ULB Plenoptic 2.0 Raytrix R8 Video with Single Moving Camera: DeerComplexToys, Part 1

The test sequence "ULB Plenoptic 2.0 Raytrix R8 Video with Single Moving Camera: DeerComplexToys, Part 1" is provided by Daniele Bonatto, Sarah Fachada, Hamed Razavi Khosroshahi, Gauthier Lafruit, and Mehrdad Teratani, members of the LISA department, EPB (Ecole Polytechnique de Bruxelles), ULB (Université Libre de Bruxelles), Belgium.

License:

CC BY-NC-SA

Terms of Use:

Any kind of publication or report using this sequence should refer to the following reference:

[1] Sarah Fachada, Daniele Bonatto, Hamed Razavi Khosroshahi, Gauthier Lafruit, and Mehrdad Teratani, "ULB Plenoptic 2.0 Raytrix R8 Video with Single Moving Camera: DeerComplexToys, Part 1" 2024.07, 10.5281/zenodo.12546958.

@misc{bonatto_ulb_deercomplextoys_moving_2024,
    title = {{ULB} {Plenoptic} 2.0 {Raytrix} {R8} {Video} with {Single} {Moving} {Camera}: {DeerComplexToys}, {Part} 1},
    author = {Bonatto, Daniele and Fachada, Sarah and Razavi Khosroshahi, Hamed, and Lafruit, Gauthier and Teratani, Mehrdad},
    month = jul,
    year = {2024},
    doi = {10.5281/zenodo.12546958}
}

Production:

Laboratory of Image Synthesis and Analysis, LISA department, Ecole Polytechnique de Bruxelles, Université Libre de Bruxelles, Belgium.

Content:

This dataset contains a dynamic scene featuring moving toys captured with a Raytrix R8 camera [1].

The dataset is split in two Zenodo links:

10.5281/zenodo.12546958 (this, sequences 0-2)

10.5281/zenodo.12668349 (sequences 3,4)

The dataset is separated into five different sequences of 300 frames each:

1. Sequence 0:

    - A mix of a non-moving camera for about half of the video, followed by the camera starting to move in a linear fashion.

2. Sequences 1 to 4:

    - The same video split into 300 frames each, starting only when the camera begins its movement.

Additionally, the dataset includes a grid of views and calibration images captured with the plenoptic camera under the same lighting conditions. These calibration images include views of a chessboard rotated in several positions.

The sequences consist of a linear movement of the camera, the images are named R8-C-A-U3-B028-RS-A - 2081_000000000X_Processed.png, where the X represent the frame number.

The grid follow the format of `X{x position}_Y{x position}_Processed.png` it is composed of images acquired between x=0 and x=20 by steps of 5cm and y=00 and y=100 by steps of 10cm.

Calibration files are provided in XML format. The datasets were acquired with a robotic bench made at ULB [2][3].

The dataset contains:

- `sequences` zip files containing:

  - one zip file for each sequence, with plenoptic images in PNG format.

  - XML calibration files.

- A `grid` folder containing:

  - Plenoptic images in PNG format of the grid views.

  - XML calibration files.

- A `checkerboard` folder containing:

  - Plenoptic raw images in PNG format of the checkerboard in various positions.

  - XML calibration files.

References and links:

[1] https://raytrix.de/

[2] D. Bonatto, A. Schenkel, T. Lenertz, Y. Li, et G. Lafruit, « [MPEG-I Visual] ULB High Density 2D/3D Camera Array data set, version 2 [m41083] », in ISO/IEC JTC1/SC29/WG11 MPEG2017/M41083, Torino, Italy, juill. 2017.

[3] D. Bonatto, « From multi-modal capture to photo-realistic view synthesis - A high-quality and real-time multiview approach », 2024.

Acknowledgments:

Sarah Fachada is a Postdoctoral Researcher of the Fonds de la Recherche Scientifique - FNRS, Belgium. This work was supported in part by the HoviTron project (no. 951989), the FER 2021 project (no. 1060H000066-FAISAN), the Emile DEFAY 2021 project (no. 4R00H000236), and the FER 2023 project (no. 1060H000075).