Case Studies
Egg Explosion (Overloadgraphics)
You’re a student, working on your final project, and you have only five months experience with RealFlow. Then you create an astonishing simulation of a bullet going through an egg in slow-motion. Is that worth a case study? Of course it is! We spoke to Anthony Chappina to get the ins and outs of his Egg Explosion project…..read and be amazed.
“I grew up outside Atlanta; my family has been here for 18 years or so. I got into 3D soon after starting my bachelor’s degree in 2001. I had an idea of what 3D was and generally how it was used, and when a fellow student showed me Maya I was totally hooked. The school I attended at the time did not teach anything related to 3D, so I pretty much taught myself, watching tons of tutorials, digging around online endlessly until I found out how to accomplish something.”
That’s how it all started for Anthony Chappina, and from there it went up, up, up. He graduated in Fine Arts and went on to do a Masters in Visual Effects at the Savannah College of Art & Design (SCAD). “There, I began to specify what I wanted to get into, which is effects. I landed an internship at Turner Broadcasting soon after starting at SCAD, and that made me realize that I had no idea what I was doing,” says Anthony
 .
He must have had some idea though as he soon landed himself a job as a full time artist at Artistic Image, and Magick Lantern after that, where he started working on a broad range of visual effects and compositing. Anthony passed his SCAD classes with ease and decided to go for a full-fledged, refined final project that mainly revolved around fluid dynamics and rigid body dynamics. Target: an egg exploding at high-speed. Weapon of choice: RealFlow.
“When I started work on this project, I had been using Maya for about 6 years, and then at SCAD I was introduced to Houdini. I started to get into particles, which at first drove me insane. What's nice about RealFlow is that it’s very easy to transition to. After using Maya and Houdini for a while in production, RealFlow was pretty much learning how fluids work, not how to use the software. The interface and workflow are pretty straightforward,” explains Anthony.
“I found a few really amazing high-speed photos of eggs being shot,” he continues. “I wanted to do a project with fluids and rigid body dynamics, which recreated the effect realistically. By choosing this specific project, I would explore a broad range of aspects in RealFlow, and fluids in general.”
“Fluids were the one aspect of effects not really covered at SCAD. I wanted to do something with fluids before finishing school, so I could put it on my reel. So I taught myself RealFlow by watching tutorials, reading on forums, and lots of playing around,” says Anthony. “I also used RealFlow on a commercial about a year ago, while working at Artistic Image, and really got into it for a couple of months. I enjoyed working with the software, and realized how powerful it was. Also, since I was already used to working with Houdini and Maya, the workflow just seemed to fit right in.”
A final course project, class discussion and feedback were essential if not compulsory. “We ended up with only 4 students in the class, and everyone did something completely different,” recalls Anthony. “Twice a week we met for class, did dailies and critiqued each other's work along with the professor. We pushed each other by really nit-picking our work apart, and with every class I was really excited to get and give feedback to others. This really inspired me to push myself.”
Anthony continues: “The RealFlow forum (www.realflowforum.com) gave me lots of inspiration as well by being able to look at projects that people had worked on or were currently working on. I received great comments and feedback from everyone on there….whenever I came across a problem, someone always helped me out with a suggestion or in many cases a solution.”
So much for the inspiration and feedback, but how do you go about re-creating an exploding egg with RealFlow, having little experience? Simple: you just do it.
RealFlow was used for all the fluids and rigid bodies in the scene. “I did research on how RealFlow integrates with Maya before even opening up the interface and beginning the project. That gave me a good idea of limitations and issues I could run into,” explains Anthony. “I started by setting up my workflow, so I could hand-fracture my shell in Maya, and then import everything into RealFlow. Specific fractures were made by looking at reference footage and mimicking that,” Anthony continues. “From there I simulated my rigid bodies in a scene of their own, with no fluids. I then took the SD file from that, and brought it into my fluid scene. This way my RBD was written to disk and no longer had to be calculated, allowing me to focus on one aspect at a time.”
“In addition, with the help of my event scripts I was able to get my RBD simulation down to about 5 to 10 minutes and that gave me the ability to tweak everything over and over to get the result that I wanted. I found out early on that attempting to do both my simulations at once created enormous simulation times and unpredictable results, so this was a good solution. After my fluids were simulated and meshed, I took everything back into Maya for lighting and shading.”
“I found the rigid bodies in RealFlow really great to work with,” continues Anthony. “I had used rigid bodies in other packages before, but the ease of working with them in RF was great and very straightforward. I originally thought that the rigid bodies would cause some separation with all my shell pieces, and I would have to do some trick compositing. But everything held up and I was able to render my scene straight without having to go back and composite in a non-fractured cell,” he concludes.
A tiny cloud appeared on the horizon when it turned out that a few pieces of shell were moving too quickly and would throw fluid all over. Anthony encountered a swift solution though: “I either turned off the global link or keyed off their activity,” he says. “Since my shell was already animated/ simulated, that gave me the ability to really sculpt my fluid. The motion of the shell completely determined how my fluid would look, so if the shell exploded, then the fluid did as well.”
Another aspect of the creation of the scene that seemed easy at first turned out a tough and time-consuming task. Anthony explains: “Meshing, lighting and shading ended up taking a while mainly because there are two fluids, one inside the other. The yoke would sometimes escape the albumen and when that happened I would get really odd rendering artifacts.”
“In order to get around that, I had to build my mesh a certain way,” Anthony continues. “I used two meshes, one for the yoke and one for the albumen. For the albumen mesh I had to add both fluids and also inflate the yoke settings, and this way when building just the yoke mesh, I knew that every frame’s mesh would be completely encapsulated by the other. By doing away with the mesh interactions, I solved my rendering artifacts.”
But Anthony wasn’t done solving his rendering artifacts issue or yet another challenge was presented to him, again mesh-related. No matter how many particles Anthony used when the mesh was built, it always went outside the thickness of his shell geometry. “I created a MEL script that would allow me to animate the thickness of all my shell pieces with a custom attribute,” explains Anthony. “First MEL was used to match all my original Maya geometry to the animated RBD shell from RealFlow. That allowed me to keep all my custom attributes. As my simulation played out I animated the local position of the inside and outside of my shell, doing away with the mesh escaping.”
Anthony used various other scripts to control if objects were active or inactive. He says: “Both the batch and event scripts helped my workflow. I had issues at first with the SD format, which ended up being my shell geometry, and because of that I was using MEL to export all my shell pieces as individual OBJ files. The batch script I used imported everything and rebuilt it inside RealFlow at the click of a button.”
Scripts, rigid bodies, rendering artifacts…. No problem for Anthony. Even the tough nut of scaling – something that can often be difficult to establish in RealFlow – was cracked effortlessly. “Scale made all the difference,” he says. “If you look at my website, all my original tests were at a large scale. Once I cut that in half I got much more realistic and stable results. That's when I really started to fine-tune everything. When I finally figured out how most of the daemons affected my fluid and RBD, I was able to make things look how I wanted instead of getting a random result. Though I cut my scale in half, it was still not ‘real life’ scale,” Anthony explains. “If I went that small I had equally unstable results as a really big scale. So I had to find a happy medium.”
It was mostly good times with RealFlow though, as Anthony confirms: “I really like how RealFlow is specifically just fluids and dynamics as you tend to focus on just that when using it. It is not bogged down with useless tools like some 3D programs are. It’s straight and to the point. The command line simulations are great,” he continues. “I would set up multiple scenes every night and just batch simulate everything. I also really feel the ease of implementing it into your 3D package. I have used RF with both 3ds Max and Maya with no real issues.”
Room for improvement? Of course! “Simulating with lower resolutions can have drastically different results than high resolutions, though it is a great deal faster,” comments Anthony. “It would be great to use lower resolution settings, and then after the simulation is baked, add resolution back to the fluid, although I am not sure this is even possible given the nature of fluids! Also, I would like to see wetmaps improved as there aren’t many settings for them at the moment. I had really random results with the maps and in the end I never used them.”
Having successfully completed the Egg Explosion final project, Anthony continues working at Magick Lantern, where he is – contentedly – getting more into effects and compositing work. “I enjoy working on projects where I am not restricted, and can go my own direction if I want,” he says. “If I could do compositing and effects all day I would be happy, maybe mixed with some fun motion graphics.”
We are certain that happy day will come sooner rather than later for Anthony.
Just watch.
Credits:
Anthony Chappina
www.overloadgraphics.com/FluidDynamics/
www.anthonychappina.com/FluidDynamics/
Thanks to:
www.realflowforum.com
http://www.rf-magazine.com
Professor Clarke Stallworth
Professor Tan Tascioglu
Marshall Petersen
Jeff Wyner
Diana Li
SCAD Students!!
Magick Lantern
Mom and Dad!!
|
