Title
Subtitles in VR 360º video. Results from an eye-tracking experiment
Conference name
Media for All 10 Conference
City
Country
Belgium
Modalities
Date
06/07/2023-07/07/2023
Abstract
The three hundred and sixty-degree (360º) immersive videos for Head Mounted Display (HMD) devices offer great potential in providing engaging media experiences. Understanding human behaviour is a fundamental issue and a departing point toward defining quality of experience and quality of service. Many challenges emerge due to the novelty of the format. In addition, the use of Virtual Reality (VR) headsets for long periods of time often causes nausea or motion sickness. These limitations have direct implications when testing user performance on tasks and preferences. There is no denying the central role of user needs, requirements and expectations in the testing stages and system definition. Still, one of the lessons learnt from previous end-user tests on accessibility services in immersive environments (XR) is user skills in VR technology.
A mixed-method design should be adopted to overcome the potential limitations of the applied testing methods, combining qualitative and quantitative approaches followed by triangulating the results. A common practice in experimental research on AVT studies is the multidisciplinary approach, fundamental when dealing with the nature of the medium and high impact of technology (Orero et al. 2018). User-centred research is the most popular methodology when defining and designing system requirements for technological solutions. In an ideal world, any system or process should be designed with accessibility in mind from the onset (Mével, 2020; Romero-Fresco, 2013). This will naturally lead to a born accessible system, avoiding expensive and complex afterthought solutions. Previous experiments involving 360º immersive video (Agulló and Orero 2017; Fidyka and Matamala 2018; Agulló and Matamala 2019) also follow a user-centric design but traditional use of focus groups or questionnaires is insufficient in immersive media as the user experience is now acutely personalised. Therefore, while standards are sought for general deployment, they must be developed through evaluation of the individual’s experience.
This study aims to further clarify which is the best visualisation for subtitles in immersive environments for all kinds of users. Feedback has been gathered from 73 participants (24 in Barcelona, 24 in Manchester, and 25 in Warsaw) regarding preferences and task load when watching subtitling content in 360º videos, along with additional data based on participants’ eye movements. A new framework (Hughes et al., 2020) that allows for subtitle editing and evaluation in 360º videos will be presented, along with a methodology based on triangulation of metrics, including psycho-physiological process metrics (eye movements), performance metrics (scene comprehension) and subjective self-reports (task-load and preferences). Results show that head-locked coloured subtitles are the preferred option.
A mixed-method design should be adopted to overcome the potential limitations of the applied testing methods, combining qualitative and quantitative approaches followed by triangulating the results. A common practice in experimental research on AVT studies is the multidisciplinary approach, fundamental when dealing with the nature of the medium and high impact of technology (Orero et al. 2018). User-centred research is the most popular methodology when defining and designing system requirements for technological solutions. In an ideal world, any system or process should be designed with accessibility in mind from the onset (Mével, 2020; Romero-Fresco, 2013). This will naturally lead to a born accessible system, avoiding expensive and complex afterthought solutions. Previous experiments involving 360º immersive video (Agulló and Orero 2017; Fidyka and Matamala 2018; Agulló and Matamala 2019) also follow a user-centric design but traditional use of focus groups or questionnaires is insufficient in immersive media as the user experience is now acutely personalised. Therefore, while standards are sought for general deployment, they must be developed through evaluation of the individual’s experience.
This study aims to further clarify which is the best visualisation for subtitles in immersive environments for all kinds of users. Feedback has been gathered from 73 participants (24 in Barcelona, 24 in Manchester, and 25 in Warsaw) regarding preferences and task load when watching subtitling content in 360º videos, along with additional data based on participants’ eye movements. A new framework (Hughes et al., 2020) that allows for subtitle editing and evaluation in 360º videos will be presented, along with a methodology based on triangulation of metrics, including psycho-physiological process metrics (eye movements), performance metrics (scene comprehension) and subjective self-reports (task-load and preferences). Results show that head-locked coloured subtitles are the preferred option.