Creating Surreal Short Film The Cradle with UE5, Houdini, & Blender

Introduction

My name is Germans Alekseikovs, and I am the CEO and Technical Director of Argentum Studio.

Argentum Studio is a team of artists and technical specialists focused on cinematic real-time graphics, character production, VFX, and rendering R&D. Our team members came into 3D from different directions, but most of our skills were shaped through real production work, experimentation, and problem-solving rather than by following a single traditional path.

A large part of our experience comes from constantly testing the limits of real-time rendering and building tools around production problems. We often had to study how Unreal Engine works internally, analyze rendering systems, modify shaders, and create our own workflows when the default tools were not enough.

As a studio, we have contributed to multiple cinematic and real-time projects presented in Argentum Studio’s ArtStation portfolio, including internal films, client-oriented cinematic work, character projects, and Unreal Engine rendering experiments. The Cradle became one of the most important projects for us because it combined all of these areas: art direction, character work, animation, VFX, rendering, and tool development.

Inspiration & References

The Cradle started as a large-scale R&D experiment. The main idea was to explore the limits of Unreal Engine and understand how far we could push real-time rendering toward the visual quality of offline path tracing.

From the beginning, Argentum Studio’s long-term goal was to achieve path-traced cinematic quality in real time. We started developing The Cradle during Unreal Engine 5.1, when many of the technologies we wanted to use were still new, unstable, or not fully production-ready.

Visually, the project was strongly inspired by the works of Zdzisław Beksiński. His art is famous, but far fewer CG projects try to build original cinematic worlds inspired by his visual language. Many artists reference H.R. Giger, but Beksiński remains less explored in CG filmmaking, and we wanted to experiment with that kind of atmosphere.

We were not trying to copy his paintings directly. Instead, we wanted to capture a similar emotional feeling: something surreal, uncomfortable, monumental, and dreamlike. The goal was to create a world that felt physically believable, but at the same time strange and symbolic.

Technically, we wanted to test systems that were rarely used for cinematic production inside Unreal Engine at the time: Hair Strands rendering, complex volumetrics, custom lighting workflows, VDB-style rendering, and internal rendering features that most users never touched.

Reason Behind Choosing UE5

The biggest reason was speed.

Deferred real-time rendering gives an enormous advantage compared to offline path tracing. In some cases, it can be 10x to 100x faster. A single frame with multiple groom hairstyles and volumetric effects could take days or even weeks in an offline pathtraced pipeline, while Unreal Engine allowed us to render similar shots in minutes through Movie Render Queue.

Even when our viewport performance dropped to 1–10 FPS, Unreal was still dramatically faster than traditional offline rendering. That speed gave us the ability to iterate, test lighting, adjust cameras, and solve visual problems much more quickly.

Another important reason was access. Unreal Engine exposes a huge amount of rendering infrastructure. Because the engine source code is available, we could study how internal systems worked, modify shaders, adjust rendering behavior through CVARs, and build tools around experimental features.

For us, Unreal Engine gradually stopped being just a game engine. We started treating it as a cinematic rendering platform.

Changing UE5 Tools to Fit the Project

Very early in production, we realized that the default Unreal Engine workflows were not enough for the visual quality we wanted. Instead of waiting for future engine updates, we started building our own internal tools, rendering modifications, and shader workflows.

One of the first major systems we built was a custom groom rendering tool - Hair Cinematic Tool. At the time, artists often relied on random console variables shared online to improve hair rendering quality, but this approach was inconsistent and unpredictable. We created a Blueprint-based system that centralized and controlled Hair Strands rendering behavior, effectively acting as a dedicated post-process layer for hair rendering.

Another major development was our native VDB rendering workflow. Existing plugin solutions had problems with translucency sorting and integration with Unreal’s rendering systems. Together with our technical team, we rewrote Unreal’s volumetric cloud shader pipeline and adapted it into a native VDB-style rendering system integrated directly into Unreal’s internal volumetric framework.

We also developed our own Alembic Cache Pipeline. Extremely dense cinematic heads could not be imported reliably through standard Alembic workflows, so we built a converter that transformed animation caches from Blender, Maya, or 3ds Max into Skeletal Meshes and Morph Targets compatible with Unreal’s native systems.

Those tools eventually became the foundation of Argentum Studio’s cinematic pipeline development.

Visualizing Hair & Moth’s Fuzz

Hair was one of the biggest technical challenges in The Cradle. Almost every major asset involved hair or fur.

Diana, the moth creature, and even the floating figures appearing throughout the film.

We didn’t separate the pipeline into “human hair” versus “creature fuzz” workflows. Instead, we built a unified approach around Unreal’s Hair Strands system, adapting it to different artistic needs.

At the time, many artists relied on undocumented console variables shared on forums to tweak hair rendering. We didn’t like that approach. It felt unpredictable, hard to reproduce, and often produced mismatches between viewport previews and final renders.

Because of that, we built our own Hair Cinematic Tool instead of relying on random CVAR tweaks. It gave us precise control over Hair Strands rendering and ensured consistent results between the viewport and final renders.

For human hair, the focus was on believable shading, softness, and preserving fine strand detail. For the moth’s fuzz, the challenge shifted toward achieving the right volume, light response, and microscopic softness while keeping the look stylized and readable on screen.

Although the artistic goals were different, the underlying workflow and toolset remained the same. For grooming and simulation, we used Ornatrix and Houdini, and then adapted the result for Unreal’s Hair Strands pipeline.

Rigging

The rigging process depended on the character.

For Diana, we used a pre-existing human rig based on Chris Jones’ work and the Universal Human Blender Rig. It was not created by us from scratch, but it gave us a strong foundation for the main character and allowed us to focus more on animation, facial performance, and cinematic integration.

For the additional characters, including the Cradle and the moth creature, we created rigs from scratch in Blender. These characters required a more custom approach because their movement language was not based on standard human anatomy.

The moth was especially interesting because the wings required additional physical behavior. We used Blender physics setups for the wings to help achieve more natural motion and secondary movement.

We also used our internal Alembic Cache Pipeline during production to transfer animation cache data into Unreal Engine in a more stable way. Instead of relying entirely on heavy cache playback, we converted animation data into Unreal-native formats where needed, which helped keep the shots more manageable inside the engine.

Animation

The most complicated part was Diana’s facial animation because we did not use motion capture on this project. Everything had to be animated manually, and facial animation is always difficult because even small mistakes immediately feel unnatural.

All scenes with the moth were also challenging. It is very hard to communicate the natural movement of a moth through animation, especially when the creature also needs to feel cinematic, readable, and emotionally connected to the scene.

The Cradle itself also went through many animation iterations. At some points, different people on the team interpreted the movement differently. Some felt like the character was playing a flute, while others felt that the fingers were simply moving unnaturally. Eventually, we found a direction that felt closer to a shredding motion, and that became the final performance idea.

So the hardest part was not just making the animation technically work. The real challenge was finding the right movement language for characters that were surreal, symbolic, and not fully human.

Conclusion

The production took around three years.

The main reason it took so long was that The Cradle had no dedicated budget and was made in our free time. At the same time, the project was technically very ambitious. We started during the Unreal Engine 5.1 era, when many systems were still unstable, and we had to solve a lot of problems ourselves.

Another major challenge was legacy. As a team, we were growing faster than the project itself. Our skills, tools, and understanding of Unreal Engine improved significantly during production, but the project still carried many old decisions from earlier stages. In a way, the legacy of the project was slowly killing the project itself.

That is one of the biggest lessons we learned. Ambition is important, but if a project has no budget, huge scope can easily turn into production hell.

In total, 42 people contributed to The Cradle, but the active core team was around 15 people.

If we made it today, we would do everything differently. Technology has moved forward, and our own tools and workflows have evolved dramatically. But at the same time, The Cradle was necessary. Without making mistakes, hacking systems apart, and experimenting aggressively, we would not have built the cinematic pipeline we have today.

Our main advice to beginner artists is to take small steps. Do not start with a giant dream project that requires years of work, a large team, and tools you do not yet have. Start with small projects that you can actually finish.

Small finished projects teach you more than huge unfinished ones. They help you understand the full production cycle, build confidence, and avoid getting trapped by your own ambition. The Cradle was extremely difficult to finish because our ambitions were very high, and we only fully understood how much work it required after we were already deep into production.

Argentum Studio