Light field video dataset captured by a R8 Raytrix camera
(with disparity maps)

Contacts: L. Guillo, X. Jiang, C. Guillemot

When using this dataset in your research, we will be happy if you cite L. Guillo, X. jiang, G. Lafruit, C. Guillemot, Light field video dataset captured by a R8 Raytrix camera (with disparity maps), ISO /IEC JTC1/SC29/WG11 MPEG2018/m42468, ISO/IEC JTC1/SC29/WG1 JPEG2018/m79046, INTERNATIONAL ORGANISATION FOR STANDARDISATION,ISO/IEC JTC1/SC29/WG1 & WG11, April 2018, San Diego, CA, US.(pdf)

Data

The R8 Raytrix dataset is composed of 3 video sequences recorded with a R8 Raytrix video camera fitted with a 35mm lens. These videos are :

ChessPieces
Boxer-IrishMan-Gladiador
ChessMovingCamera

The first two videos, "Chess" and "Boxer-IrishMan-Gladiator", have captured a scene placed on a turn table. The camera is then located at 35cm from the scene. In the third one, the scene is static but the camera is moving. For this last video, the camera is located at 60 cm from the scene.For all of these videos, each frame has a 1920x1080 resolution. These videos are presented hereafter (the video quality is reduced due to compression, the quality of extracted views from the light fields are not degraded):

Examples of frames extracted from video sequences

For each of these ligthfield video sequences, the Raytrix API was used to extract twenty-five views per frame from the lightfield. More precisely, the function Rx::ApiLF::RxSetPar with the parameter "Rx::ApiLF::EPar::VirtCamPinholeStd_ViewOffsetMM_g" allows to specify an offset of a virtual camera in the plane perpendicular to z-axis of main lens. The offset has an horizontal value of 4mm and vertical value of 4mm and are measured from the center of main lens. The extracted views are numbered in the downloadable zip files as depicted in the diagram below (e.g.: sa_1_3.png for the coordinates (1, 3)).

Twenty five views extracted per frame — 25 virtual cameras per frame one for each extracted views

The following videos are made using the 25 extracted views from the first frame. The order of display is a serpentine order from the left up corner to the right bottom corner; i.e. from (1,5) up to the right of the row then downward to the second row and up to the left of this row and so on up to (5, 1)):

Videos of views belonging to the first frame

For each frame, 25 views are extracted from a Raytrix raw file related to the current frame and also a raw picture corresponding to the picture captured by the sensor. These raw pictures can be downloaded with the 25 views. Here are below as a video, the raw pictures related to each frame for all videos :

Raw frames extracted from video sequences

For each extracted views, a disparity map was computed. Here are below, as a video, the disparity maps related to the central view for the 3 videos:

Disparity maps associated to the central view

ChessPieces

The video sequence "ChessPieces" contains four chess pieces, two knights and tow pawns turning on a turn table and with a calendar as background. It is composed of 300 frames. The 25 extracted views per frame are available below for download. Data have been split into three files to ease the download:

The calibration parameters used for this video are provided as a XML file downloadable here.

A disparity map for each view of each frame was computed. Data have been split into three files to ease the download:

For this light field scene, the minimal value of the estimated disparity (between adjacent views) is d_min = -1.52 pixels and the maximum value is d_max = 0.45 pixels. The adopted convention is the one where the disparity value in the foreground is superior to the one in the background.

Boxer-IrishMan-Gladiador

The video sequence "Boxer-IrishMan-Gladiador" contains three Lego characters turning on a turn table and with a calendar as background. It is composed of 300 frames but the first one is numbered 400. The 25 extracted views per frame are available below for download. Data have been split into three files to ease the download:

For this light field scene, the minimal value of the estimated disparity (between adjacent views) is d_min = -1.50 pixels and the maximum value is d_max = 0.39 pixels. The adopted convention is the one where the disparity value in the foreground is superior to the one in the background.

ChessMovingCamera

In the video "ChessMovingCamera" a moving camera films a static scene composed of chess pieces. A calendar is used as background. It is composed of 300 frames but the first one is numbered 210. The 25 extracted views per frame are available below for download for the three values of dx: 4, 8 and 12 mm. Data have been split into three files to ease the download:

For this light field scene, the minimal value of the estimated disparity (between adjacent views) is d_min = -1.67 pixels and the maximum value is d_max = 0.11 pixels. The adopted convention is the one where the disparity value in the foreground is superior to the one in the background.

Acknowledgement

This work has been supported by the EU H2020 Research and Innovation Programme under grant agreement No 694122 (ERC advanced grant CLIM).

Copyright

Creative Commons License All datasets and benchmarks on this page are copyright by us and published under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License. This means that you must attribute the work in the manner specified by the authors, you may not use this work for commercial purposes and if you alter, transform, or build upon this work, you may distribute the resulting work only under the same license.