Computer Graphics World

July-Aug-Sept 2022

Issue link:

Contents of this Issue


Page 43 of 47

42 cgw j u ly • a u g u s t • s e p t e m b e r 2 0 2 2 O ur understanding of history mostly comes from artifacts, but what if we could see the expressions on the face of an ancient human? What if we could actually see them move their facial muscles to reflect surprise, anger, or fear—or even to smile at you? Those were the questions posed by Sofija Stefanović, Profes- sor of Physical Anthropology in the University of Belgrade's Depart- ment of Archaeology, and her team when they set about creating a trailblazing reconstruction of the face of a 10,000-year-old shaman from the famous Lepenski Vir archaeological site in Serbia. Conveniently also located in Serbia is 3Lateral—a leading devel- oper of digital human technology and now part of Epic Games—who partnered with Stefanović on the project. The team at 3Lateral has been working alongside colleagues from Cubic Motion—another member of the Epic Games family—and the Unreal Engine team to develop the MetaHuman framework, which opened up early access to its free, cloud-based app MetaHuman Creator last year. Meta- Human Creator enables anyone to create unique high-fidelity digital humans, derived from its database of scanned data, in minutes. The latest MetaHuman release, however, not only brought new features to MetaHuman Creator but also launched a new Meta- Human Plugin for Unreal Engine. The first major feature of this new plugin is Mesh to MetaHuman, sharing the core technology that was used to reanimate the shaman. This exciting development, now freely accessible to everyone, enables users to create MetaHu- mans from their own custom models and complete similar projects without expert knowledge. Physical facial reconstruction using traditional techniques The Lepenski Vir settlement in Eastern Serbia, located on the banks of the Danube River, was first discovered in the 1960s. It is a site of huge historical significance as it showcases the early stages of the development of European prehistoric culture. Around 500 skeletal remains were unearthed at the site, including the skeleton that has come to be known as the 'shaman' thanks to the unusual cross-legged lotus position that it was found sitting in. The shaman is thought to have lived at around 8,000 BC. From the skeleton, a team of experts was able to determine his height, weight, and even the fact that he lived on a largely seafood-based diet. To begin the reconstruction, the team needed to create an exact physical replica of the prehistoric skull in order to preserve the original's integrity. Under the guidance of archaeologist Jugoslav Pendić from the BioSense Institute in Novi Sad, Serbia, the team began recreating the shaman's prehistoric facial features by captur- ing hundreds of 2D images using a full-frame camera. These were imported to RealityCapture to form a 3D virtual model, which was then 3D printed to produce a physical model. The replica skull was then passed to Oscar Nilsson, a forensic art- ist and archaeologist from Sweden who is an expert in reconstruct- ing models of ancient faces for museums around the world. Nilsson was able to begin creating a forensic facial reconstruction, adding muscle and skin layers with clay, with the thickness determined by the gender, age, ethnicity, and estimated weight of the subject. Turning a static mesh into an animatable model To bring the actual shaman model to life, the clay reconstruction— still without skin textures or hair—was scanned using a Peel 3D scanner and reconstructed with RealityCapture to create a digital model. With a basic texture for skin and eyes applied, a prerequisite for the next step, it was time to put the mesh through the Mesh to MetaHuman process. Using automated landmark tracking in Unreal Engine 5, Mesh to MetaHuman fit the MetaHuman topology template to the shaman scan. This new mesh was then submitted to the cloud, where it was matched to a MetaHuman with similar facial geometry and propor- tions to the shaman and automatically bound to the MetaHuman facial rig. The technology used MetaHuman Creator's extensive database of scans of real human expressions to produce 'statisti- cally estimated expressions' for the ancient face while preserving the deltas to retain the original likeness. Once the process was complete, the character could be opened in MetaHuman Creator. Immediately, the team could press the play WINDOW TO THE PAST METAHUMAN TECHNOLOGY BRINGS A 10,000–YEAR–OLD SHAMAN BACK TO LIFE

Articles in this issue

Archives of this issue

view archives of Computer Graphics World - July-Aug-Sept 2022