Fractals and Procedural Production

Heribert Raab talked to 80.lv about his most recent series of incredible 3d fractals, created with Houdini. He also was kind enough to share his thoughts on the future of game and movie VFX. Heribert is a VFX artist, who worked on a bunch of movies, including Rachet & Clank, Captain America 2: Winter Soldier, Pixels, Thor & Avatar! You can find his full portfolio here at IMDB.

Introduction

My name is Heribert Raab and I am originally from Germany. I am certified electrician, but for some mysterious reason I was converted into VFX. I’ve started computer graphics as young kid in 1992 with an Amiga 500. In 2000 I’ve started my carrier in VFX as Animator / 3D Artist. Over the last couple of years, I’ve become an effects technical director. I’ve worked for most of major animation / VFX studios worldwide, such as Weta Digital, Dreamworks Animation, Digital Domain, Animal Logic, Sony Imageworks. Currently, I am working at Rainmaker Entertainment as Lead Effects TD. In my free time, beside being a dad, I am doing lots of different art.

Series of Fractal 3D Images

I was always enthusiastic about fractals. I love this awesome mix between creative and technical. It all started with simple Java scripts to draw points and lines. It was quite simple and had basic coloring. Later, I played around with 2D fractals software. You can get a nicer shading out of it, but still 2D.

After a long time, I had enough of the 2D look and wanted to create something in 3D. I looked into the Mandlebulber Software, which is a great tool, but it is a pixel based calculation and no real 3D. It requires too much calculations to use Mandlebulb equations in a real 3D. Also with the Mandlebulb you fight a lot with mirror-effects and level of details poppings. I wanted a system which I can use in a real 3D package, with real camera’s and all tools which a 3D software provides you. I know some guys, who use a Mandlebuld system in Houdini or Maya, but it always comes with loss of details or render quality.

Last year, I’ve started exploring 2D fractals, with spline shading technique. I am very satisfied with it, it’s more like 2.5D fractals. Now I’ve moved to “phase 2” to bring 2d fractals into 3d Space.

With the current series of 3D fractals, I am trying to explore meshing and render techniques. These are very basic and simple fractals meshed into polygons. I am planing to use this “Fractal Primitives” with new attractors to create much more complex scenery. With fractals, you can create billions of polygon in minutes, the key is to manage the data and reduce it to the minimum without losing too much details.

I’ve decided to reverse the fractal imagine generating. Starting with a simple geometric shape like a line or cube and going more and more complex. The advantage against math equations, I have full creative control and I can use all kind techniques to make meshing and rendering more efficient like render-instances, delay rendering or curves with normals etc ….

I am about to start new series of images to explore different meshing methods like volumes before I am going to the next level and populate this fractals into complex scenery.

Using Houdini

I’ve started with Maya and then Modo before I moved to Houdini. In Modo, I have fast raytracing engine which makes it possible to render huge amount of polygons with realistic materials. I enjoyed it a lot, because I prefer good looking fractals to boring fractals with great math solutions. But the procedural nature of fractals lead me finally to Houdini. I really know this software well, because of my daytime job as Effects TD. Houdini gives me a lots of tools to create any kind of fractals and the concept of the software makes it possible to handle an insane amount of data.

Production Process

Mostly, I am starting with simple polygon line and use some deformers, mostly Point SOP or VEX snippet, to create some interesting shapes. After, I am using copy nodes,Vex snippets or Foreach nodes to loop through many iterations to generate complex structures.

I am using point clouds to calculate attributes like density or tangent and mixed it together with color ramps.

At last, I am using poly or volumes tools to mesh it out for rendering, and thats it! The beauty of fractals is simplicity, it’s just doing the same thing over and over again, with little randomisation. The challenge is to manipulate the iteration in a way, so that it looks good.

Rendering

Mostly, I use Houdini’s Mantra and sometimes Modo. lately, I’ve been using more Mantra and started to look into Arnold. Modo has a great Renderer but it is to isolated from Houdini. I need a direct render access to curves and instances in Houdini. Mantra is pretty cool, because it uses REYES render method. It is very efficient to render huge amount of data, when using pre-generated colors and occlusion, it renders also very fast.

Procedural Tools

I love procedural tools. They let you focus more on the creative part and not worry about technical part. Procedural tools can create realistic content much faster and in amore efficient manner, but it still requires a creative artist. There will be a point, when people get bored about super realistic content and desire more creative non-realistic content. You are going to movies because you wanna see something different, something which takes you away from your daily life. I think movies will split into to directions: event movies and story movies.

You will experience event movies like fancy motion rides with VR-Headsets, where the movie puts you in the middle of action, when aliens are destroying whole city blocks or so. And there will story movies where story is the key. Pixar movie with a VR-headset does make sense, because you have much freedom in visual respect. In a VR movie you can look around, look what is behind you, but you may miss important dialogue and gesture from characters. If you tell a story, you want the audience to pay attention to story and its important elements, driving the story.

Are Tools Changing?

I think they will not change much. The need for more screen resolution (special for VR) will need more computer power.  Tools will be smarter, faster and overall better, but a real evolution will take a longer time.

There are some VR Tools for sculpting, which is cool, but at the end, you get really tired when you have to move around all day to sculpt a character in a “Virtual Reality Zbrush”. It’s much more comfortable to sit in you chair and use tablet for sculpting.

The realistic content creation will be much easier. Image modelling software will create most of the 3D models. Modellers will do clean ups or shift into more non-realtistic modeling, as some for animators or lighting artists.

Artists will shift into more creative tasks and a wide part of realistic content creation will be taken over by software. Biggest advancements will lie in the details, software and hardware will grow into more pleasing experience. It will be much easier to be creative.

Movie-quality VFX in Interactive Entertainment

One day – yes, but right now I don’t see it. The demand for movie-quality in games is not that big. People seem to be happy with current quality, they want more social interactivity in games, creative gameplay and games on smartphones. People are not as crazy about computer technology (geeking out about graphics cards etc.) like years ago. This slows down the race for more processing power on consumer level.

Interesing links

BUILDING PROCEDURAL EFFECTS WITH HOUDINI

TAKING ENVIRONMENT ART TO THE NEXT LEVEL

USING HOUDINI IN GAME DEVELOPMENT WORKFLOW