Epic Games launches MetaHuman 5.8 to create real-time game character crowds
Epic Games announced at its Unreal Fest event in Chicago today that MetaHuman 5.8 is now available for developers. MetaHuman 5.8 delivers substantial advances that mean devs can scale their real-time MetaHuman populations into the thousands, as well as extending proven facial workflows to the entire body. Epic is also making core MetaHuman libraries available as open source under MIT for the first time. MetaHuman debuted six years ago with Unreal Engine 5’s debut.What’s new in MetaHuman 5.8: Highlights Populate your world with crowds of MetaHumansMetaHuman Collections is a new asset type that enables you to populate your real-time world with crowds of MetaHumans, scaling up to hundreds of characters on mobile and thousands onhigher-end platforms. Introduced as Experimental in this release, the feature enables you to combine highly realistic close-ups with optimized performance and memory footprint by seamlessly transitioning characters between high-fidelity individual Actors and lower-fidelity Instanced Skinned Meshes(ISKMs) based on camera proximity. You can control the look of the crowd by building collections of modular character components(head, body, hair, clothing) and composing them by hand or procedurally using Blueprints. The new crowd creation workflow also relies upon Mass for crowd orchestration, and can berendered using Nanite virtualized geometry where available. Animation is achieved using Animation Blueprints or simple sequences, depending on distance. New scalable normal maps—developed to support this workflow but useful in any scenariowhere multiple characters appear on screen simultaneously and texture memory is a concern—can optionally be used for faces, using a magnitude less memory while maintaining a level of visual quality that is almost indistinguishable from cinematic at close range. You can opt in to using this feature during the assembly process. To get started, download and explore the new MetaHuman Crowds Sample, which demonstrates a performant and scalable approach to populating your scene with thousands ofMetaHumans. Turn full-body meshes into a MetaHuman character, ready to animate. Now fully integrated into MetaHuman Creator, Mesh to MetaHuman—which previously only worked for heads—is now also able to conform bodies, separately from or simultaneously with heads. It accepts input meshes of any topology and automatically generates results powered by the MetaHuman database. This powerful enhancement means you can turn any human character mesh with arbitrary topology into a fully-rigged MetaHuman with MetaHuman topology in a single workflow. It’s ideal for characters created from scans or external DCC tools, meshes generated with tools like Meshy or Tripo, or for converting existing characters. While the workflow is constrained to humanoids, it is able to handle stylized characters. You can see a great demonstration of this in the MetaHuman Vampire Asset, available now on Fab. Generate body animation from a single camera. Another feature previously restricted to faces is now also available for bodies in this release:MetaHuman Animator can now capture your full character performance: face, body, or both at once, from a single off-actor camera. The same workflow that set the standard for facial performance now drives the body too, with no need for mocap rigs, helmet cameras, or markers, and no separate pipeline to stitch together. Because processing happens offline on your machine, the capture setup stays simple: a webcam is enough to deliver high-quality results. Introduced as Experimental in this release, this new ability comes from the integration of Meshcapade’s Markerless Motion Capture, and is made available via the MetaHuman Animator Markerless Motion Capture Plugin, which is available on Fab for Windows. Get open source access to RigLogic and DNA via OpenRigLogic Developers looking to provide MetaHuman-compatible capabilities in third-party applications now have access to core technology components—the RigLogic and DNA libraries—on GitHub, under an MIT license. This opens up the potential for using MetaHumans in your platform of choice, while maintaining compatibility with MetaHuman Creator and MetaHuman Animator in Unreal Engine. The OpenRigLogic repository is the start of the MetaHuman Devkit, an evolving collection of MetaHuman character technology that developers can integrate into the platform or applicationof their choice outside of Unreal Engine.Other key new features in MetaHuman 5.8 Unbaked textures Artists wanting to match a specific look can now export unbaked texture and material assets to edit using standard Unreal Engine editors or external DCC tools, and then apply them as overrides in MetaHuman Creator—so you can still benefit from the material baking when running default assemblies to maintain performance optimization.Custom lighting setup previews Character artists, lighting artists, and lookdev teams can now import custom lighting scenes in MetaHuman
MetaHuman 5.8 delivers substantial advances that mean devs can scale their real-time MetaHuman populations into the thousands, as well as extending proven facial workflows to the entire body. Epic is also making core MetaHuman libraries available as open source under MIT for the first time.
MetaHuman debuted six years ago with Unreal Engine 5’s debut.
MetaHuman Collections is a new asset type that enables you to populate your real-time world with crowds of MetaHumans, scaling up to hundreds of characters on mobile and thousands on
higher-end platforms.
Introduced as Experimental in this release, the feature enables you to combine highly realistic
close-ups with optimized performance and memory footprint by seamlessly transitioning characters between high-fidelity individual Actors and lower-fidelity Instanced Skinned Meshes
(ISKMs) based on camera proximity.
You can control the look of the crowd by building collections of modular character components
(head, body, hair, clothing) and composing them by hand or procedurally using Blueprints.
The new crowd creation workflow also relies upon Mass for crowd orchestration, and can be
rendered using Nanite virtualized geometry where available. Animation is achieved using Animation Blueprints or simple sequences, depending on distance.
New scalable normal maps—developed to support this workflow but useful in any scenario
where multiple characters appear on screen simultaneously and texture memory is a concern—can optionally be used for faces, using a magnitude less memory while maintaining a level of visual quality that is almost indistinguishable from cinematic at close range. You can opt in to using this feature during the assembly process.
To get started, download and explore the new MetaHuman Crowds Sample, which demonstrates a performant and scalable approach to populating your scene with thousands of
MetaHumans.
Turn full-body meshes into a MetaHuman character, ready to animate.
Now fully integrated into MetaHuman Creator, Mesh to MetaHuman—which previously only worked for heads—is now also able to conform bodies, separately from or simultaneously with heads. It accepts input meshes of any topology and automatically generates results powered by the MetaHuman database.
This powerful enhancement means you can turn any human character mesh with arbitrary topology into a fully-rigged MetaHuman with MetaHuman topology in a single workflow. It’s ideal for characters created from scans or external DCC tools, meshes generated with tools like Meshy or Tripo, or for converting existing characters.
While the workflow is constrained to humanoids, it is able to handle stylized characters. You can see a great demonstration of this in the MetaHuman Vampire Asset, available now on Fab.
Generate body animation from a single camera.
Another feature previously restricted to faces is now also available for bodies in this release:
MetaHuman Animator can now capture your full character performance: face, body, or both at once, from a single off-actor camera.
The same workflow that set the standard for facial performance now drives the body too, with no need for mocap rigs, helmet cameras, or markers, and no separate pipeline to stitch together.
Because processing happens offline on your machine, the capture setup stays simple: a webcam is enough to deliver high-quality results.
Introduced as Experimental in this release, this new ability comes from the integration of Meshcapade’s Markerless Motion Capture, and is made available via the MetaHuman Animator Markerless Motion Capture Plugin, which is available on Fab for Windows.
The OpenRigLogic repository is the start of the MetaHuman Devkit, an evolving collection of MetaHuman character technology that developers can integrate into the platform or application
of their choice outside of Unreal Engine.
Artists wanting to match a specific look can now export unbaked texture and material assets
to edit using standard Unreal Engine editors or external DCC tools, and then apply them as
overrides in MetaHuman Creator—so you can still benefit from the material baking when running default assemblies to maintain performance optimization.
lighting.
Meanwhile, the real-time audio-driven model adds procedural blinks (drawn from a library of
motion-captured blinks and modeled on natural blink frequency) and automatic emotion detection from the audio when no emotion is specified.
Plus, curve output is now more animator-friendly, with cleaner activation on key poses for easier editing and refinement.
MetaHuman Animator is now available on Linux and macOS platforms, where it joins MetaHuman Creator. Real-time facial animation from Live Link Face and offline facial animation are supported across all three platforms, while support for other features varies across platforms.
These are just some of the new features and enhancements in MetaHuman 5.8. You can
check out all the updates in the Release Notes.
The post Epic Games launches MetaHuman 5.8 to create real-time game character crowds appeared first on GamesBeat.
What's Your Reaction?