AIquarium
A New World of AI Alien Sea Creatures from POP Creative Studio
Somewhere between the deep sea and deep space lives AIquarium, POP Creative Studio’s surreal underwater world filled with alien sea creatures that feel strangely alive. Glowing textures, translucent forms and slow drifting movement give the project a cinematic quality that feels immersive and atmospheric. What makes the work so compelling is the sense of curiosity running through every frame. AIquarium feels less like a technical experiment and more like an exploration of world building and visual storytelling.
We spoke with POP Creative Studio about the inspiration behind AIquarium, shaping an entirely new underwater ecosystem and creating creatures that feel unsettling, beautiful and believable all at once.
What was the original idea behind AIquarium?
AIquarium is an internal R&D project exploring how AI could support a more flexible approach to creature design and world building. We were interested in whether AI could be used more like a simulated evolutionary system, where the environment itself shaped the outcome. The idea was to explore speculative biology and create alien sea life that felt as though it had genuinely evolved under unfamiliar conditions. It also became a way for us to test new hybrid AI and CGI workflows, particularly around realism, consistency, refinement, and moving concepts closer to production-ready assets.
Instead of designing the creatures directly, you focused on the environments first. Why?
That became one of the most interesting parts of the project. Instead of describing what we thought each creature should look like, we defined the conditions it needed to survive in, things like water salinity, temperature, pH levels, predators, camouflage requirements, diet, or how much light reached its habitat. We then allowed AI to interpret how and why a creature might evolve to survive in those conditions. The creatures evolved differently in each response, often developing their own internal logic and biological reasoning. It allowed us to tap into AI’s learned understanding of science and nature, creating lifeforms that felt unfamiliar, but still believable.
What did this workflow allow you to explore that a more traditional pipeline might not?
Speed and breadth of exploration were huge advantages. In a traditional pipeline, broad experimentation can become expensive very quickly, especially once modelling, texturing, rigging, and look development begin. That often means concepts stay relatively close to familiar forms early on. AI allowed us to explore ideas much closer to a final visual experience far earlier in the process, including texture, atmosphere, environment, before committing more production resources. This let us test hundreds of directions quickly, exploring unusual silhouettes, anatomy, textures, and behaviours while still maintaining a sense of biological plausibility.
Consistency is often one of the biggest challenges with AI-generated imagery. Was that something you were trying to solve?
Definitely. One of the biggest goals was moving beyond a single successful image. We wanted creatures to feel like a consistent species that could exist across multiple views, lighting conditions, poses, and environments while still maintaining a recognisable identity. Through iterative refinement and controlled workflows, and custom-trained AI models, each creature gradually developed a more stable visual identity. Once a species became consistent enough, we could effectively “clone” it across a tank, creating populations that felt recognisable, but subtly unique, as in the natural world. That shift from isolated outputs to repeatable, developable assets became a really important part of the project.
How did CGI fit into the process?
CGI became an important bridge between exploration and control. Once stronger creature designs emerged, selected specimens were translated into 3D. This helped validate their overall structure, proportions, and anatomy from multiple angles, while also allowing us to refine poses, camera perspectives, and integration within the aquarium environments. Importantly, the process wasn’t a one-way transition from AI into CGI. Once posed and refined in 3D, the creatures could then be reintroduced back into our AI workflows, where textures, lighting, atmosphere, and motion could continue evolving while maintaining a stronger sense of consistency and control.
What did the project ultimately become?
AIquarium evolved into much more than a creature experiment. It became a proof of concept for a broader AI-supported creature pipeline, connecting character design, concept exploration, realism refinement, consistency solving, 3D translation, environment building, and motion development into one connected workflow. More importantly, it allowed us to develop new tools, workflows, and production experience that can now be applied to future live projects with far greater confidence, control, and creative flexibility.