Hybrids student film

Award-winning student filmmakers tell the story of Hybrids and how Arnold helped bring it to life

For their graduation short film, students from MoPA in Arles, France set out to highlight the issue of pollution. Hybrids was born and went on to win multiple awards including the 2018 VES Student Award, The Rookies’ 3D Animated Film of the Year, and Best in Show at SIGGRAPH which qualified them for an Oscar nomination. We spoke to the five co-directors – Florian Brauch, Matthieu Pujol, Kim Tailhades, Yohan Thireau, and Romain Thirion to learn about their tools of choice and the inspiration behind this stunning piece.

How did you come up with a story and theme that you could all connect with?

(FB) The idea came from Romain who’s from Cannes in the south of France. He used to dive to gaze at the underwater animals. As time went by, he noticed that there were fewer fish and more trash. Every time he would spot something shiny buried in the sand thinking it was some interesting form of life, it turned out to be a bottle cap instead, igniting the first spark which inspired the film. He shared this experience with the rest of the team and it drove us to create a short film to denounce this change.

What story did you want to tell and how did you want your audience to feel?

(RT) It is a story about evolution and the way it is being profoundly affected by the human action, showing that our ever-growing unawareness towards pollution is taking root deeply into nature, irreversibly. We wanted the audience to be entertained visually in order to connect to the grave subject of the film.

You only had nine months to complete the whole film. How did the team manage with such a tight schedule?

(FB) We tried to be careful on the time spent on each production stage, and we tried a lot of different things at first to find the ideas and the look we wanted. We did early tests and RnD of everything at early stages to be sure it was doable. After that we went quickly to a previz stage to lock all the key moments of the film and try to avoid redoing the same shot or sequence multiple times.

We knew that we would have to learn new stuff through the production, because we had never done crowds before, and almost all of the film is underwater, and that changes everything. We had to adapt the animation, the lighting, the way we were doing the fx and the compositing. It’s almost like if we had to learn everything again or at least transpose the techniques we knew. But all of us had already done some professional internships and we tried to bring everything we learned to achieve what we wanted.

We did a lot of dailies and each one had the opportunity to push the idea further, but in the end, it was their teammates within their department who would decide if it was doable or not in the condition we had, as if they were given a veto card or final approval.

We were able to render on the school computers during the night, but there were eight teams who had the same deadline, so you can imagine how it happened and that you had to share a lot of resources with everybody. We had to be smart and ultra-efficient with the renders or simulations to avoid long render times.

Are there particular features of Arnold that you relied on more heavily? 

(RT) We relied heavily on the fog and volumes to play around with beams of light in order to create specific moods for the shots. Arnold’s integration with compositing software is also great and allowed us to tweak our shots very precisely.

(KT) To work with Arnold allowed us to easily use the .tx process, having 8K textures wasn’t a problem, we had a lot of camera moves so it was really helpful to use this tool.

Tell us about your look development process.

(RT) We textured all our characters with Substance Painter, working specifically on the transition between the skin and the non-organic objects. We used the built-in IPR at some moments, but quickly jumped into Arnold with a turntable template to review our materials and shade our assets properly. The assets were then referenced into all our scenes.

(KT) Each environment had its own building process. Sometimes we needed to create each asset before building the background, like in the first sequence, and we used Houdini to add the seaweed. In some cases the environment was laid out first and then we had to find an efficient process for texturing all the elements efficiently.

Can you tell us a bit about the shading work?

(MP) We used Substance Painter to texture most of our assets, then exported the maps for Arnold. We were using the Alshader library so we had to color correct in Maya the exported maps from Substance Painter. All the assets were tested in a basic scene with HDRI to check if they were matching the desired look and were coherent with one another.

(KT) For the environment, we used a mix between Substance Painter and procedural shading, like in the cemetery for example.

(RT) The complexity of some materials such as the translucent/emissive shader of the shrimp required some complex network of maps.

What was your process for designing the lighting? How does the lighting help tell the story?

(KT) The first step was to define what atmospheres we wanted to put in the movie that would serve the story. When we were able to define the story and the events of each sequence we proposed several moods, starting by creating a colorboard/lightboard. Following this we defined what would be the master shot of each sequence, trying to get closer to what we wanted at the start and each sequence was defined following these shots. It was important that the lighting not only served to show what was in each scene but could also tell in a different way what was happening with different sets of light.

Can you tell us a bit about your lighting and rendering workflow in Arnold?

(KT) Each sequence required a different lighting process. Even if the base was the same in most cases, that is to say - to define the key light direction, an HDR, a rig of light for the caustics - it was necessary each time to adapt.

We still tried to keep things simple: to put a base of lighting, to make a very draft render for a fast rendering time, one frame every 10 frames to have a global overview of what we would need more precisely as an adjustment in terms of the passes for compositing, but also which elements would be optimized and what compromises could be made.

What were some of the lighting techniques you used to achieve the underwater look you were going for?

(KT) Surprisingly, apart from the work on the fog (more dense and always present) and the caustics rig, lighting an underwater environment was not so different from outdoor lighting. It was at the FX and compositing stages that the work was more important to create the underwater look (e.g. particles, fluid dynamics, color treatment, and chromatic aberration). Constant back and forth between lighting and compositing had to be done.

In the ‘making of’, you talk about the challenge of finding a balance between realistic and artistic. How did you manage this?

(KT) We wanted a progression in the lighting throughout the film, to start with enough natural and realistic lighting and to finish under water with a sequence with more direct lights and more fantastic atmospheres. By first working each master shot of each sequence we already had a global idea of the evolution of light. For realistic rendering we decided to stick with more natural blue colors, and a simple lighting set: HDR / Caustics / Sun. For more fantastic shots we did not hesitate to play with different lights and colors, we also pushed the general grading in compositing, putting aside realistic underwater principles: like the blood that we wanted vivid red. We also pushed more saturations and lowered the fog at certain times for a better reading of the actions or more dramatic rendering. However, it was important to maintain a water effect thanks to suspended particles, animations, and so on.

The turtle shot is one of our favorites. How did you approach the rendering of this complex scene?

(MP) We established our artistic goals early on for this shot. Knowing that we wanted to go full 3D motion blur, 3D DOF and 3D fog paired with a complex reflective and refractive character topped with more than 100 point lights, we were expecting insane render times at that point.

We started testing each main component separately. We locked the animation and then animated the focal point as well as the aperture on low res simple grey renders to speed up the process. In parallel we were developing the pipeline to export a particle simulation from Houdini to Maya and convert each one of those points into a point light with a random color and intensity (within a defined range, of course). The next step was to set up the fog. Once again, low res tests were made with some lights and the turtle to fine tune the look.

It was now time to merge all these different render elements together. We test rendered a few frames at a low sample rate, then tested the full sample rate before submitting the entire shot to the farm. Due to the huge render time we couldn’t afford any mistakes and had to make sure that all the settings worked well together.

What were the toughest lessons learned from the lighting and rendering process? Any surprises?

(KT) Especially for underwater lighting we quickly realized that it was very important to have a process of back and forth with the compositing, we also realized that the lighting process had to be developed in parallel with the creation of the environment. Creating an environment in its entirety and trying to light it at the last moment with the characters was not a good process. Each environment and character reacted according to a light, and sometimes requires adjustment. It was very important to keep a flexible pipeline between the departments.

What was a surprise for us was to realize that lighting an underwater or outdoor environment was finally not so different - it was especially in the fx and compositing stages that everything played out.

What were your render times like?

(MP) On the best part of our shots we managed to keep fairly low render times: between 25 to 45 mins per frame, except for the turtle shot that ramped up to 5 hours per frame.