PI: Seth Holladay

We received a MEG grant for researching the simulation of granular materials for computer graphics and animation. Materials such as sand and snow simulation is very expensive but necessary for productions, as well as hard to control the look, so I have been mentoring undergraduate and graduate students researching how to make sand simulations both efficient and art directable.

The grant money was used to hire these research students

We really appreciate the grant money to make this possible!

Our method of optimizing the simulation of granular materials is to combine 3 existing granular simulation methods – particles (slow simulation time but most accurate results), fluids (faster but less accurate), and static surfaces (fast but least accurate). In a single granular material, our system subdivides the material and automatically determines which areas need which amount of accuracy (e.g. a splash needs more detail/accuracy than the flat part of the surface nearby) and which simulation method to use in which section of the material. It automatically figures out transitions between these states frame by frame as the material moves and changes, using efficient calculations where possible and detailed methods where necessary.

The project is still moving forward as proposed. Students have been able to combine the methods and simulate sand using are method. They are currently helping debug the process and address encouraging feedback that we have received from submitting to Siggraph, the top tier computer graphics conference, and other submissions.

Here is a list of students are working or have worked on the project

• Kevin Munns (2015) – surface particle motion (particle state
• Wesley Hauwiller (2015) – update the fluid solve method (fluid state)
• Jeremy Oborn (2015-present) – helped with the fluid solve portion of the granular simulation and artistic control of fluids
• Samuel Giraud-Carrier (2015-present) – working on transitioning states (determining when, where, and how each state should transition), and figuring out the surface (static state)
• Nathan Zabriskie (2016-present) – working on transitioning states

The major results so far is that this can reduce not only simulation time but also manual labor on the setup of the simulation, since it automatically determines which simulation method to use where within the granular material. Current optimization methods in industry requires and artist to manually set up multiple simulations and control the interaction between the two. Each setup has to be customary. Ours is showing promise in minimizing simulation setup time.
The final publication has yet to be completed and submitted, but we have a built a sand tool that shows promise.

Please let me know if you need additional information.

Thanks so much for supporting our research area that has a high impact on graphics and film industries.