Giacomo Gerotto

Motion capture TDshoots, capture and pipeline

Five years on motion capture stages for game studios, immersive productions and live events: setting up and running Vicon volumes, keeping capture quality high during takes (sync, markers, props, fingers, face), solving and retargeting, and building the tools and real-time pipelines around the capture. Alongside the shoots, I install, train and support customers on Vicon, Xsens, Manus, Captury and Faceware.

  • Vicon, Xsens, Manus, Faceware, Captury
  • MotionBuilder, Shōgun, Unreal, Unity
  • Python, C#, NDI, Lab Streaming Layer
  • Genoa, Italy, open to relocation and travel
Email me Download CV LinkedIn
Player and ball solved live in Vicon Shōgun. Footage: GPEM

How I run a shoot

My role on client shoots at GPEM, sole Vicon distributor for Italy, Malta and the Balkans, 2021–2026. Clients hired us for the capture service, so I am not in their credits; projects under NDA are anonymised.

  1. PrepTake list from the client about two weeks ahead; I scope takes per day and the number of shoot days, and build the session plan and sequencing with the lead animator, producer and director.
  2. Capture strategyI advise the animation team on how to approach each move set, balancing what they want creatively against the limits of the capture system and their in-game or in-engine pipeline.
  3. Stage & propsCamera layout, calibration and threshold/strobe tuning per session; prop and set-piece builds to the client's 3D models; matte tape and low cameras where reflections and occlusion threaten the data. When something unusual has to be tracked or worn, such as a fast ball, a small handheld prop or a wearable that won't take a standard marker set, I design the marker layout, mounts or rig that make it solvable, 3D-printing custom parts when needed.
  4. On setI lead a crew of 1–2 technicians through suit-up, markers and ROM, brief performers, co-direct with the animation lead or director, call retakes when the data needs it, and own every technical issue during takes.
  5. Before wrapTakes named, noted and selected; data checked before the session ends, so problems are fixed on the day instead of in a reshoot.
  6. DeliveryRaw solved data or cleaned, retargeted FBX, depending on the job; afterwards I stay the client's technical contact for integration issues.

Selected productions

As service provider via GPEM. Camera counts refer to GPEM's 14 × 4 m Vicon volume unless stated.

Performance capture, Unreal / MetaHuman

Card game — dealer performance capture

Game studio2026client under NDA

  • Half-day pre-capture + 2 shoot days, single performer: full body and high-fidelity fingers (10 markers per hand)
  • 26 Vicon cameras: 16 Valkyrie VK26 + 10 Vero
  • Table prop built to the client's 3D model (rough wooden build by the crew, the client's blackjack felt on top), so hand-to-table contacts matched the game set
  • I recommended scaling the MetaHuman character to the performer's proportions from the pre-capture, and supported the CG artist through it
  • Quick live retarget to Unreal during takes for on-set previs; final retarget in post
  • Led the post with a technician reporting to me: body and finger cleanup, retargeting, clean FBX delivery
  • After delivery, diagnosed bone-scaling issues at Unreal import and in the client's hand-off to integration, using Maya and Blender
VR / MR, Venice Immersive 2025

Black Cats & Chequered Flags

MIAT, Milanshot Oct 2024

  • Multiplayer VR/MR experience on F1 champion Alberto Ascari: Official Competition at Venice Immersive 2025, shown at MAUTO Turin, Museo Ferrari Maranello and Silverstone Museum
  • Half-day pre-session with a quick Unity preview, then 1 shoot day with the director and art director on set; 3 performers including a ~10-year-old child
  • Took measurements from the client's 3D car models and directed minimal wooden rigs at the contact points only: steering wheel, seat, trackside railing, plus taped floor marks
  • ~25 Vicon cameras (Vero + Vantage); cleanup and retargeting onto 3 characters done personally

Venice Biennale page

Game, recurring client

Unreleased action game

European publisher's studio2023–2024client under NDA

  • 4 shoot days across 18 months for a returning client, with a consistent setup from one session to the next
  • Locomotion, combat, deaths, interactions and cutscenes; up to 3 performers at once
  • Tracked props: up to 2 rifles, a table and a chair
  • Finger capture on 2 performers simultaneously (5- and 10-finger marker sets)
Game, large reflective prop

Licensed motorsport racing game

Italian game studio2025client under NDA

  • 2 shoot days: riders on a real rented motorcycle, celebrations, podium, mechanics, crowd layers, cutscenes; up to 3 performers
  • Per-session setup against the bike: low-mounted cameras to recover leg and pelvis markers, matte tape on helmets, bodywork and reflective surfaces, per-camera threshold and strobe tuning
  • Delivery: raw solved data for the client's in-house cleanup
Object tracking

Football app — player and ball capture

2025client under NDA

  • A former professional footballer, now a coach: shots, juggling, dribbling, headers
  • Ball tracked for position and rotation: benchmarked several retroreflective tapes, then built the ball as a Shōgun object with ~30 flat markers in six clusters, one per side, so it solves from any orientation
  • 150 Hz, aperture opened to keep markers bright, circularity threshold and environment drift tolerance tuned
  • Cleanup by a technician under my direction, with hands-on fixes on the hardest takes
Remote crew, Venice 2024

Weightless — Arilena Ara

Vnova (ImmersiX) and Chokolate TV2024

  • Cloud-rendered cinematic VR experience and music video, previewed at the 81st Venice International Film Festival
  • Paris shoot (Jan 2024): planned the sessions and trained the GPEM field crew on Xsens MVN with Awinda and Manus Metagloves; they ran the days on their own with phone support from me
  • Nov 2024: sole cleanup in Shōgun Post of US-captured Vicon body and finger data, working from video reference with director Guido Gallo Fabris
Live, single take

Novo Nordisk 100th anniversary

Velvet GroupVilla Miani, Rome2023

  • ~20-minute live talk by a real-time avatar of founder August Krogh, performed by an actor in one continuous take
  • Built and operated the pipeline: Xsens MVN + Manus Metagloves → MotionBuilder live retarget → Unreal Engine for the agency's show output
  • Hot-standby MotionBuilder machine on the same stream
  • Anticipated IMU drift over the long take and had the shot reframed to exclude the exit door, so the actor's small offset at the end never showed

Project page (GPEM)

R&D, performance capture and tools

Work where I built or integrated the capture system itself.

Live streaming, real-time, remote

Remote live workshop for IIDEA

IIDEA, Italian games industry associationJan 2026

Footage: GPEM
  • Real-time performance capture workshop held on Discord; I ran it with a GPEM technician reporting to me
  • Designed the streaming setup: three PCs wired directly and linked over NDI (NDI Tools) into an OBS multi-view with an Unreal screen share, a Shōgun screen share and a webcam on the lab
  • Two performers on the production marker template with 10 markers per hand, plus tracked props: a glass, a rifle, a table, a cigar and a lighter
  • A Vicon-tracked virtual camera (iPad on a handheld rig) for live camera moves in Unreal
  • Everything solved and streamed live onto MetaHumans in Unreal Engine
Body + face + voice, credited

GNVR Experience

2024Tech Transfer call, CTE Genova

  • VR visit of the Abbazia della Cervara with an AI guide; I found the call, wrote the proposal, led the project and built the Unity app
  • Split performance capture of the guide, actor and art historian Hans Di Maria: Vicon body and fingers in Genoa; face and voice at Sud Sound Studio, Rome, where I ran the Faceware Mark IV headcam with Faceware Studio, locked to audio with Tentacle Sync timecode

Project page and credits (GPEM)

Sync, multi-sensor, credited

MIMOSE

2024–2025PNRR-funded R&D with Politecnico di Torino, University of Genoa and Wondertech

  • One time-synchronised platform for Vicon, Movella DOT IMUs, Cometa EMG and an Orbbec RGB-D camera, over Lab Streaming Layer, with live output to Unity and Unreal
  • Named Technical & Quality Lead in the funded project documents; day-to-day project manager; later extended the platform with the LSL LabRecorder
Real-time IK, C#

Full-body avatar from sparse rigid bodies

Politecnico di Torino research lab2021–2022

  • Unity IK plugin driving a full avatar in VR from 6–11 live Vicon rigid bodies, leaving the body free for clinical markers
  • In-build subject calibration and editor tools for character setup; commissioned on site in May 2022
Underwater, markerless

DIP Working — diver training in VR

2021regional R&D consortium led by BK

  • Optical markers ruled out because water absorbs near-infrared; switched to Captury markerless with 8 cameras in sealed housings at 5 m depth
  • Dry feasibility tests in the lab (Vicon vs Captury into Unity) and Unity integration, then a final test in a 5 m deep pool; prototype tested with professional divers

Project page (GPEM)

Art installation, credited

Amarcord

2022–2023artist Meryl Bougo, choreography by Eric Sourdeau

  • Dance captured on 28 Vicon Vantage cameras and turned into CNC-bent steel and a 3D-printed shell
  • Rigged the volume and ran the sessions; found the trajectory export via FBX that the artist and fabricators needed

Project page and credits (GPEM)

Genlock, timecode, camera tracking

Sync and camera tracking know-how

2023–2024GPEM in-house know-howdemos and customer support

  • Tasked with building GPEM's in-house know-how: researched the setup, sourced the hardware (sync and timecode cabling, camera, tracking crown) and worked out the integration closely with Vicon's support engineers
  • Wired the chain end to end: Blackmagic SDI camera with a Vicon tracking crown → Vicon Lock → Evertz timecode → Blackmagic sync generator
  • Used it for demos, and supported Digital Kinetics (Zagreb) and Rupture (Malta) on timecode and sync
Tools, 2025–2026

Pipeline tools

Python and C#, using AI-assisted development that I design and review

  • Pose2Sim, OpenSim and FreeMoCap integrated into a Vicon Nexus workflow
  • Unity → Blender export scripts to FBX and GLB
  • Shōgun Post and Maya scripts for repetitive capture and cleanup steps
  • Support chatbot over GPEM's knowledge base for Vicon, Xsens, Manus, Captury, Theia and Cometa

Also: Vola come una farfalla (2021), support on a temporary 10 × 5 m Vicon install for an AR book with Olympic medallist Alessia Maurelli.

Installations, training and support

I joined GPEM in 2021 to help grow a company of mostly biomedical engineers into entertainment and development. Beyond shoots, that means technical pre-sales, installing systems, training teams on site and supporting them afterwards.

Systems and skills

Beyond shoots, I install, train and support GPEM customers across the whole product range, with Vicon as the main line: entertainment studios as well as research labs and API integrations.

Optical

Vicon Shōgun Live & Post, Nexus, Valkyrie / Vantage / Vero, Lock Lab

Inertial, hands, face, markerless

Xsens MVN (Awinda), Manus Quantum Metagloves, Faceware Mark IV headcam with Studio, Analyzer and Retargeter, Captury, Movella DOT

Other capture & XR

iPhone ARKit face capture, Rokoko, HTC Vive, Meta Quest / Oculus

Sync & real-time

Genlock, timecode and camera tracking (Blackmagic SDI camera with tracking crown → Vicon Lock → Evertz timecode → Blackmagic sync generator), Tentacle Sync timecode, Lab Streaming Layer, live streaming to Unreal and Unity, MotionBuilder live retarget

Post & integration

Cleanup and solving, retargeting incl. MetaHuman, FBX delivery; MotionBuilder, Maya, Blender, Unreal Engine, Unity

Development & fabrication

C#, Python, Unity plugins and editor tools, Shōgun/Nexus scripting; Claude Code and Cursor; 3D printing of custom mounts and marker carriers (Bambu Lab, Cura)

Demos, workshops and events