Azimuth: Creating a Space Mercenary Character for a Sci-Fi Movie

Introduction

We are eight artists who worked on Azimuth, our graduation film at ArtFX. We all come from different backgrounds, but we share the same drive: having fun while creating something we truly love. That's how the whole adventure of this film began.

The team: 

• Thomas Courtois (FX Artist)
• Thomas Teisseire (Director)
• Chloé Coudray (Character Artist)
• Cassandre Cinier (CG Generalist)
• Damien Lagadeuc (FX Artist)
• Mathis Giraudeau (3D Animator)
• François-Clément Michez (Compositing Artist)
• Martin Bluy (CG Generalist)

Thomas Teisseire: It started as our graduation film at ArtFX, but it quickly grew into something far more challenging. Our team shared the same goal: to create a small-scale sci-fi 'mini-blockbuster' and immerse the audience in another world within just five minutes. We also wanted to make something fully CG and semi-realistic, which allowed us to create the world exactly as we imagined it. The only real limits to our imagination were our computers and the time we had.

I wrote the film in the summer of 2023, and the story kept evolving through the strong artistic input of the team, from concept art to visual research and worldbuilding. Despite the technical constraints, the core remained unchanged: an intimate story about the mutual sacrifice of a brother and sister. Hadenna wants to save someone she loves, but ultimately realizes that this person is the one who saves her. Her impulsive choices come from the fear of losing something she believes is gone, until she understands it was never truly lost.

As director, I aimed to keep a clear vision and ensure that the technical aspect serves the emotional arc rather than overshadowing it. Even though we still wanted to preserve a strong sense of spectacle.

Hadenna Character

Chloé Coudray: For Hadenna, our space mercenary marked by a painful past, we aimed to create a design that felt unique yet aligned with our semi-realistic art direction. I began by gathering references for silhouettes, proportions, and style before shaping a distinctive concept.

Her look blends medieval-knight inspiration with a futuristic sci-fi aesthetic: a form-fitting suit, lightweight armor, heavy-duty helmet, and a burning sword. Her bond with her tribe is engraved on the armor, while her stoic mask exposes only her eyes, contrasting her hardened warrior side with a fluid, feminine silhouette. Above all, we wanted viewers to sense the untold story behind her.

I started by sculpting the naked body and face from a base mesh in ZBrush to define proportions, musculature, and silhouette. After validating the proportions and facial structure, I completed the retopology in Maya before building the suit to work with clean, animation-friendly geometry. I then separated the head from the body, the face becoming a hollow mask since only the eyes would remain visible, while the body served as the base for the spacesuit.

I then refined the face in high resolution in ZBrush, adding pores, folds, wrinkles, and scars using alpha projection, and later exported it as displacement maps for texturing and lookdev.

The suit, armor, and sword were modeled in Maya based on the concept art. For the helmet, I experimented with shapes in ZBrush to explore variations freely, then finalized and retopologized it in Maya.

Texturing and Grooming

Once the model was complete, I unwrapped the mesh and created clean UVs in Maya. I then textured in Substance 3D Painter the character using a PBR workflow, adding materials, roughness, normal/displacement maps, and all surface details as skin marks, dirt, fabric fibers, leather, metal wear, and decals.

To support Hadenna's evolving story, I created three texture variations:

• Battle-worn: scratches, wear, and tribal engravings.
• Rain-soaked: droplets and streaks.
• Post-explosion: soot, scratches, and facial injuries, including a damaged eye.

Lookdev was done in Houdini Solaris with Arnold, defining, for example, skin subsurface scattering, armor speculars, mask reflections, and all shader properties for realistic integration under lighting. 

For character grooming and simulations, Houdini was perfect, especially for the final shot where Hadenna appears without her helmet. We wanted her hairstyle to be natural and indicate a recent battle. I created her long eyelashes and natural eyebrows, and added peach fuzz to enhance realism in the final render.

The animated character, like all our hero assets, was rendered on separate layers to simplify compositing, using the USD-based Solaris workflow with Arnold in Houdini.

We had a lot of different environments, going from sci-fi industrial bases to Thailand-inspired landscapes, so we needed to be efficient and save time while keeping a visual consistency all along. First, we modeled an entire kitbash library and reused it through production on every hard surface environment.

Then we build and use tools on Houdini Solaris to speed up the workflow, such as an asset builder. This one allowed us to export each asset with its variants, LODs, proxies, and materials, into our asset gallery. Finally, we used DMP and camera projections as much as possible.

As an example, in the first sequence of the movie, all tropical mountains are projected or matte-painted. Of course, working closely with the FX and animation departments was crucial to avoiding assembly mismatches or surprises.

Creating the Scenes

This shot was created by doing a lot of comes and go with the FX artist. As optimization was the key to this heavy shot, we split the terrain, creating one on both Gaea and Houdini, from the simulation. We instanced on the points of the simulation several layers of small mud, pebbles, debris, leaves, grass, wild plants, and more to fit the references and make the close-up shot credible.

The middleground needed less definition than the foreground. We scattered using decreasing density, mesh, and texture LODs regarding the distance. Then, the background was made entirely in DMP to save time while adding more realism. Finally, lighting with a setting sun and strong atmosphere helped bring more depth and a humid feeling.

Mathis Giraudeau: To give Azimuth its cinematic style, I filmed myself performing the scenes and used the footage as a reference. Thomas (the director) was always present to validate or adjust my acting choices. Then, in Maya, I animated based on those references, refining and exaggerating some of Hadenna's expressions and poses.

For the action scenes, especially the fight sequences, I wanted them to feel both impactful and believable. So I found and edited some references from movies, video games, and real-life swordfights to achieve the most realistic and professional approach possible.

And of course, the layout of these scenes was essential in emphasizing the action, for instance, through tight framing, single-shot sequences, and slow-motion scenes.

Thomas Teisseire: At the beginning, we made several previs tests using a simple iPhone, with cardboard stand-ins for the robots and a paper plane as our spaceship. Even during these rough tests, I already knew I wanted a camera that rarely stayed still, a camera that kept moving forward as if drawn toward something inevitable.

When we transitioned to layout, I kept that intention in mind and intentionally avoided classic establishing shots. I wanted the audience to feel that we had no choice but to keep advancing, just like the characters. In our approach, establishing shots appear only after the action, almost as a consequence of what just happened, rather than as a way to introduce the environment beforehand.

Throughout the film, it was important for me to preserve a human-level point of view. I wanted the camera to remain physically close to the characters so the audience could experience each moment as if they were right beside them.

On the technical side, we added subtle rotational noise to break the perfection of CG movement and give the impression of a real handheld camera. Most shots were approached as shoulder-mounted moves. And because our environments, especially the Extoria base, are entirely fictional, we put a lot of effort into designing movements that emphasized parallax and depth, helping the world feel grounded and giving viewers a strong sense of scale. 

Thomas Courtois: For the VFX, among the shots I worked on, the solar beam and the large explosion were the two main effects I developed. For the solar beam, I built a setup combining a mesh-based core, pyro simulations for a more organic shape, a dense particle layer for extra details, and several custom AOVs.

The goal was to create something physically intense and blinding while keeping it fully art-directable in compositing. With the custom AOVs, the compositor had deep control over shaping the energy, pushing the heat distortion, and amplifying the blinding impact of the beam.

For the large explosion, I handled both the destruction and the pyro work. I split the effect into three separate explosions to gain more control over the timing, scale, and readability of the shot. For the RBD setup, the environment was extremely heavy, so I first created a proxy version of the fractured geometry to speed up the simulation and keep it stable.

Once the proxies behaved correctly, I added custom constraints across different sections of the structure to control how each piece bent or broke apart. On top of that, I added several secondary FX such as particles, debris jets, shockwaves, and smaller sims to bring more complexity and realism to the shot.

The challenge here was to combine both simulations so they felt connected and accurate. Overall, my goal with these FX was to make them feel dynamic, grounded, and easy to integrate into the final shots.

Damien Lagadeuc: We had to think carefully about how to optimize and reuse our setups to save as much time as possible. Most of my FX shots, especially the first one with the robot crawling through the mud, were extremely time-consuming to render because of the high level of detail and the proximity to the camera.

This meant optimizing everything based on the camera's perspective to give the illusion that the entire ground was simulated, even when only part of it was. Another particularly challenging FX was the bridge explosion. There were huge amounts of polygons and many groups to create before we could even start simulating anything.

A significant amount of prep work was needed just to set up the simulations. We decided to use USD throughout our pipeline, which was new to us, so we had to quickly adapt and find ways to render our FX in USD without overloading the network storage.

Martin Bluy & Cassandre Cinier: Regarding the lighting, consistency, and diversity were the main challenges. In the early stage of production, after gathering references, we blocked and lit all the key shots to define the mood across each set. Then we could easily duplicate and adapt the lighting sets to the needs of each shot.

Regarding the big scenes, mainly in the second part of the movie, we used a lot of light instances in Solaris, allowing us to save render time. We worked closely with the compositor to answer all the needs he might have regarding aov's and light groups.

François-Clément Michez: During our first meeting about rendering, one thing quickly became clear, every element of the scene needed to be rendered in its own layer. Meaning that, during the rendering setup, we would separate characters, environments, FX, and atmospheric elements so that I could fully manage each one of them in compositing.

Because we were aiming for a very intense "in-camera" feeling, keeping a close eye on Depth of Field was essential, especially since it would be prominent in several shots. Working with multiple layers allowed us, on the technical side, to achieve accurate defocus on each element without artefacts, particularly in cases involving semi-transparent or very thin objects.

On the other hand, on the artistic side, it also gave me the freedom to cheat the DOF when needed, using it intentionally to help storytelling and improve cinematography. Another crucial aspect for me was how our lights were rendered. We needed them separated into individual light groups as well.

Since Azimuth is a sci-fi film that relies heavily on lighting to establish its different moods, it was important for us to retain control over the lights even after the lighting stage. By splitting every light, the compositing phase became as much an artistic process as a technical one.

It was truly the moment when the atmosphere, color palette, and overall feeling of each shot came to life. On top of that, having the lights separated allowed me to create flickering effects. Especially in the final act, when our protagonist is surrounded by fire.

Conclusion

Bringing Azimuth to life, from writing and directing to every stage of production, took 22 months of work. We faced many challenges, starting with learning how to collaborate as a team using new tools, and pushing ourselves to create something believable, emotional, and memorable.

We are extremely proud of the final result, but even more grateful for having experienced this adventure together. All these moments brought us closer as a team. And we couldn't make this possible without our friends and family who helped us a lot from the very, very beginning.

Damien Lagadeuc, FX Artist 

Interview conducted by Gloria Levine