Research scientist · Madrid

Making digital humans feel human.

I work at the intersection of computer vision and graphics—building models that understand human shape and motion, and turn them into compelling visual experiences.

Portrait of Javier Romero
Javier Romero
Madrid, Spain

About

Research that moves between pixels and people.

I am a Research Scientist Lead at Meta, where I work on human data collection and modeling toward VR communication that is indistinguishable from reality. For more than fifteen years, my research has explored how computers perceive, reconstruct, and generate people.

My academic work helped establish influential parametric models of the body, face, and hands—including SMPL, FLAME, and MANO, and MHR. In industry, I have led and contributed to work spanning personalized garments, fitness, new shopping experiences, and photorealistic avatars.

15+years in human modeling
4foundational human models
CV + CGresearch at the intersection

Research highlights

Models, methods, and virtual people.

Avat3r teaser showing high-fidelity reconstructed head avatars 03

ICCV · 2025 · Avatars

Avat3r

Large animatable Gaussian reconstruction for high-fidelity 3D head avatars.

Tobias Kirschstein, Javier Romero, Artem Sevastopolsky, et al.

Paper
Paper teaser showing a sequence of expressive controllable 3D avatars 04

3DV · 2025 · Avatars

Drivable 3D Gaussian Avatars

Controllable, high-fidelity human avatars represented with 3D Gaussian splats.

Wojciech Zielonka, Timur Bagautdinov, Shunsuke Saito, et al.

Paper
SMPLpix pipeline from body model and camera inputs to neural avatar rendering 05

WACV · 2021

SMPLpix

Neural avatars built from a compact 3D human model and sparse posed images.

Sergey Prokudin, Michael J. Black, Javier Romero

Paper
FACSIMILE teaser showing a person reconstructed as a detailed 3D body scan 06

ICCV · 2019

FACSIMILE

Fast, accurate, textured 3D scans of a person from a single image.

David Smith, Matthew Loper, Xiaochen Hu, et al.

Paper
MANO paper teaser showing articulated 3D hand and body models 07

SIGGRAPH Asia · 2017

Embodied Hands / MANO

A learned model for capturing and representing expressive hands together with the body.

Javier Romero, Dimitrios Tzionas, Michael J. Black

Paper
FLAME teaser showing expressive parametric 3D face models 09

SIGGRAPH Asia · 2017 · Human models

FLAME

A model of facial shape and expression learned from expressive 4D scans.

Tianye Li, Timo Bolkart, Michael J. Black, Hao Li, Javier Romero

Paper
GRASP taxonomy teaser showing diagrams of human grasp types 10

IEEE THMS · 2016 · Perception

The GRASP Taxonomy

A systematic taxonomy that organizes the breadth of human grasp types.

Thomas Feix, Javier Romero, Heinz-Bodo Schmiedmayer, et al.

Paper
Keep it SMPL teaser showing 3D human pose and shape reconstruction from photographs 11

ECCV · 2016 · Perception

Keep it SMPL

Automatic estimation of 3D human pose and shape from a single image.

Federica Bogo, Angjoo Kanazawa, Christoph Lassner, et al.

Project
SMPL paper teaser showing a family of realistic posed 3D human body models 12

SIGGRAPH Asia · 2015

SMPL

A learned, skinned, pose-aware model that made realistic 3D bodies practical to animate.

Matthew Loper, Naureen Mahmood, Javier Romero, et al.

Project

Complete bibliography

All publications

74 records · newest first · select a year

Curriculum vitae

A career in human-centered vision and graphics.

Research leadership across Meta, Amazon, Body Labs, and the Max Planck Institute for Intelligent Systems.

Open full CV
  1. 2021—nowMetaResearch Scientist Lead · Madrid
  2. 2017—2021Amazon / Body LabsPrincipal & Senior Applied Scientist · Barcelona
  3. 2012—2017Max Planck InstituteResearch Scientist & Postdoctoral Researcher · Tübingen
  4. 2007—2011KTH Royal Institute of TechnologyPhD in Computer Science · Stockholm

Get in touch

Ideas are better
in conversation.