This post was written by Henry Winchester.
We wanted to do something different. Our most recent short film, The Ningyo, had just died in development hell. We had put years of blood, sweat, and tears into that film, and it all went up in flames.
Needless to say, we were heartbroken and disenchanted with the whole industry. But, after moping around for a few months, I found comfort in the cinema of the '60s and '70s, captivated by its raw and imperfect qualities. The use of 16mm film and bleak endings imbued the films with a sense of punk-rock rawness, making them feel real and unrefined.
I wanted to make an homage to this era, but do it in animation, and from that came The Voice in the Hollow, an African-inspired horror film in Swahili.
The movie drew inspiration from classic Westerns, with Sergio Corbucci's The Great Silence serving as a major influence for its bleak ending. We also drew from The Good, The Bad, and The Ugly for its color scheme and tense stand-off. Other sources of inspiration included Tales from the Crypt, Texas Chainsaw Massacre, Mario Bava's Black Sabbath, Dario Argento, and Apocalypto.
We typically worked with V-Ray, an offline renderer, but taking part in the Unreal Fellowship — an intensive, month-long course — opened our eyes to the possibilities of real-time, which uses game engine technology to deliver great visuals almost instantaneously.
My partner Tran and I both teach at Gnomon School of Visual Effects, so we presented the idea to Founder and President Alex Alvarez. He loved it and agreed to fund the film, but there was a caveat: we had to document the process and put it online. Which was terrifying for us! When I'm working for companies, I'm the kind of guy who sits with his back to the wall and only lets people see his work once it's complete.
So as well as trying something completely new in an engine we had only used for a few weeks, we had the added pressure of showing our work every week. We thought we would fail on an epic scale, and it would be online forever.
'The Voice in the Hollow'Credit: Half M.T Studios
In some cases, there were shots, entire sequences, and camera angles and edits, that we created live on the stream, and they're in the movie to this day. But on other days, it was two hours of just total disaster that we didn't use in the final film. But I think that's kind of cool too. The creative process is NOT linear; it's two steps forward, three steps back.
One of the humorous advantages of recording the stream was that we would figure out how to do something, and then two months later, we completely forgot how to do it. But we could go back and watch our videos to figure out how to do it again!
When we completed the project, it totaled 90 hours of making-of videos. In hindsight, I am very glad we documented it; hopefully, it will help someone out there.
'The Voice in the Hollow'Credit: Half M.T Studios
The Story
The Voice in the Hollow was inspired by a mysterious cavern in northern California where we filmed some scenes for The Ningyo. The cavern was notorious for its ghostly moans, resembling the cries of a young girl, that echoed from within. As people approached to help the "girl," they would accidentally slip in, falling 250 feet to their death.
In the 1900s, the cave was excavated, revealing a pile of skeletons. And what was at the bottom of the pile? A 10,000-year-old skeleton of a teenage girl. How cool is that?!
Credit: Half M.T Studios
'The Voice in the Hollow'Credit: Half M.T Studios
Production Design
For our previous film, The Ningyo, set in Victorian times, we had a wealth of cinema references to draw from, but we didn't have that here. I knew I didn't want this to feel like The Lion King but more like a 1970s fantasy art piece, more in line with Frank Frazzetta or Masters of the Universe than Disney (I love them all, for the record).
With the core team being Tran and me, we had to find a style that fitted within our technical and time constraints. We wanted to avoid "wrestling" with the technology. The solution was to create a wooden doll-like look and feel. To achieve this, Tran omitted subsurface scattering (which gives skin a realistic translucency) and eye refractions, using technology limitations to drive the aesthetic. The aim was not to create a technology-driven showpiece but instead mimic the charms of low-tech, stop-motion animation.
We wanted the costumes to reflect the sisters' abilities and experiences. The costume for Ala was designed to show her hunting prowess, with pelts and leather from her previous kills. On the other hand, Coa, as the less experienced sister, has none.
The father had minimal lines, so we had to deliver a lot of his character in his appearance. We wanted him to be likable but aloof, unaware of the sisters’ conflict until it was too late. I love how he looked, and it's a shame he had to wear a mask the majority of the time.
'The Voice in the Hollow'Credit: Half M.T Studios
Environmental Design
We used Quixel Megascans assets, which are free for Unreal users, to build the village, and we structured it in a very modular way so we could make variations of tents or stairways. The Morning Star tree represents the devil, so we went for the horn design. We created the hollow in Quadspinner's Gaea terrain design tool. I wanted it to look cinematic and not just like a hole in the ground, so I looked to the Sarlacc Pit in Return of the Jedi for inspiration.
'The Voice in the Hollow'Credit: Half M.T Studios
Camera Optics
A huge part of the film's look is its grain and camera work. A big inspiration for the optical feel of the film was The Texas Chainsaw Massacre. To achieve this, we scanned 16mm film and overlayed it onto the film; we added gate weave and halation with the Dehancer plugin.
I made it a point to always move the camera the way a camera operator would on a lower-budget 1970s film. He had a dolly, a small crane – and a lousy first AC. The intentional "lost focus" moments or film scratches were used for dramatic effect; the scratches ONLY appear when Coa is mad or scared. I imported shots of crash zooms from films I admired into Premiere Pro and got the frame length and camera wobble to match exactly the same. The lenses were all 50mm+ with a tendency to go longer for a more shallow depth of field.
'The Voice in the Hollow'Credit: Half M.T Studios
Cast
We initially intended to record voice performances in English, but it just didn't feel authentic to me. So, we switched to Swahili, mistakenly thinking it would be easier due to its widespread use in Africa. Despite hundreds of submissions from African actors in LA, none spoke Swahili fluently, leading us to reach out to African talent agencies with no response. As a last resort, I reached out to audiobook translators and was lucky to find Janeth, Rosalie, and Goodluck Gabriel, who proved fantastic and brought the authenticity we sought.
'The Voice in the Hollow'Credit: Half M.T Studios
Motion Capture
This was our first time doing mocap; we used Xsens' Awinda suit. Unfortunately, we didn't have the budget to go to Africa to mocap our actresses, or to bring them over here, so I mo-capped myself playing Coa and Ala. But it was so obvious that it was a 40-year-old dude puppeting the girls. I even tried to move my hips more, which just made it more terrifying.
So we started working with actresses Kaitlyn O'Connell (who starred in our short "The Green Ruby Pumpkin"), and it was one of the best things we ever did. She brought an unbelievable amount of life to the characters. There are certain poses she hit that I can't believe were not hand-animated.
Giallo films inspired some of 'The Voice in the Hollow’s' lighting
'The Voice in the Hollow'Credit: Half M.T Studios
Facial Capture
For the facial capture, we used the iPhone app MocapX. Our good friend Chris Bostjanick sculpted the blend shapes; these 52 shapes were the basis for all the characters' facial expressions.
Then I had to memorize the dialog in Swahili, get the timing exactly right, and record myself doing it to drive the facial animations for all the main characters. As the body mocap and facial mocap were coming from two different programs, we had to blink and clap to synchronize movements; it became like our clapperboard from animation.
A lot of the faces were redone entirely by hand to improve the timing, eyelines, and such.
'The Voice in the Hollow'Credit: Half M.T Studios
Hand and 2D Animation
Mocap proved to be useful for many shots, but from the midpoint of the film onwards, when there's more action, we used a lot of hand animation for the characters, done by a few animators we had from around the world. Some in Thailand, Europe, and some from Gnomon.
Initially, we were concerned that mixing various animation styles would clash, but we were pleased to find that the film's visual style was very forgiving. For example, when the vines open up to reveal the Leopard mask, the shot lacks motion blur due to technical issues in generating motion vectors from SpeedTree, but the result gives the film a unique touch. It looked odd in a good way, like an old Polish stop-motion film or something out of Sesame Street in the '80s.
So we just kept it as is, and it was fine. If this was one of our live-action projects, we would have felt like we had to get the motion blur to work because it's photo-real. I love that freedom to put 2D animation in or not add motion blur here, and it's all fine.
The trance sequence was masterfully created by Babak Nekooei and his talented 2D animation team, using traditional Disney-style ink and paint techniques.
The fire in the hollow was created through hand-drawn frames that were then composited in Nuke. The goal was to create a unique and tactile feel for the sequence, and what could be better for that than hand-drawn animation at 24 frames per second?
'The Voice in the Hollow'Credit: Half M.T Studios
Thoughts on Unreal Engine
I think the coolest thing about Unreal is that sometimes the camera would snap to a random location that I had never looked at the scene from. I'd think, wow, what a great angle; let's go with this.
These happy accidents felt like something I would do on a live-action set. On our previous short, I walked in on actress Tamlyn Tomita rehearsing on the floor, and the beauty of the moment inspired me to adjust the blocking to match her actions. These serendipitous discoveries are what make the filmmaking process so exciting.
From an editing point of view, real-time rendering allows me to set up multiple camera coverage angles and render the entire sequence efficiently. In a traditional rendering pipeline, that could take weeks or months, so you only render what you absolutely need. But again, by using Unreal Engine, and having all the angles rendered, I could find those happy accidents in the edit.
Unreal Engine also allows us to do things that would be impossible in terms of filmmaking. For example, you can change the F stop to get a more shallow depth of field without it affecting the brightness of the shot. But you can always revert to a physical-type camera.
I'm happy with what we were able to do. And it just shows you the power of the program that if you come in with experience in VFX you can pick it up in four weeks. We had some minor issues, but that's just part of the process for all VFX and animation.
What's Next?
The Voice in the Hollow has once again attracted a lot of attention in Hollywood, and it's definitely opening up some huge opportunities for us. I can't get into details YET, but I will say we could not be happier with the outcome. So be on the lookout for that soon.
Thanks to everyone who watched and shared the film, we really appreciate it.
Special thanks to Alex Alvarez and Gnomon for helping to make this possible. And, of course, Unreal Engine.
This post was written by Henry Winchester.
