Research on methods for optimising the performance of VR applications in low-bandwidth browsers

The relevance of the research lay in the need to improve the performance of WebVR applications in browser environments with limited network bandwidth, which is important for users with slow or unstable Internet connections. The aim of the work was to develop and evaluate effective methods that reduce scene loading times, optimise the amount of data transferred, and maintain a stable frame rate without compromising the quality of interaction. The study used experimental modelling of a virtual reality environment with limited network parameters. The work used open libraries for three-dimensional graphics with the ability to progressively load 3D resources. Several optimisation methods were implemented and compared, including multi-level model detailing, deferred loading, server-side geometry simplification, data compression, and texture decompression. The effectiveness was evaluated based on loading time, traffic volume, smoothness of playback, and system response. The results of the study showed that combining multi-level detailing with texture compression can reduce the amount of data transferred by up to 70% without noticeable loss of image quality. It was found that the use of progressive loading significantly reduces the initial scene rendering time, while client-side caching reduces repeat traffic by 90%. The most effective strategy was found to be the initial loading of simplified models with the subsequent asynchronous addition of detailed objects, which supports stable operation even at low Internet speeds. The practical value lies in the application of the developed methods in the creation of WebVR applications for education, medicine, commerce, and entertainment, especially in conditions of limited internet connection. This contributes to expanding the audience and reducing infrastructure costs.