Creating Panorama Virtual Tour Systems for the Built Environment: A Practitioner Perspective

In the evolving landscape of digital representation and exploration of built environments, the creation of panorama virtual tour systems has become a focal point for enhancing our interaction with physical spaces. A panoramic image is a wide-angle view photograph that captures a scene or environment in a broad and panoramic format (see Fig. 1). It can be created using panoramic camera equipment or through the stitching of individual images taken by digital single-lens reflex cameras [1, 2]. While these images are static by themselves, connecting multiple panoramas allows the creation of virtual tour systems. This digital approach has left an indelible mark across diverse fields, including education, tourism, and entertainment [3, 4]. Although panorama virtual tour systems are not as immersive or interactive as a fully reconstructed virtual reality (VR) environment, they preserve a great level of details of the surroundings and are often more accessible than VR applications that require the use of a head-mounted display. Notably, a web-based approach emerges as the commonly adopted way to host panorama systems due to its cross-platform support. Despite the strides made in virtual tour systems, a noticeable gap exists concerning the smoothness and degree of freedom in user movement [5]. Many systems primarily support transitioning between scenes, lacking the nuanced step-by-step movement effect crucial for a heightened sense of presence and spatial awareness [6]. In this paper, we present the design and development of a panorama virtual tour system that allows step-by-step continuous movement in the scene. We also discuss the practical constraints and share the lessons learned from our practices in the four-step procedure encompassing path planning, data collection, tour design, and multimedia embedding. Our work offers valuable insights to guide practitioners, such as VR developers and technical teams of universities and public institutions in the development of panorama virtual tour experiences.

Creating a virtual tour system requires a series of planning and data collection work before the system development. In this section, we summarize a four-step procedure that details the design and development of a virtual tour system (see Fig. 2). For the data collection, we used a Nikon D7500 DSLR camera with a 14 mm f/2.8 ultra-ultrawide lens that can cover a wide 114° angle of view. The camera was mounted to a tripod for stability. Using the north campus of the Xi’an Jiaotong-Liverpool University (XJTLU) as an example environment, we captured over 700 images (4096 x 2160) in RAW and JPEG file formats. These image data were used to create 65 panoramic images (6528 x 3264), with a total size of 2.61 GB. Adobe Photoshop and PTGui² were used for image processing and stitching. We used the krpano³ and everpano⁴ to design the virtual tour and embed multimedia information. These tools are used because they are the mainstream approaches in the construction of web-based virtual tours and have rich plugin support for extended features.

We invited 12 participants (9 males, 3 females, aged M = 25.5, SD = 7.8) to use the virtual tour system and evaluate the usability using the System Usability Scale [12]. Overall, the system had a usability score of 72.71 (SD = 15.94), indicating an acceptable (>68) system usability. Detailed results are shown in Table 1.

Overall, participants had a high acceptance of the virtual tour system. They found the system easy to use (M = 4.5, SD = 0.52) and agreed that the system was easy to learn (M = 4.42, SD = 0.51). They also felt confident using the system (M = 4.25, SD = 0.75) and were willing to use it frequently (M = 3.92, SD = 0.9). The evaluation also pointed out some areas of improvements. For example, some participants found the system complicated and had unconnected places. Future improvements can be made to address these issues to improve the usability and user experience of the system, such as shortening the distance between hotspots and supporting automated touring.

In this paper, we present the design and development of a panorama virtual tour system that supports step-by-step continuous movement in the scene. Instead of delving into the algorithms and techniques, we focus on the practitioner’s perspective and reflect on the development process. In addressing the practical constraints and challenges encountered during the development of our panorama virtual tour system, we implemented a series of strategic solutions and gained valuable lessons in the process. For the challenge of path planning and the size of the environment, we first collected data from a small area of the scene (i.e., the entrance), planned the distance between two panoramas, and estimated the number of images needed. This involved multiple iterations between path planning and data collection. Through experimentation, we discovered that choosing cloudy days and leveraging holiday or weekend time yielded optimal results for data collection. This helped avoid having bright reflections, sharp shadows and too many people in the scene. To mitigate challenges related to data collection complexities, we checked weather reports and conducted all data collection during similar times of the day to maintain consistency. Additionally, we iterated between the data collection and image processing, adopting an agile approach with short loops so that we could refine image stitching and modelling tasks. It was learned that having objects with recognisable features in the overlapping areas will increase the efficiency in image stitching. For example, buildings usually have more recognisable features than trees and roads. The embedding of multimedia requires a careful balance between visual effects and web response time. While high image quality enhances visual effects, it comes at the cost of longer times needed for loading and rendering. Subjective factors, such as expertise for content creation and system maintenance, are also important constraints. Our project experience demonstrated that a motivated individual with some technical background, such as a master’s student, can self-learn the necessary techniques within a reasonable timeframe (~30 h). Finally, it is often inevitable to include passengers when capturing panoramas in public open areas. It is important to implement transparent and ethical practices, ensuring that user concerns are addressed and respecting privacy throughout the virtual tour system’s development and deployment.