Camera Tracking
RIG

The POPART camera tracking solution provides a camera RIG with two witness cameras shutter synchronized with the main camera. We provide scripts to conform data and perform fully automatic camera tracking of your footage. Additionally, we provide scripts to generate scenes for your preferred camera tracking software for manual corrections.

Post-Production & Previz Servicing

LABO and Mikros Image propose Previz integrated with Post-Production servicing.

Open Source

We build a fully integrated software for 3D reconstruction, photomodeling and camera tracking. We aim to provide a strong software basis with state-of-the-art computer vision algorithms that can be tested, analyzed and reused. Links between academia and industry is a requirement to provide cutting-edge algorithms with the robustness and the quality required all along the visual effects and shooting process.

The POPART software builds on a set of core libraries, both newly written ones and such that are adapted from existing software. All of these core libraries are based upon open standards and released in open source. This open approach enables both us and other users to achieve a high degree of integration and easy customisation for any studio pipeline.

Beyond our project objectives, open source is a way of life. We love to exchange ideas, improve ourselves while making improvements for other people and discover new collaboration opportunities to expand everybody’s horizon.

  OpenMVG

Open Multiple View Geometry

OpenMVG library solves classical problems in Multiple View Geometry and provides a full Structure-from-Motion pipeline with a strong focus on accuracy. The library scope is not limited to the creative sectors but can be used in a wide spectrum of fields, like urban planning, industrial design, archeology, medical applications, etc.

  MayaMVG

Photomodeling plugin - Autodesk © Maya

MayaMVG allows graphic artists to do photomodeling on top of a 3D reconstruction (point cloud and cameras) with pixel precision.

  ofxMVG

OpenFX Camera Localization and Calibration plugins.

The CameraLocalizer plugin estimates the camera pose of an image regarding an existing 3D reconstruction generated by openMVG. The plugin supports multiple clips in input to localize a RIG of cameras (multiple cameras rigidly fixed). The LensCalibration plugin estimates the distortion parameters according to the couple camera/optics.

  CCTag

Concentric Circles Tag

This library allows you to detect and identify CCTag markers. Such marker system can deliver sub-pixel precision while being largely robust to challenging shooting conditions (e.g. motion blur, occlusion and poor lighting conditions).

  PopSIFT

Scale-Invariant Feature Transform (SIFT)

This library provides a GPU implementation of SIFT.
25 fps on HD images on recent graphic cards.

Open Datasets

We released 2 datasets allowing to evaluate 3D reconstructions and camera tracking.

Levallois Town Hall

This dataset consists of 596 rendered images along with a 3D ground truth. It allows to run 3D reconstruction solutions on these 596 images to measure the reconstruction quality according to the 3D ground-truth. We also provide the rendering of a virtual camera to evaluate camera tracking with different rendering options.  This dataset consists of 596 rendered images along with a 3D ground truth. It allows to run 3D reconstruction solutions on these 596 images to measure the reconstruction quality according to the 3D ground-truth. We also provide the rendering of a virtual camera to evaluate camera tracking with different rendering options. 

Green Screen Studio

This dataset is an example of an everyday production with professional cameras and green screen studio with artificial lighting environment. The dataset contains raw video files shot on RED EPIC, with additional witness camera footage shot in a mixed reality studio setting with CCtag fiducial markers attached to the wall and ceiling. The goal of this dataset is to provide a reference footage for improving the open source CCtag fiducial marker library. It contains a reference camera tracking and all material required to build a virtual composite using the POPART camera tracking system.

Toulouse Capitole

This dataset collects 56 images taken in Place du Capitole in Toulouse (FRA) and 4 related videos taken while moving around in the place. The aim of the dataset is to test and evaluate the camera tracking algorithms developed for the POPART project, and help the reproducibility of the experiments. The camera tracking algorithms developed for the POPART project are based on model tracking of an existing 3D reconstruction of the scene. First a collection of still images are taken and a SfM pipeline is used to perform the 3D reconstruction of the scene. As result of this first step, a 3D point cloud is generated. The camera tracking algorithms are then based on camera localization techniques: each frame is individually localized w.r.t. the 3D point cloud using the photometric information (SIFT features) associated to each point. This allows to align the point cloud to the current frame and thus compute the camera pose.

Cap Digital Courtyard

This dataset collects 197 images taken in courtyard of Cap Digital building in Paris (FRA) and 2 video sequences taken with a camera rig composed of 1 main camera and 2 witness cameras on the side looking outwards. The aim of the dataset is to test and evaluate the camera tracking algorithms developed for the POPART project, and help the reproducibility of the experiments. The camera tracking algorithms developed for the POPART project are based on model tracking of an existing 3D reconstruction of the scene. First a collection of still images are taken and a SfM pipeline is used to perform the 3D reconstruction of the scene. As result of this first step, a 3D point cloud is generated. The camera tracking algorithms are then based on camera localization techniques: each frame is individually localized w.r.t. the 3D point cloud using the photometric information (SIFT features) associated to each point. This allows to align the point cloud to the current frame and thus compute the camera pose. 

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 644874.