Creating a Short Film at HTC: Lessons from a Team-Based Animation Project

Navigating Between 2D and 3D Skills

Team-based animation projects can be both exciting and challenging, and my short film project at HTC was no exception. Over the course of a semester our class was engaged in the production of Robot Retribution. Among my peers and me, our skillsets and experience varied greatly with little overlap as some students started out on the 3D oriented classes while others were more familiarized with 2D animation from the get-go; I stood somewhere in the middle with a strong foundation in Blender while also recently completing some of the 2D oriented courses. Consequently we settled on using Blender as our primary tool to accommodate everyone where they were in their studies; this choice however would prove far more complex than any of us anticipated.

Setting the Scene

A bit reluctantly, torn between the role I wanted and the role my team needed, I stepped in as an Environment and Material Designer to fulfill a crucial gap in our pipeline rather than pursuing animation; I was skilled enough with the Material Editor and 3D modeling in general, but this was my first rodeo fleshing out a fully realized environment.

Our storyboard featured two primary locations: a gas station rooftop where the protagonists scout things out, and a multi-level parking complex patrolled by robots. While the animation teams awaited the finalized environments, I ensured progress by providing skeleton sets, simplified placeholder environments later to be replaced with detailed assets.

Built for Convenience

The gas station was intentionally minimal. I began with three core forms: two to block out the store structure, and one for the fuel pump overhang. Despite some of these details never making any screen time, I refined the model by beveling one corner and insetting some faces to represent the storefront, then added flat planes to mark areas for parking. Prioritizing this set allowed the 2D animation team to start production early, while the 3D team advanced on their character rigs. The Finalized version only built upon the original skeleton; adding on details such as the AC units, support beams, and fuel pumps before applying the corresponding materials.

Concrete Ambitions

The parking complex evolved a bit more dramatically. My first draft was overly ambitious, featuring multiple three core shapes to the structure, a toll booth, and a collapsed crane envisioned for an action sequence.

However, as the project's scope tightened, I simplified the design to a more manageable L-shaped layout. By reusing assets from my earlier version, I was able to stay on schedule and support the 3D team's workflow. We had planned to include collapsed floors, and while some of that is visible on the model, it was never fully realized or used in the final cut.

Ruined by Ruins

To connect both main sets, I needed a transitional cityscape, a sprawling ruin to bridge them both into once cohesive world. Building this was easily the most daunting part of the production, requiring numerous assets.

I started with environmental props: trees, tree guards, and powerlines, most modeled using Blender's Skin Modifier for organic and wired forms. The trees being the most complex as I'd vary the width of the trunk and branches then apply a subdivision modifier afterwards giving more of an organic shape. Power poles were modeled a bit more standardly using several simple privatives; connecting them with cables however also made use of the skin modifier.

Achieving the rubble I sculpted several meshes into blobs that spilled onto the streets, then scattered debris using a particle system. The ruined buildings came from a single duplicated asset repeatedly fractured using VFXGuide's Cracker add-on. In hindsight, this approach was far from efficient; with every implementation of this combination of assets our poly-count would rise and severely impact performance, especially on campus hardware.

Refining the Workflow: Reflections and What I'd Do Differently

In retrospect the city ruins likely wasn't the most efficient use of my time. Outside of that one slip I couldn't help but feel destroyed by this project when it was all over. To be fair, we were all still learning, none of us had mastered scene optimization, composition, or workflow management yet, it was even as much of a learning curve for us as it was for our instructor.

If I could redo this project given the insight I have today, I would start with stronger storyboards and shot composition accounting for optimized environments with a priority on key details to save time. Take for example my Binocular Shot, instead of a dolly-in through the scene to the parking complex I'd instead frame our destination between two blocks of skyscrapers, using the perspective lines apparent on the sky scrapers to guide the viewer towards our destination, then cut to the binocular perspective with a 400-600mm lens. Providing a clear focal point while avoiding the countless assets in our initial scene.