I majored in Japanese Language Education at college back in 2001 in South Korea. About that time, I fell in love with many Disney's animated movies, so I started learning 3D graphics by double-majoring in Media Design at the school. Because I had no idea which part I wanted to specialize in, in the field, I just kept focusing on 3D modeling - probably the easiest way to enter for beginners - for many years. I experienced various modeling tools, such as 3ds Max, Maya, Rhino and AutoCAD, which I enjoyed a lot.
In 2007, I began my career at a small place called IA Studio, an animation studio in Seoul that makes mostly TV series. One day, I went to the movies to see the first Transformers with all my colleagues. The movie shocked us all. I never watched such realistic CG robots being smoothly composited into live-action footage, with spectacular heavy effects, like explosion, destruction, and the robots transforming under the same camera and light conditions. Since then, I dreamed of being an FX artist, and that brought me to the United States in 2009.
I attended Savannah College of Art and Design to study visual effects. The school provided me with the perfect courses to become a VFX artist. I learned a broad knowledge of VFX from classes covering many different kinds of tasks in the field, especially basic classes, like Renderman Shading Language and C++. This gave me the chance to think about what VFX are really about for the first time. That naturally led me to the Houdini world, which uses fundamental mathematical and programming knowledge. After the long student life away from Korea, I got finally hired as a Houdini generalist by Method Studios in New York City in 2013.
GE - Childlike Imagination
There were no technical difficulties for the job. The creative director and I spent a lot of our time trying many different kinds of looks until the client found their favorite one.
I started the job by simply bringing the famous GE logo into Houdini. I then set up a simple and quick simulation, where particles were being generated from the logo. For a more interesting, swirly motion, I applied a vortex force to those particles with a little bit of noise. Then the animation was played in reverse so it looked like particles were forming the logo. The plexus effects were created out of those particles. Houdini has a convenient node called "Add SOP" that automatically connects points, making lines under certain conditions.
I also paid attention to shading, lighting and rendering. I ended up increasing the number of particles and lines quite a lot, making sure the moving sculpture looks as a whole with clear highlights and shadows in render time while those tiny individual elements fly free. The render time of each frame went pretty long because I applied the subsurface scattering shader to all the points and lines in the scene, which was quite expensive.
Samsung - Center Stage
Although the latest Houdini 17 has just been equipped with the strongest cloth system than ever, the tool used to be known for its horrible cloth dynamics. It was really slow and inconvenient compared to the Maya's Ncloth. The FX supervisor challenged me to go for Houdini for the consistent pipeline at Method Studios.
As for the cloth folding shot, if I didn't work in reverse, those cloth would be tangled to each other without arriving to the right place because cloth was colliding with each other. Instead, I started with the folded state. I intentionally made cloth pieces fly off one by one, from top to bottom, so that the cloth would naturally sit on the ground from bottom to top when reversing.
In Houdini's cloth dynamics, if one specifies a certain shape as a rest object, the original piece would have a tendency of returning to the rest shape during the simulation. Similarly, I was given unfolded cloth models paired up with folded ones for the effects. This way, I could successfully change the folded cloth to unfolded seamlessly while floating around in the air.
Finally, I let the cloth move to the left side of the picture frame by simply adding a wind force, which was the laundry machine location. This way, cloth could look like coming from the back side of the machine, when playing in reverse. The project was pretty successful. We all were proud of the result as the concept of the cloth effects was not easy.
Monster.com - Opportunity Roars
The ad was one of the biggest jobs at The Mill. For the hair’s movements, I did the fur simulation with small amount of hair first, then applied its result to the rest of them. Of course, we worked back and forth with animators and hair groomers all the time whenever there were any changes.
The entire workflow was very intense and slow, which required a lot of patience. Every time there was an animation change, I had to bring the new animated model into Houdini and do the muscle simulation, which took forever to calculate. Only when going through the muscle sim was successfully, was I able to do the low-res fur sim that interacts with the updated muscle, which then went to the conversion process to the high-res fur.
The most challenging FX part to me was how to describe the natural motion of the giant monster's hairs. If I made them too flexible in dynamics, it would make the monster's size small and light. If too stiff, on the other hand, it would lose the liveliness of the creature. It took me a while to have these kinds of controls in Houdini. Another challenge was the fact that the supervisor wanted to see more variation in terms of hair movements. For this, I did multiple hair simulations with different results, then applied these onto different spots on the monster's body, which helped make the creature look enormous.
Sonnet
The project was a large-scale collaboration of many talents from different departments. Basically, we had to recreate NASA’s rocket launching scenes, including the rocket ship and its facilities. Thousands of modeling pieces, along with their textures, were provided by modelers then lit and rendered by lighters.
As for the FX elements, the most important thing was to make the rocket ship looks big enough. For this, I had to keep the speed of all simulations as slow as possible and the resolution of volume as high as possible. When working on the CO2-like smoke being released from parting lines of the ship's cold body, it was difficult to get the high-res sim for the whole body at once. Instead, I did five or six different high-res sims, then combined them together, which caused a lot of data usage.
This applied to the blast-off shot as well. Multiple different smoke/pyro simulations were done and combined for the massive scene. The shot was tricky because I had to manage the smoke to rapidly be poured out from the rocket due to the massive energy, then to be slowed down somehow so that it maintained the large scale. I also created additional effects for the shot, such as ice debris being chipped off from the body due to the strong heat.
I tried to re-create these effects as close as possible to what happens in the real world. I spent a lot of time on watching NASA's real footage of what's really happening when a rocket launches. The ad turned out really successful, and I enjoyed it during the entire process.
Sewang Kim is currently working as a VFX artist at The Mill in New York City. He came to the United States from South Korea just over seven years ago and has worked at Method Studios, Sony Pictures Imageorks and Buck TV. He can be reached at: sewkim80@gmail.com.