As part of the Master of Interactive Technology (MIT) in Digital Game Development program at SMU Guildhall, I completed a thesis project focused on VFX. I built an environment centered on an interpretation of the Casa Batlló from Barcelona in Unreal Engine 4, which I subjected to a volcanic eruption and destroyed with explosions and fire (sorry Gaudí). Through the thesis, I demonstrated mastery in three areas: EXPLOSIONS, DYNAMIC WEATHER, AND DYNAMIC MATERIALS.
The Facade of the Casa Batlló consists of curving sandstone masonry and balconies evocative of organic forms like bones and faces topped with a roof covered in colorful tiles akin to the scales of a dragon. Much of it is covered in a mosaic of varied tiles referred to as trencadís, characteristic of many of Gaudí's buildings.
I broke the facade into modular pieces, then used a comibination of a color look up table and masking to create the appearance of sandstone blocks of varying colors.
To create materials for the rest of the environment, I used a combination of Adobe Substance Painter and Designer. I used mask textures to create the many details of the trencadís. For some tiling materials, I created alternate texture sets that I would swap between to keep them from looking repetitive.
Through my research into explosions, I learned that they have three parts: a kick, a high point, and a fadeout. The kick represents the initial detonation and shockwave. During the high point, the fireball and smoke evolves and grows. Finally, during the fadeout, the fire and smoke dissipate, leaving only the remnants of the destruction behind: broken glass, streaks of soot, and shattered windows.
A good explosion has carnage. I used 3DS Max to build custom destructible meshes and destroyed versions of different areas on the façade. I used blueprint to swap the pristine meshes out and replace them with the destruction, using the explosion VFX to cover the transition.
I used a combination of material transitions, shader tricks, and particle systems to create a transition from a serene, partly cloudy day to the cataclysmic scene of a volcanic eruption.
Using dynamic material parameters changed in real-time using blueprint, I changed the color of the clouds and their density. I faked orange underlighting on the clouds from the volcano and fires in the city. For the falling ash closer to the camera, I used GPU particles, but further away, I switched to using a mesh with a material that used a mask for multiple panning layers to simulate depth.
While searching for reference for my volcanic weather, I found many videos showing volcanic lightning arcing through ash clouds. I liked how this conveyed the power of an eruption, so I used a particle system to add lightning sprites to my scene. I used a reveal mask to make the tendrils of the lightning bolt grow from starting point and included the flash lighting up the clouds as well.
I used a material parameter collection float controlled by blueprint to make ash build up on top of everything after the volcanic plume spread over the sky.
I approached dynamism in a variety of ways, using materials that change in response to outside inputs to represent smoke and fire.
Initially, I approached the smoke using a particle system but I found that the overdraw of the smoke sprites had a significant negative impact on their performance. Instead of particles, I built a fire and smoke material based on an approach by Klemen Lozar (https://twitter.com/klemen_lozar/status/1012028137247346689#m). I created distortion textures in Substance Designer, and used multiple layers of distortion applied to multiple sphere masks to define the different parts of the fire and smoke. I then created material instances so I could quickly make different smoke for the various fires in my scene.
The sycamore trees can catch on fire as a result of the explosions or the volcanic eruption. I created a lighter version of my smoke for them, and used vertex colors to control the progression of fire across their branches, leaves, and trunk. They also sway in the wind and shake due to the shockwaves from the explosions.
Mastery Pillar 1: Explosions - Retrospective
This mastery had the most progress early on and I felt comfortable from the beginning of the thesis project with achieving it. However, I had difficulty adding variety to my explosions to show off the mastery and struggled with incorporating the explosions into my environment through destruction and the shape of the fireballs. I also had difficulty optimizing things like smoke so that my artifact remained performant. I could have chosen a different format for my artifact to justify and show the different kinds of explosions more effectively. Also, while I did a lot of research into Embergen early in the project, I could have spent more time with the program later to achieve better looking fireballs with more volume. I learned how important subtle details like light, embers, and smoke were to selling an explosion. I also found that getting the timing right was critical for explosions, as even small variations could make an explosion feel powerful or underwhelming.
Mastery Pillar 2: Dynamic Weather- Retrospective
I achieved much with the weather transition I created from the direction of both particle systems and materials. I managed significant improvements in the appearance of my volcanic weather each milestone and feel that, by beta, it felt well fleshed out. However, I wish I had made the clouds look more realistic and added in ash building up on the ground to push the mastery even further. I realized that using a material to fake distant ash falling from the sky in combination with a particle system for the closer motes and sparks gave the weather an all-encompassing impression. I also learned quite a bit about how to create the impression of three dimensional clouds with an unlit material, including faking shadows.
Mastery Pillar 3: Dynamic Materials - Retrospective
I had a lot of experience working with dynamic materials going into the thesis project thanks to my role as the primary VFX artist for the cohort game project in the spring. As a result, I was generally confident in my approach to Unreal’s material graph system throughout the thesis process. A different artifact environment could have provided me with more opportunities to show a variety of different dynamic materials apart from in the weather and explosions. I learned quite a bit over the course of this project about what could be achieved with Unreal’s material system including the use of flow/distortion maps, how to create material functions, and the variety of things that could be achieved through material parameter collections.
Artifact - Retrospective
My recreation of the Casa Batlló looks good, despite being a complex structure. In hindsight, I would have chosen a different environment better suited to showing off my masteries. I would have also chosen an easier environment to create, so that I could have spent less time on it and more on my masteries. Despite this, I learned much about pseudo-organic, hard-surface modeling and the ways to use the material graph to make it easier to build up the environment.
Why is this Mastery
Through the development of different kinds of explosions for different parts of the structure, refining the timing, and making the explosion effect the environment, I created explosions that felt powerful and satisfying while having a method of implementation flexible enough to be adapted to a variety of use cases. By developing contrasting weather states and showing the changes in the sky, in the precipitation, and in the environment, I showed my ability build simultaneously immersive and dynamic weather through VFX and blueprint. With a variety of applications in almost every part of my artifact and a demonstration of technical competency in usage of the material graph, I showed that I have a strong grasp of dynamic materials and how they can be used to enhance existing VFX and improve performance through visual tricks.
I would be interested in seeing a project approaching explosions and/or dynamic materials that are stylized. During my research, I also found that there were more technically involved approaches to clouds, like using volume textures, which I think could make for a compelling, though difficult mastery. Later in the project, I learned about 2D ray marching as a way to light smoke and cloud meshes and think it would be good for a future student working on an artifact containing those to look into it. Having spent so much time working on the Casa Batlló, I think that a future student could use historically accurate Art Nouveau or Gaudí as the basis for a more environment art focused mastery. The most important thing for future graduates to consider is how well their artifact will show off their masteries.
When I began this project, I had a lack of confidence in my ability to create a complex environment or take on a project of this scale. I also had issues committing to a single approach and would often become bogged down in research, nervous about whether a certain way of doing things would work. However, over the course of the thesis, I found that even the most intimidating tasks were doable if I committed to them. I also faced a lot of setbacks and failures (both technical and personal) over the approximately three years it took to complete this thesis, and I had to learn to forgive myself for them and keep moving forward instead of letting them overwhelm me. Finally, I had to learn to find balance between working on the project and taking time to relax and recuperate. It was far more effective to commit to doing a realistic amount of work and giving myself the ability to have fun than it was to repeatedly work myself into burnout, then feel guilty about being unproductive.
Thank you to Professor Boris Fisher for advising me throughout the thesis project and designing the thesis process, SMU Guildhall for providing me with the opportunity to acquire an MIT in Game Developement, and my family for supporting me throughout.