The Sea's Heart
The Sea's Heart is the project I hold most dear, because with it I learned that consistency and discipline are key to achieving any goal. At that time in my life, my time was extremely limited: I worked at a VR studio and also trained and competed as an elite athlete. This made it take six months to finish the project, moving forward little by little, despite months with repetitive tasks like lookdev. Not giving up paid off, and today I can say it is, without a doubt, my best personal project.
From fearing the sea to diving with bull sharks
Everything started with the movie Avatar 2 and an obsession with water. I went from having a fear of the sea to being so fascinated that I ended up diving with bull sharks just 30 centimeters from my face in Mexico. But that's another story... As I was saying, everything began with my dream of working on major VFX productions and, combined with my obsession for Avatar 2, the idea for this project emerged. Originally it was going to be called The Heart of the Mountain, a short film about a boy who traveled to a fantasy place through a passageway under his bed. I had a storyboard made, but it was too ambitious for my knowledge and resources. When Avatar 2 came out, I decided to adapt it to the sea: a single 10-second shot, a sci-fi ocean full of alien life, and a large pulsing crystal at the center. This made the project realistic and achievable.
Understanding water to be able to create it
I didn't know how to achieve a convincing underwater look. Initially I was going to use Maya and Arnold, my usual tools, but the quality and complexity of Avatar 2 led me to change programs. I made a first version in Maya (a shell with a treasure inside) using references, but my computer couldn't handle all the geometry I wanted to include. I tried USD in Maya, but I hadn't yet discovered the full potential of real Solaris and USD. Finally, I found Isotropix Clarisse, which allows handling billions of polygons thanks to its architecture and references, which changed the game. With prior research and this new software, the real path to realize the project began. My idea was to create the underwater effect entirely in Nuke, like in Avatar 2. The lighting would have to be realistic, without added atmosphere, just taking care of caustics and hard shadows.
Building a living ocean
I did the blocking and camera animation in Maya, then transformed those arcs into realistic rocky geometry in Houdini using VDB and noises. After that I combined everything in Clarisse and added Megascans rocks manually. For textures, I did advanced shading to combine multiple complex textures, with displacement and a sand shader projected on Y to simulate sediment on the rocks. These shaders provide great realism and variety.
The asset list was extensive: corals, anemones, fish, and alien flora, seeking the diversity and complexity of Pandora's oceans. I bought some assets, modeled others, and some were free, but all went through cleanup in Houdini, UV correction, and lookdev. For lookdev, I built a scene in Clarisse that simulated realistic underwater lighting, achieving an optimal result. Once lookdev was done, I assembled the scene using Clarisse's scatter tools. Some anemones needed to be animated: I used 3ds Max with a noise modifier on the geometry to create four variants and distribute them randomly. Others I animated directly in Houdini. Most assets were placed strategically by hand.
Believable movement for a living ecosystem
The fish were animated individually in Houdini with a bend and POP Flock particle simulation following curves, for realistic school behavior. I created three variants to avoid repetition. Some anemones were animated in 3ds Max with noise on the geometry, others directly in Houdini depending on the need for realism or artistic control.
The pulsing heart
The crystal required internal volumes and complex ramps that Clarisse couldn't reproduce at the desired level. I generated the geometry and internal volumes in Houdini (VDB + noises), then exported to Maya for lookdev with Arnold, achieving realistic refraction and volumetric response. I rendered the crystal in Maya on a separate layer and in Clarisse I made a holdout of the environment to integrate it perfectly in composition.
Details that define underwater realism
I made the godrays in Maya with Arnold: light with caustic gobo, volume and render with holdouts to optimize. The suspended particles were simulated in Houdini and rendered in Maya as an independent layer. Both layers were essential to achieve a sense of depth and water density.
3 billion faces
The final scene exceeded 3 billion polygons, but Clarisse allowed viable performance. Each frame of the ENV layer took only 25 minutes, allowing completion of all material in five days of distributed rendering.
Where everything becomes real
During production I researched and tested several versions to guide lighting and final look decisions. For a real underwater look, two aspects were essential: color loss and diffusion according to distance. Light underwater travels at a different speed than in air, and water absorbs colors according to depth and distance. Red disappears at about 5 m, yellow at 15 m, green at 25 m, and blue last. For example, an object at 1 m depth but 4 m lateral distance loses red (1 + 4 = 5 m). Applying this, I dispersed color according to the depth channel of the renders and added gradual saturation loss, achieving the effect of colors being bathed in blue with distance.
For diffusion, I did a cumulative blur controlled by depth, combining the original image with the blurred one using keymix to avoid artificial blurring.
For the aquatic atmosphere, I created a DMP using constants, noises, and masks, masked by depth, achieving volume and directionality. I integrated the crystal from Maya, added godrays, dust, lens FX, crystal pulse animation, and a final color adjustment. Finally I added SFX in Premiere.
A project of technique and character
The Sea's Heart taught me that with dedication, ambitious projects can be achieved. It forced me to master multiple tools and make practical decisions: Maya, Houdini, 3ds Max, Clarisse, and Nuke played a key role. What mattered wasn't the software, but how to use it to solve real problems. Pre-production and R&D were essential: everything was designed to replicate reality and achieve a believable result. Special credits to Steffen Hampel and Split The Diff for their videos, which helped me better understand the physics behind the underwater look.
