Final Motion Capture Project — Capturing Movement in Gym Workouts
Project Context
Our final project — “Capturing Movement in Gym Workouts” — explores the nuances of athletic motion through motion capture.
The idea was to record realistic gym movements (push-ups, squats, yoga poses, balance ball work, etc.) and bring them into a stylized virtual environment. Our focus was both technical (recording, streaming, rigging, and retargeting workflows) and aesthetic (expressing workout through body motion and visualize with metahuman).
Streaming to Unreal Engine: Our Week 6 Focus
During Week 6, our team (Michael, Ryan, Galt, and I) concentrated on one major technical goal — live streaming full-body, hand, and facial motion data directly into Unreal Engine.
We wanted to bypass the traditional MotionBuilder workflow and instead test real-time retargeting through OptiTrack → Unreal via Live Link, combining:
- OptiTrack Motive for body motion capture
- Manus Quantum Metagloves for high-fidelity hand and finger tracking
- Unreal Live Link Face for facial motion streaming
- MetaHuman as the target character for visualization
In theory, this setup would allow us to see our captured performer instantly drive a MetaHuman inside Unreal — no post-baking or manual retargeting required.
Pipeline Overview: Real-Time Route vs Traditional Pipeline
To understand the difference, here’s how the two pipelines compare:
MotionBuilder Workflow
- Record body data in OptiTrack Motive
- Import to MotionBuilder
- Rigging & Retargeting onto a skeleton
- Import baked animation to Unreal Engine and apply it to our MetaHuman
Real-Time Streaming Workflow (Our Goal)
- Stream body data from OptiTrack to Unreal using Live Link
- Pair the live skeleton to a MetaHuman Control Rig
- Add Manus Glove data and facial Live Link feeds
- Visualize the performance live within Unreal
Essentially, the second approach collapses the pipeline into a single step — ideal for virtual production, where you can preview or even record in real time.
Week 6 Session – October 4
For this session, I invited a friend to perform as our motion-capture actor.
After calibrating the capture volume, we attached the Manus Metagloves and set up Live Link Face on an iPhone. The Manus gloves connected successfully to Motive through the Manus Plugin, and we could see clean hand and finger tracking data.
However, the body skeleton from OptiTrack did not align correctly with the MetaHuman rig inside Unreal. The naming conventions and bone hierarchy between Motive and MetaHuman skeletons differed too much for automatic mapping.
We spent roughly an hour troubleshooting — checking retargeting setups, bone axes, and Live Link sources — but the data still didn’t sync cleanly. Eventually, we decided to pause the live-streaming route and focus instead on perfecting the body + hand capture in Motive itself.
So, for the remainder of the week, we captured clean body and hand data to use later in MotionBuilder and Unreal.
Week 7 – Final Session (October 10)
By our last session, we had already captured enough motion assets for the final video collage — multiple gym actions from different performers.
Our main focus became retargeting and polishing.
Since the live streaming pipeline didn’t work reliably, we returned to the standard workflow:
- Exporting OptiTrack data to MotionBuilder
- Performing rigging and retargeting onto our MetaHuman skeleton
- Baking and exporting animations back into Unreal Engine
This gave us full control over cleanup and keyframe accuracy. It also ensured the motion quality stayed intact for editing and rendering.
Reflection
This project taught us how complex real-time motion pipelines can be.
While the Manus Gloves and OptiTrack Motive integrated beautifully on their own, combining them with Live Link Face and MetaHumans introduced synchronization challenges that go beyond simple data streaming — skeleton naming, bone mapping, and rig structure all need tight compatibility.
Still, the attempt was valuable. We now understand both workflows:
the traditional retarget-and-bake approach for reliability and the real-time streaming route for future virtual-production flexibility.
Next step — we plan to compile our captured motions into a “Gym Workout Collage” video piece that visually compares captured, retargeted, and rendered outputs, showcasing how motion capture can translate physical effort into digital expressiveness.
Credits
Team: Michael Katz, Ryan Burke, Galt Almogy, Summer Hwang
Capture Date: September 22 & 26, October 4 & 10, 2025
Tools: OptiTrack Motive + MotionBuilder + Manus Metagloves + Unreal Engine + MetaHuman + LiveLink
Project: Capturing Movement in Gym Workouts
This is our final submission for MoCap for Archive class highlighting our captured body motions. However, our team plans to continue refining the visual presentation in Unreal Engine — polishing lighting, camera work, and rendering to better communicate the energy and realism of the gym movements.