Courtesy of Half M.T Studios

Award-winning short film The Voice in the Hollow takes real-time animation to dark places

Animation can take a village, but you don’t need hundreds of people to make something great.

Just look at The Voice in the Hollow. From the titles on, you can tell you are looking at something unique. What you might not know is that this isn’t the work of a major studio. It’s a passion project created by a small team of professionals who are using real-time tools to fast-track projects, creating the type of intriguing visuals that would stun viewers even without a great story guiding them. Thankfully, they have that, too.
 

What’s old is new

Director and VFX supervisor Miguel Ortega and writer and production designer Tran Ma came up with the idea for The Voice in the Hollow while filming scenes on their previous project, The Ningyo, deep inside a cavern in Northern California known as “the moaning cave.” “When we were down there shooting, they actually said, don’t shoot over there because that’s where all the skeletons are,” says Miguel.

Skeletons?

The story goes that there was a moaning sound emitting from a hole in the cave that sounded like a girl calling for help. “From a scientific point of view, all it really was like was essentially blowing into a bottle—a 250-foot drop being the bottle—and it gave out the sound of a little girl,” he says.

Would-be amateur rescuers assumed the cave wasn’t as deep as it was and would sometimes fall victim to the perilous drop. When the cave was excavated in the early 1900s, a pile of skeletons was discovered presumably from the many people that fell to their deaths. At the bottom of that pile was a ten-thousand-year-old skeleton of a young girl.

This became Miguel and Tran’s inspiration for The Voice in the Hollow, a story that taps into the essence of the Cain and Abel story, but places it in Africa with two sisters as its leads.
 

Inspired by crime dramas, gritty Spaghetti Westerns, and grindhouse horror movies from the ‘60s and ‘70s, Miguel and Tran used a saturated color palette not unlike what you’d see in violent Italian cinema of that time. Sergio Corbucci’s The Great Silence, in particular, was a big inspiration for Miguel because the ending was so bleak.

Italian cinematographer Mario Bava’s color choices can be felt in the vivid reds and bright teals lighting up what would otherwise be a fairly dull yellow and brown world. These reds also help differentiate the two sisters, Coa and Ala, as one wears a striking red face paint, while the other wears white.
Courtesy of Half M.T Studios

Directing characters in real time

With no perfect precedence for what they were creating, the team’s references were starting to lead them in some novel directions. Which was great, because they wanted to stand out.

“What we didn’t want was this to feel like The Lion King or something like that,” says Miguel. “We wanted it to feel a little more fantastical.”
Courtesy of Half M.T Studios
And Miguel and Tran know what fantastical feels like, as both have prior experience working on VFX blockbusters like Superman Returns, The Mist, Speed Racer, and Transformers. They also aren’t strangers to small projects, drumming up a lot of buzz over their cryptid-inspired short, The Ningyo.

However, once they started working on The Voice in the Hollow, the constraints became about bandwidth. They knew where they wanted to go with the characters and look, but how do you feasibly animate an ambitious film with such a limited crew?

The answer motivated a lot of their creative choices. For instance, it’s no accident that the film’s characters resemble carved wooden dolls. “By making them look like wood, no one is going to say that the refraction of the iris is not correct,” says Miguel. But even then, getting to a design they were comfortable with required multiple iterations. And while Tran has said she isn’t always comfortable showing her in-process sculpts, it’s engaging to see her open up about her process on the many Gnomon-driven BTS segments that followed the production. She clearly knows what she’s doing.
Courtesy of Half M.T Studios
In practice, the character models were sculpted in ZBrush and textured in Substance 3D Painter. Tran notes that typically with photoreal models, you want to hide your paint strokes. But since these characters are doll-like, she was deliberate about showing the brush strokes on their bodies.
Miguel and Tran also wanted Coa and Ala’s outfits to naturally suggest who was the better hunter between the two. “If you look closely at Coa’s clothing, she doesn’t have any animal skins because she hasn’t been very successful [at hunting],” says Tran. “Whereas, Ala’s design is asymmetrical, it’s supposed to be more freeing, and you can see that everything she has is the skin of something she hunted down.”
Courtesy of Half M.T Studios
Once the characters were refined, they were animated in Unreal Engine, a tool with which the team only had a few weeks of experience. Yet, even with that limited experience, Miguel and Tran knew they could use features like UE5’s shallow depth of field focus to emphasize important visual cues, like the miniature appearance of Coa and Ala. Not only would this help capture a distinct aesthetic for the characters, but it gave the project a stop-motion look that many audiences like.

Crafting an African world

The ominous terrain around the hollow was created in Gaea, which helped the team blend manual procedural elements and Megascans assets. Special attention was also placed on the crowning spiked pillars of the pit, which symbolize Lucifer and help lead Coa toward her dark destiny.
Courtesy of Half M.T Studios
While the village doesn’t exactly take center stage in The Voice in the Hollow, Miguel and Tran still wanted to create a complete environment to increase shot flexibility. To save time, they built it up using a blend of ready-made Megascans assets and customized Mudbox sculpts, tapping the right tool as they went. Miguel likened the process to LEGO pieces. “It shows you that just because you’re using Megascans assets, it doesn’t have to look like Megascans assets,” says Miguel.

What may not be noticeable on the first watch is just how detailed the village and other environments are. Even with pre-made assets, much of the film includes custom-modeled statues and structures that are often only barely perceived in the film, but remain essential to its overall feel. “If you’ve taken my classes, that’s something you’ll always hear me say, ‘It doesn’t matter if you see it, you have to feel it,’ ” notes Miguel. And in The Voice in the Hollow, you can sense the complexity.
 
Courtesy of Half M.T Studios

Balancing the cultural aspects of the story also required Miguel and Tran to think broadly about their presentations. Rather than drawing from a single tribe or source, research was conducted into designs from all over Africa. This helped the team select a variety of clothing, hair, and body paint choices that wouldn’t be boxed in by a specific aesthetic but would still remain tied to the region.

Miguel and Tran also didn’t want to mimic a singular group of people in Africa. Instead, they designed the characters so that they could easily create variants of them in Unreal Engine using masks. “By creating masks, you could go in, change the clothing color and change the paint color on any of them,” says Miguel. Tran adds, “It was a very flexible system in order to get variation. It’s a really nice material system that you can build in Unreal.”
Courtesy of Half M.T Studios
Once Miguel and Tran nailed the aesthetic, the next challenge was pushing through the grueling schedule they gave themselves. They’d work daily as a small team of professional artists, only taking breaks to teach their classes. Tran admits that if this project was done in a traditional pipeline, it probably wouldn’t have been finished, crediting how productive Unreal Engine’s real-time workflows can be.

Adding vocal and kinetic depth

To further immerse viewers in their African-inspired world, Miguel and Tran decided to alter their script from English to Swahili, a language commonly spoken in East Africa. Casting it was easier said than done though, a fact Miguel and Tran learned the hard way when they began searching for Swahili-fluent actors in Los Angeles. “We got hundreds of submissions and none of them spoke Swahili,” says Miguel.

With LA out, you’d think they’d find their solution in an African talent agency. Only they couldn’t find one. Seeking a solution, they turned to audiobook translators for the voiceovers. It was a huge success. Even with no prior acting experience, all three actors amazed Miguel and Tran with their performances. And now, Miguel and Tran can confidently say they’ve not only made one of the only animated African horror stories told in Swahili, but they did it with people from Africa.

There was still a problem, though. The actors were on another continent, which means they would have had to travel thousands of miles to attend a shoot. So Miguel took the mocap process on himself, recording the facial capture with a head-mounted iPhone and MocapX, which was later touched up in Maya.

Ala and Coa’s body capture data, on the other hand, was captured on a professional green-screen stage at their producer Gnomon's campus using Xsens' MVN Awinda suit and gloves. Initially, Miguel recorded himself to create the temporary animations for the characters, but for the final animations, they enlisted the skills of Kaitlyn O'Connell, who had previously collaborated with them on The Ningyo. This ensured more accurate and authentic movements, particularly in terms of capturing the essence of a young girl.
 
Courtesy of Half M.T Studios

Illuminating the journey

Lighting and rendering were done within Unreal Engine 5, utilizing light blockers to manipulate the positioning of shadows in the environment. To achieve greater control over the final result, Miguel and Tran embraced a different approach.

They opted to render and composite raw lighting effects and passes directly in real time within the engine, diverging from a previous workflow of generating render passes in an offline renderer and compositor. The benefit? With everything in one place, freedom and flexibility came easily. Lumen proved particularly advantageous, as its dynamic global illumination lighting capabilities closely resembled their former offline renderer, and further enhanced the quality of their work.
 
Courtesy of Half M.T Studios

Real-time takeaways

Now that the project has wrapped, Tran noted how difficult it would be to go back to their old ways after exploring real-time animation. “It just saves us so much time, especially with the storytelling process,” she says. “The response time, the quality, just seeing everything instantly…it’s so much faster.”
Courtesy of Half M.T Studios
Just take the title shot—with its majestic mountains—alone. In the past, that would’ve taken them weeks to create. But in UE5, it only took a few days. And for a team like Miguel and Tran, who are constantly thinking about how to do more with less, that immediacy is motivating.

“With other tools, it can be time-consuming to get to that Eureka moment. But with Unreal, it’s so much easier to get there,” says Miguel. “And that’s going to keep us coming back.”

    Start animating in Unreal Engine

    Explore the Animation Hub for informative case studies, resources, and other online learning materials that will get your project off on the right foot!