For those who don’t know the tool: MetaHuman Animator sits inside Unreal Engine for performance capture on MetaHuman characters. The new Fab plugin extends it with markerless body capture from video.
Why this news? Because at the release of Unreal Engine 5.8, we brushed past this, but since then, we realised that this is kind of a big deal. And we were yelled at by excited authors. (Hi Manuel!)
The webcam enters the mocap chat
Epic Games has released MetaHuman Animator Markerless Motion Capture Plugin, a free add-on for Unreal Engine 5.8. The plugin generates body animation from standard video footage and extends MetaHuman Animator inside the engine.
The pitch is simple enough to make mocap stage operators stare into the middle distance: shoot a performance on a phone, webcam, or studio camera, process it locally, and use the result as animation in the editor. The plugin targets full-body animation, including hands, and can drive the MetaHuman skeleton directly or support retargeting to other character types.
The useful part for production artists is not the absence of Lycra. It is the pipeline position. The generated take can be refined and adapted in Sequencer like other animation. The plugin uses the same assets and processing flows as the rest of the facial capture workflow, with automation available through Python and Blueprints.
One camera, two solves
The plugin supports single-camera capture for body, or body and face from the same video. Face and body solves run independently on the same footage. That matters for small teams because one camera can cover both streams without a separate stitching pipeline, and dual solves means more focussed code for each.
A dual-camera path also exists. In that setup, face and body footage come from separate cameras and combine in a level sequence, like a head-mounted camera for face capture and an off-actor reference camera for body capture. When a final game character does not yet exist, an identity stand-in MetaHuman can be used first. All MetaHumans share the same skeleton, so the animation maps to the final character without retargeting.
The body capture path works offline. The video-to-mocap pipeline processes footage locally on the workstation. The Fab listing also states that footage never leaves the workstation. That is the boring sentence supervisors like, because cloud upload policies are not usually where shots get prettier.
Experimental means experimental
This is an experimental feature introduced with Unreal Engine 5.8. Experimental features are not production-ready and may change or be removed in future releases. Translation for anyone building a show bible: do not wire the whole animation department around it before a proper test.
The requirements are clear. The plugin needs Unreal Engine 5.8 or later, the MetaHuman Animator plugin enabled in the editor, and the markerless plugin installed and enabled from Fab. It requires one or more performance takes captured with a single video camera or the Live Link Face iOS app configured to use a non-TrueDepth camera.
Platform support is the next practical filter. The markerless motion capture plugin is supported only on Windows. MetaHuman Animator on Windows requires a DirectX 12 compatible graphics card and the default RHI set to DirectX 12. The recommended hardware page is at least 16 physical high-performance CPU cores, 32 GB RAM, and a GPU such as an NVIDIA RTX 3070, AMD RX 6800 XT, or Apple M2 Ultra with 8 GB VRAM for MetaHuman Animator and MetaHuman Creator features.
The workflow in plain editor language
The workflow starts with footage import. Video comes through Capture Manager in Live Link Hub, creating a capture data asset in the project. Then the artist configures solve settings in the MetaHuman Performance editor, runs the solve, and exports the result as an animation sequence. Finally, the exported animation gets assigned to a MetaHuman in a level sequence.
It is a fairly normal offline animation workflow, minus suits, optical markers, and a dedicated capture volume. The plugin removes the requirement for a mocap rig, helmet camera, or markers for this motion capture path. It also removes the need for a separate pipeline to stitch face and body results together when using the single-camera route.
For motion capture teams, this makes the plugin more interesting as a rough blocking, indie capture, previs, or fast iteration tool than as an automatic replacement for a controlled stage. So far we have not had accuracy metrics, solve times, cleanup effort, camera movement limits, occlusion behavior, or multi-actor support.
Pricing, licensing, and the fun-free bit
The Fab listing marks the plugin as Free, including commercial use.
Where Meshcapade fits
The plugin arrives after Meshcapade was acquired by Epic. The Max Planck Society madeMeshcapade as a Max Planck spin-off from Tübingen that develops advanced technologies for creating and animating digital humans. The team joined Epic’s Research team to contribute to technologies for Unreal Engine and MetaHuman.
What artists should actually test
The first test should be ugly. Use an ordinary webcam, a phone clip, a clean studio clip, and a clip with occluded hands. Try the same action from a static camera and from a less stable camera. Compare foot locking, finger motion, torso rotation, and contact with props. Then push the result through the exact Sequencer and retargeting path used on a real shot.
The second test should be boring. Confirm processing time, storage use, RAM use, GPU behavior, and whether the solve survives batch work without babysitting. Local processing is good. Local processing that monopolizes a workstation all afternoon is still a scheduling problem with nicer branding.
For small teams, students, indies, previs artists, and realtime prototyping, this is a practical addition to the Unreal Engine plugin shelf. For serious shot work, it is an experimental plugin that needs ruthless testing before anyone sells it to production as a solved problem.
https://www.fab.com/listings/4095b8e0-3eff-44f1-acb4-cb40b99228b9
