Skid Mark/Smoke
Using the attribute transfer nodes, I was able to get a skid mark effect pretty quickly in Houdini, though I will be investigating ways to push it further in the next week. To achieve this effect, its a simple grid that passes a custom attribute to the ground plane anywhere the motorcycle touches. A SOP solver was needed to make the attributes persistent from one frame to the next.
For the smoke, I found some tutorials that approached a burn out effect by creating a pop network that defines the velocity fields based on the tires movement. These velocities are then imported into a pyro simulation to create the smoke. I have a version of the effects working, but it needs more refinement and I would like to explore ways to adjust this network to be more controllable.
Select reference: