From Sandbox to Studio: Leveraging R&D Readiness for Production Efficiency
Edward Chiu from the motion design studio ManvsMachine joined us to talk about his work, explaining how the studio integrates technology and art, how his role is to take the research into new visual styles and techniques, and how his passion for 2D style has helped him with other creations.
Introduction
I'm Edward Chiu, the Research and Design Lead at the motion design studio ManvsMachine. We create visuals by deeply integrating technology and art. In a sense, we produce commercials for brands, but instead of explaining a product literally, we communicate ideas through shapes, text, and physics. This approach requires us to use simulations quite often, letting math and physics do a lot of the heavy lifting.
R&D Lead
My role involves leading the research into new visual styles and techniques. A clear example is our exploration of the MPM solver in Houdini. We work on many ice cream projects for brands like Talenti, Walls, and Viennetta. These products usually have multiple ingredients, such as a chocolate shell over vanilla gelato or caramel sauce mixed with truffles. Because these materials have different physical properties, getting them to interact realistically is difficult.
In the past, managing the interaction between FLIP and RBD solvers was always a struggle. When the MPM solver was released, it made sense to focus our R&D efforts there because it is physically accurate. While we still spend time fine-tuning material settings and directing particle movement, we now have a library of setups we can use to speed up future production.
2D Style
Those are my passion projects. I have always liked traditional art, and I'm thankful for tools like the Copernicus module that allow us to bring a different feel to 2D styles. I enjoy exploring the possibilities within our software. There are many ways to achieve the same result, but everyone's process is a bit different.
Passion Project
On a recent project for Carte d'Or Chunkies, I used Copernicus to generate the textures for the cookie dough, chocolate bits, and vanilla seeds. Without the deep dives I did on my personal projects, I probably wouldn't have been comfortable using it so quickly in production. I was able to create the Diffuse, Roughness, Normal, Displacement, and custom SSS passes directly. By using noise, I captured details like the air bubbles in the dough.
One advantage of using Copernicus over a Redshift shader network is the real-time preview, and it also avoids messy "noodle" networks at the shader level. For the vanilla seeds, I imported points and stamped spheres onto them, then used a noise attribute to drive the blur. This created the look of the seeds being embedded inside the ice cream layers. Sometimes it makes more sense to create food assets entirely in 3D rather than using photogrammetry because it gives you more control and faster iterations once the setup is done.
