Two years after the last project with Rafael Nadal for the Vivo brand, we were asked again to fill the stadium with cgi crowd.
In 2016, we used Massive in it´s first incarnation of Arnold compatibility. Was 1 month working nonstop and until late, each day.
Two years later plus houdini with crowd system, I decided to go 100% with houdini in this project and our boss and co-workers got onboard.
In just a few days our crowd was working and ready to render; in the end, our main problem was improving the poorly exported stadium geometry from 3dmax.
The two adds for the campain and a short breakdown.
Screenshot from the movie with the cgi crowd and stadium.
Screenshot from the movie with the cgi crowd and stadium.
Screenshot from the movie with the cgi crowd and stadium.
Screenshot from the movie with the cgi crowd and stadium.
This kid is so fanny and cute!
The stadium alembic exported from max, with thousands of objects and uv problems.
The improved version of the stadium, with instanced and delayed load geometries and fewer objects.
Object (layer) variation tool developed for this project.
Since our other crowd guy, Thiago Faravelli, stopped working on the advertising projects to focus on the entertainment business, I begun to develop a complete crowd pipeline inside houdini, without the need to rely on other artists (or sofwares) to handle crowd animation and rendering.
It´s in the beta stage, without error handling, but has forward and inverse kinematics translation. Python transform nodes are used to compute and translate the animation.
Probably I will transfer the system to chops transforms, since it seems cleaner and easier to handle constraints (wasn´t available when I started developing the tool), but in my early tests seems equal or slower to compute than the python transform nodes.
A short video showing the mocap retarget system in it´s first incarnation.
The tree with  the retarget nodes.
Back to Top