11/12/2026 updated

**** ******** ****
100 % available

Software Engineer - Computer Vision & Real-time Graphics

Kawasaki, Japan
Only remote
Master's-equivalent (DESS), Digital Image Engineering, Universite Paul Sabatier, Toulouse (2004)
Kawasaki, Japan
Only remote
Master's-equivalent (DESS), Digital Image Engineering, Universite Paul Sabatier, Toulouse (2004)

Profile attachments

2026-10-25_Paul_Lacroix_Resume.pdf

About me

Software engineer in computer vision and real-time graphics, 20+ years. Specialties: camera/projector calibration, multi-view geometry, physics simulation, real-time rendering, 3D scanning and medical imaging. C++, C#, Python, OpenCV. Hands-on from prototype to deployment. Available remotely.

Unity 3dComputer Animation3D Computer Graphics3D Computer Graphics Software3D Rendering3D VisualizationAugmented RealityC++ (Programming Language)Computer-Generated ImageryPython (Programming Language)Visual SystemsVisual Technology3d ImagingVisualization SoftwareImaging DevicesVirtual RealityRendering EngineMixed Reality
Computer Vision & Camera/Projector Calibration
Expertise in camera and projector calibration, multi-camera 3D marker reconstruction, markerless face and body tracking, multi-camera stitching, 3D scanning, photogrammetry, mesh processing, depth-map and point-cloud processing.

Real-time Rendering & Graphics
Proficiency with OpenGL, GLSL/HLSL shaders, Unity, Ogre3D, custom multi-display media server development, and volumetric rendering for real-time applications and live events.

Programming Languages & Software Development
Strong command of C++, C, C#, Python, PHP, HTML, JavaScript, and assembly languages, applied across research, design, and production of custom technical solutions over 20+ years.

AR/VR & Sensing
Development of AR and VR content, use of depth cameras (Kinect, Orbbec, RealSense), motion capture with OptiTrack and MotionBuilder, IMU-based systems, and sensor-driven interactive installations.

Physics Simulation
Real-time cloth simulation, grid- and particle-based fluid simulation, and data-driven custom physics models applied to interactive installations and virtual fitting room systems.

Libraries & SDKs
Hands-on experience with OpenCV, dlib, CUDA, Qt, .NET, Kinect/Orbbec/RealSense SDKs, Adobe Photoshop/After Effects SDKs, and Nuke NDK.

Embedded & Hardware
Custom electronic-board design and development with Raspberry Pi, Arduino, and ESP32 for interactive and sensing applications.

Medical & Scientific Imaging
Real-time rendering of volume data from MRI and CT (DICOM files), ultrasound, confocal microscopy, extraction of 3D vascular models, and 3D mesh reconstruction from laser profiles.

Projection Mapping
Technical direction and development of large-scale projection mapping projects, including automatic color calibration, production workflow design, and custom After Effects and 3ds Max plugins.

Experience

Professional experience10+ years
Freelancer since2011

Languages

EnglishFluentFrenchNative speakerJapaneseGood

Project history

Independent R&D Engineer / Technical Director

Independent / Self-employed
Independent R&D Engineer and Technical Director based in Tokyo, working on computer vision, real-time graphics, projection mapping, interactive installations, and custom technical solutions for live events, museums, and broadcast.

Color Calibration and Compensation System for Projection Mapping (Software Developer)

Various clients
Irregularities in the materials composing walls make projected colors drift from the intended look. To correct this, I developed a color compensation process: the projection area is scanned with a light pattern and a camera, from which a compensation map is generated. This allows a much better projection quality, with more accurate colors.

The first version was developed for the Seikei University projection mapping. It was later deployed for a permanent installation at the Dockyard Garden, in front of the Yokohama Landmark Tower, where the show was automated and played every day, and the calibration process could be launched remotely at any time for maintenance. It was also used in other locations that cannot be disclosed because of NDA.

Interactive Art and Immersive Experience (Software developer / Content creator)

Various clients
In 2009, I conceived and managed an interactive digital signage project. I developed a proprietary sensing library (presence, motion, face detection and tracking) exposed through an API, and used it to build interactive content driven by computer vision, with customizable content templates, rendered in C++/OpenGL or Flash. I handled R&D, content design, client relations, installation and maintenance. Over the years, I created many interactive contents of various sizes. Notable examples include Aquamarine Fukushima (2010), the interactive seashore floor projection at Enoshima Night Aquarium (2014), and a permanent installation on a giant LED screen in a shopping complex in Hiroshima (2018). Most of them rely heavily on real-time physics simulation. At Hiroshima, a second content was also synchronized with a real elevator through a custom electronic interface I built. More recently, I worked on an immersive installation for Bandai Spirits (2023).

Multi-camera stitching (Software Developer)

Undisclosed client
I developed the calibration system for a real-time stitching program that combines three 4K cameras into the equivalent of a single 12K wide camera, running at 60 fps. Designed for sports, the calibration had to limit the deformation caused by the cameras' perspective correction, so that the size of the players remains constant on the left, center and right of the panorama. Being automatic, the calibration process can be run by a technician alone without advanced calibration knowledge. At such a high frame rate, real-time stitching needed advanced low-level (assembly) optimization, which was achieved in collaboration with another member of the team.

Custom multi-display media server (Software Developer)

Various clients
I created a media server solution for synchronized playback of high-resolution video across multiple displays, also handling the deployment of content to every machine. The initial version was developed during the extension of Haneda International Airport's terminal, for a 36-screen display wall. While the other members of the project focused on the creation of the video content, I was in charge of the media server development. Synchronization and smooth frame-by-frame playback were the main challenges, all while maintaining the highest possible image quality. The system was also designed to make it easy to integrate interactive content when no video was playing.

The solution was later deployed in other locations in Japan and abroad, including Osaka Castle Park, the Kitora Tumulus Museum, and immersive rooms and entrance lobbies for luxury brands in Shanghai, Ginza and Paris.

Custom optical motion capture system (System developer)

Independent project
I created my own optical motion capture solution, optimized for face and body projection mapping, and have been using it in most of my projects involving motion capture since.

Starting from industrial-grade cameras, I designed and 3D-printed custom enclosures and integrated custom electronics into each one to align the pulse of its infrared emitter area with the frame shutter. I also built a device to synchronize all the cameras together.

I then wrote the software that reads the raw camera data and processes it to obtain, in 3D space, the positions of markers placed on performers or objects. Calibration, both intrinsic and extrinsic, was a key part of this work.

Self-funded, I optimized the system to work with a minimum number of cameras for greater flexibility, rather than the large setups common in commercial systems. Inertial sensors (IMUs) were later added for a more stable orientation of tracked objects. Although it started as a personal project, such end-to-end development is rare, and it became a major building block of my later work.

Real-time Body Tracking and Projection Mapping (System developer / Technical Director)

Various clients / independent projects
I developed real-time full body tracking and projection mapping systems, adapting the tracking technology to each project. Unlike the face, which relied on a rigid body model, the full body requires tracking every joint, bone and muscle, which called for a much higher level of expertise. Latency and accuracy were again critical aspects of these projects. In 2015, I built a Kinect-based system for a promotion video in Paris. In 2017, I worked on EXISDANCE in a collaborative setup, using an original Panasonic tracking system and hardware. It was created for a promotion event, and we were later invited to present it at America's Got Talent. In 2020, PARTED was published, a personal project born from a collaboration with other artists and running on an infrared marker-based motion capture system that I developed.

Real-time Face Tracking and Projection Mapping (System developer / Technical Director)

Various clients
I created the world's first real-time face tracking and projection mapping system, and kept improving and maintaining it through years of research. The first version tracked the face using infrared markers. One of the main challenges was latency, so everything possible was brought into a single standalone application, from the motion capture process to face pose estimation and 3D rendering with animated textures. Great care was also taken on the hardware side. Another critical aspect was the various calibrations involved in the system. The system was used for the art video OMOTE, seen more than five million times on the Internet in a single week, for a Japanese TV show (SMAPxSMAP), and for other collaborations.

To make the technology easier to experience, I developed in parallel a markerless version. It was used for a Van Gogh face projection at the Tokyo Metropolitan Art Museum, then taken outdoors for The Lady in the Wood, another art video that I directed in a forest near Mount Fuji. At the Eurovision Song Contest 2019, I was invited by the French delegation to join their team, and the system was installed on stage in only 40 seconds for a live event watched by almost 200 million viewers. These projects were made in close exchange with the artists and directors involved, and their ideas shaped how the technology evolved.

Software R&D Engineer

Widget Inc.
Software R&D Engineer at Widget Inc. in Tokyo, working on research and development of software solutions.

Software Developer

Polygon Pictures Inc.
Development of a real-time task management web app for an animation studio, handling task assignment and scheduling for over one hundred employees. The app served two sides: supervisors could create and assign tasks, while creators (modelers, animators, editors, etc.) tracked and updated their work through it. Because of the studio's multi-cultural environment, it also managed automatic task creation for the translation team when necessary. This project stood out as an enriching experience, requiring interviews with each category of artist to understand their workflow and integrate it into the system. The app was used in the context of an American TV series.

Software R&D Engineer

Voxelan (3D scanners)
Software research and development across multiple tools for a 3D scanner in the dermatology and medical fields. Building on prior experience in this domain, the work especially focused on improving the quality of the scanning process, mesh generation, and texturization. It also included developing a software tool for semi-automatic body measurement, designed to process large groups of subjects.
  • Digital measurements on 3D models
  • Scan anonymization
  • Scan stitching
  • Documentation, demonstration and training related with the software
  • Upgrade of the company's website
  • Research and evaluation of methods especially in the following fields:
  • Processing of images and depth maps (Digital Elevation Model)
  • Scanner calibration, correction of the lens distortion
  • Mesh operations (polygonization, merging, simplification, comparison, positioning...)
  • Use of various data structures for optimization (octree, kd-tree...)

R&D Intern

Pierre Fabre (medical / dermo-cosmetic)
Development of the control software for a high-definition profilometry scanner built for the dermatology field, designed to capture precise body measurements. The software controlled the scanner hardware, captured image sequences, converted laser profiles into a point cloud, and reconstructed a 3D model from it. The project involved significant R&D to get the reconstruction pipeline accurate and reliable.

R&D Intern

Pierre Fabre (medical / dermo-cosmetic)
Development of the control software for a high-definition profilometry scanner built for the dermatology field, designed to capture precise body measurements. The software controlled the scanner hardware, captured image sequences, converted laser profiles into a point cloud, and reconstructed a 3D model from it. The project involved significant R&D to get the reconstruction pipeline accurate and reliable.

Portfolio

Contact form

Log in to get in touch

You need to be logged in to use the contact form.

Sign upLog in