Speeding up fluid FX simulations
Igor Zanic, freelance FX artist and co-founder of online school Rebelway, shares his thoughts on the Adaptive FLIP solver in Houdini 18
Igor Zanic discusses the Adaptive FLIP solver in Houdini 18
Igor Zanic
Igor is a freelance FX artist based in Serbia, and is the co-founder of Rebelway. He got into 3D after being inspired by movies like The Abyss and Jurassic Park. igorfx.artstation.com
3D World: It has been a few weeks since the reveal of Houdini 18. You were one of the artists who was working with Sidefx before the public launch, using the new adaptive pressure projection for FLIP fluids. What do you think about this new feature, and how does it compare to performance and timesavings in older versions?
Igor: FLIP simulations are a really demanding and slow process; over the years when I started doing simulation work, I didn’t have a powerful enough workstation, so optimisation was a big part of my process. Now everyone can have a powerful workstation and they don’t think about these things so much, but trust me even on a powerful workstation, some demanding simulations can be slow.
An Adaptive solver is solved on an adaptive (octree) background grid to significantly improve performance over solving with a uniform grid. In order to maintain highresolution details in the simulation, a band of fine grid cells are constructed along the boundaries of the fluid. The grid is then progressively coarsened into the interior of the fluid. To minimise the difference between the adaptive and uniform grid pressure solves, the adaptive pressure is upsampled and smoothed back to the uniform grid before updating the velocity field.
An Adaptive solver helps on FLIP tank setups and it can speed up your simulation up to 1.5x-3x, but keep in mind that to gain the most out of Adaptive, you should have proper depth in your simulation. A shallow river will not benefit that much from Adaptive, compared to a large ocean with 5m-20m depth tank. On a shallow example, there is no room for Adaptive to happen. I did a lot of comparison examples where you can see how non-adaptive and Adaptive work on the same shots, and where it is better to use one over the other.
In the past, large simulations with deep tank (e.g. a submarine emerging from the water, or a huge animated monster) would be a nightmare to sim. Those required a lot of memory and CPU power. With Adaptive, that is now easier to solve and allows for more detailed simulations.
3D World: You are well respected in the FX community for your meticulous testing and mastery of fluid FX simulations, not only in Houdini but other capable packages. Where would you say FX tools are most lacking in, and what would you like to have created or improved?
Igor: Software can now do amazing things; artists really push boundaries and make beautiful artwork. For me, in terms of water simulation, all of these tools give you 80 per cent, sometimes even less, of what you need. Big studios have their own solution; they have people and resources to make everything that they need, but for the single artist, that can be a big problem sometimes. I understand that software developers need to focus on the entire software package and they want to add all new stuff and support many artists, but from my point of view, that can be a problem because you just do base support or make tools that cover some aspects, but problems may arise if you want to do something else, or it is not supported at the moment.
I would like something along the lines of a complete setup or a setup that is more optimised and polished. You need to do something, but you need to use several applications, so even simple tasks can be a nightmare to do, because you don’t have tools and support in the current software version.
I would select one task, and see what is needed to achieve them, and try to make support for everything that is needed. Then, the artist can try and see if it is possible to use the same tech and support different examples, or if something further needs to be added.
3D World: Over the years you have worked on many different projects, with many clients. What still drives your passion today? What would be your dream project?
Igor: In the past, there were a lot of things that were very hard to do or even impossible, because of software and hardware limitations. Big studios had a lot of hardware support and also amazing people that wrote and still write in-house tools. But as a freelancer, you just didn’t
have that power. Today, it is a lot easier. You have tools and hardware that are cheaper and more powerful, and that really give you a lot of freedom.
So I always look back and see what I did or what I wanted to do. Things like movies and other people’s work that pushed me into this industry – what I could not do before, I now have the chance to make them. So I try every day. When I do projects and personal work, I learn a lot from it, and try to push and do better next time.
As for a dream job, that’s something that is for me hard to say. I would like to work in the studio on a large-scale water project, working alongside other artists and learning from them. On the other hand, I would like to make something on my own, and leave a mark on the community. Working in the studio is amazing, but you do what they tell you and how they want, which is fine. But when you do your personal work, it is your vision and you put your body, mind, heart and soul into that artwork. Most of the time people think, ‘ahhh, it’s too hard to do at home, don’t waste your time’. But artists can do amazing stuff at home.
3D World: Aside from learning all the different controls/parameters (and there are quite a few), what advice can you give about being patient, persistent and motivated?
Igor: Because there are many parameters that a user can change, which can impact the look of your simulation, they might give you different looks depending on what kind of task you need to do. I always like to spend a few more days exploring how settings work on the current task, no matter how many sims I did in the past. I know which parameters I can use, but still you learn a lot when you test them all over again; in this way you create your own library of tests that you can use in future projects.
Also an important thing for me is that many people tell you this parameter will do that or this, but sometimes it’s really hard to visualise, so performing comparison tests and learning from them can save you a lot of time later. I know artists don’t like to do that, because it takes time; sometimes I do more than 200+ compare tests in one week, but for me that’s fun, and many times I got some new ideas from just seeing what I did the day before. Challenge yourself every day, and try to make the best out of it. Sometimes it will not work, but that’s fine – just relax and try another day.
“AN ADAPTIVE SOLVER HELPS ON FLIP TANK SETUPS AND IT CAN SPEED UP YOUR SIMULATION UP TO 1.5X-3X”