About Magnopus and the Coco VR Project:
Founded in 2013, Magnopus are an Academy Award winning team, made up of talented artists, producers and engineers, from the film, game, and software development industries. Magnopus have created experiences for some of the biggest franchises and multinational companies in the world, including the likes of Disney and Warner Brothers.
Magnopus prides itself on being an Experience Company utilising art, science, and technology to create new experiences across virtual and augmented reality. Magnopus work on both original and licensed IP for VR experiences that are game engine-driven, footage driven or a mixture of both. They have developed projects for clients on Oculus Rift, HTC Vive, Samsung Gear VR, Magic Leap and Hololens with the use of the Unity and Unreal game engines.
Disney contracted Magnopus to create Coco VR, Disney Pixar’s first VR game, it was released November 15th just prior to the film’s worldwide release on November 22nd. Coco VR won significant critical plaudits, being nominated for an Emmy at the 70th Primetime Emmy Awards and more recently, winning VR Marketing of the Year at the second annual VR Awards.
Coco was a global box-office success, a cultural juggernaut, and a powerfully important piece of storytelling within an era of particular political turbulence in the USA. The best-selling movie ever in Mexico, and a global phenomenon, Coco won the Academy Award for Best Animated Feature in 2018, as well as swept the board at the 2018 Annies. Coco VR was designed to grant audiences a glimpse inside the land of the dead – literally inside it.
The Goals of Coco VR:
Our goal was to introduce users to an expanded world of the dead that touched on the themes of the film, without giving anything away. As a marketing piece, CocoVR was released before the film, so we didn’t want to inadvertently spoil anything. Among the great many challenges on the road to release, the clearest of all was a commitment to film quality; creating a VR experience worthy of Disney•Pixar’s pedigree is no small task. In a practical sense, we have the challenge of translating something that takes hundreds of hours to render, and packing it into 22 milliseconds in VR.
The first order of business was to create a pipeline allowing us to use assets straight from the film. The models are lovingly crafted, and very dense by real-time standards. Once we began the optimization process, we didn’t stop until the project was released. In fact, I think one of the last pushes to source control was the reduction of a few more verts on some of the plaza buildings behind the selfie-station.
The Challenges of Coco VR:
In terms of challenges, reducing the polycount was only the first. We also had to port the textures from Pixar’s pipeline into ours, however, none of the Pixar assets had UVs. This ended up being a blessing in disguise because it forced us to come up with a unique solution that could only work on a project with an established CG world that exists in another medium. Our solution involved using a series of 360 renders out of Pixar’s version of the world – which contained all of the lighting and texture information – and projecting that onto our reduced geometry using a custom shader. The shader would then determine, based on view angle, distance, and amount of stretching, which 360 projection to render on any given triangle.
Because we were getting the 360 renders directly out of renderman, we were able to adjust many of the values within the image to match what we were seeing in the HMD via Unity to what you’re seeing on the movie screen. These renders would take hundreds of hours in some cases, they are the highest quality texture assets you could ever ask for. In a very real sense, the renders from the film are the same as what you’re getting in the headset.
Making Coco VR a Social Experience:
Creating a social experience was one of our primary goals. The logic for this was something we set up very early in the project and allowed for some unique development experiences. We would do ‘location scouting’ in VR with the Pixar team, who was up in Emeryville, and Magnopus, who was in Los Angeles. We would all log in, sometimes six or seven of us, and just walk around the land of the dead to make gameplay decisions, design flow, or move entire buildings from one block to another. We even created a camera that would take pictures in world and store them locally on your hard drive with the coordinates of where the photo was taken. We essentially achieved teleconferencing in virtual reality, but imagine everyone’s a skeleton and we all have mustaches.
Something we didn’t do early enough, however, was to gate access to matchmaking based on the development version of our app. So we at Magnopus would be playtesting some new feature when suddenly unfamiliar voices start to pop up – it’s the Pixar team giving a demo to an executive. On one occasion that executive happened to be Ed Catmull, so we had to make sure everyone was presentable. When you’re working in a fantasy world, anyone can show up at any time – even Ed Catmull.
The Critical Response to Coco VR:
CocoVR was very well received. The project was nominated for an Emmy, won the VR Marketing of the Year Category at the VR Awards, as well as a Grand Clio, and a Cannes Silver Lion. Our audience seemed to enjoy the opportunity to appear on stage in a fantasy world, where they are themselves – and yet, not. I think this broke down some barriers, and allowed people to come out of their shells a bit.
Advice to Other Developers:
If I were to give some advice to others making a similar project, I would say, establish your locomotion constraints early. A significant portion of the design challenge for VR is ‘where can I go’ and ‘how do I get there.’ Making sure your locomotion mechanic is solid, reliable, intuitive, and satisfying is a difficult task, but establishing it early will inform nearly every other aspect of design – including how close you get to walls, how far you need to be to grab objects or interact with props, how many colliders you need to make, how many vantage points there will be in a given scene.
