3D Visual Discomfort Database

Introduction

The production of high quality stereoscopic videos becomes more challenging than conventional 2D film shooting, since multi-camera aspects such as inter-camera positioning must be addressed. A major challenge is optimizing for the highest viewing comfort, best depth impression and high image quality aspects. Therefore, we have developed a small stereoscopic three-dimensional (S3D) video database with scenes of different content and characteristics. For all the videos, we provide subjective assessment data considering visual comfort, depth quantity and image quality. The resulting scores can be used to evaluate performance of quality metrics, visual discomfort prediction models and disparity mapping algorithms.

Video database

Videos

The database contains

  • Ten low-resolution (640x480), non-expert, high-motion, outdoor videos, captured by means of a Bumblebee 2 camera from Point Grey and re-rendered using two different depth comfort zones, see [A].
  • Eight high-resolution (1920x1080), expert, low-motion, indoor videos with four original videos from [1] and four re-rendered videos using our disparity mapping algorithm for quality optimization [A].
  • Three anaglyph videos for comparison of results from our optimization algorithm [A] and the one of Yan et al. [2].

Subjective assessment

17 subjects, from twenty five to fifty years old, participated in our subjective assessment. After testing them for stereo-blindness, colour-blindness and low vision according to [3], we had to break the experiment for three users because of low vision.

The videos were shown on a 55-inch stereoscopic display with the eyes of the viewer’s horizontal centred and 3.1 times the images height [1] away from the display.
Since, the subjective assessment aims at providing evidence regarding subjective image quality, depth quantity and visual comfort, discomfort, we designed four questions, similarly to [2], as follows:

  • Question 1: What is the level of image quality of the video (Bad-Poor-Fair-Good-Excellent)?
  • Question 2: What is the level of the depth quantity of the video (Bad-Poor-Fair-Good-Excellent)?
  • Question 3: What is the level of visual comfort associated with the video (Extremely uncomfortable-Uncomfortable-Mildly uncomfortable-Comfortable-Very Comfortable)?

Copyright

Permission is hereby granted, without written agreement and without license or royalty fees, to use, copy, modify, and distribute the data provided and its documentation for research purpose only. The data provided may not be commercially distributed. In no event shall the Software Competence Center Hagenberg (SCCH) be liable to any party for direct, indirect, special, incidental, or consequential damages arising out of the use of the data and its documentation. The Software Competence Center Hagenberg (SCCH) specifically disclaims any warranties. The data provided hereunder is on an "as is" basis and the Software Competence Center Hagenberg (SCCH) has no obligation to provide maintenance, support, updates, enhancements, or modifications.

If you use this database in your research we kindly ask you to reference this website and the paper below:

[A] W. Zellinger, B. A. Moser, A. Chouikhi, F. Seitner, M. Nezveda, and M. Gelautz: “Linear Optimization Approach for Depth Range Adaption of Stereoscopic Videos,” Stereoscopic Displays and Applications XXVII, IS&T Electronic Imaging, 2016.

Download of the database

The database contains:

  1. anaglyph videos: Contains three anaglyph videos showing results of [2] side by side with our optimized results.
  2. left and right views: Contains left and right stereo views of test videos. Please note, that videos '01-01_original', '06-01_original', '08-01_original' and '11-01_original' are originally taken from [1].
  3. scores: Subjective assessment results for all videos

Here you can download the zip-archive of the database.

References

[1] Goldmann, et al., "Impact of acquisition distortions on the quality of stereoscopic images." Proc. Int. Workshop on Video Processing and Quality Metrics for Consumer Electronics-VPQM, 2010, pp. 1–6. [Online]. Available: http://mmspg.epfl.ch/3diqa.

[2] Yan, Tao, et al. "Depth mapping for stereoscopic videos." International Journal of Computer Vision 102.1-3 (2013): 293-307.

[3] ITU-R, “Methodology for the subjective assessment of the quality of television pictures,” Tech. Rep. BT.500-11, 2002.