TY - GEN
T1 - ISR
T2 - 2026 ACM Multimedia System Conference, MMSys 2026
AU - Bakhtiariazad, Ghazaleh
AU - Rehman, Haseeb Ur
AU - Shirmohammadi, Shervin
AU - Amer, Ihab
AU - Hefeeda, Mohamed
N1 - Publisher Copyright:
© 2026 Copyright held by the owner/author(s).
PY - 2026/4/6
Y1 - 2026/4/6
N2 - Cloud-based Virtual Reality (VR) gaming enables immersive experiences without the need for costly high-end consumer hardware. However, it imposes substantial bandwidth requirements due to the need to stream high-resolution, high frame rate, and stereoscopic frames to maintain immersion and prevent motion sickness. Existing techniques like foveated rendering and encoding face challenges such as reliance on costly eye-tracking hardware and sensitivity to sudden gaze shifts. In addition, prior super-resolution methods can improve fidelity but are often too computationally heavy for practical deployment. To address these challenges, we propose iSR, a system that integrates stereo-aware colorization and super-resolution to reduce transmission cost while preserving visual quality. The key idea of iSR is that it first downsamples both stereo views to reduce the total number of transmitted pixels. Then, it transmits one view in full color and the other in monochrome. This removes redundant chrominance information and further reduces the required bandwidth. On the client side, iSR reconstructs full-color, high-resolution stereo frames by transferring chroma between views and enhancing spatial resolution. Extensive experiments across multiple VR games show that iSR achieves substantial bitrate reductions while maintaining high visual fidelity. These results highlight its potential for enabling high-quality VR streaming in bandwidth-limited environments.
AB - Cloud-based Virtual Reality (VR) gaming enables immersive experiences without the need for costly high-end consumer hardware. However, it imposes substantial bandwidth requirements due to the need to stream high-resolution, high frame rate, and stereoscopic frames to maintain immersion and prevent motion sickness. Existing techniques like foveated rendering and encoding face challenges such as reliance on costly eye-tracking hardware and sensitivity to sudden gaze shifts. In addition, prior super-resolution methods can improve fidelity but are often too computationally heavy for practical deployment. To address these challenges, we propose iSR, a system that integrates stereo-aware colorization and super-resolution to reduce transmission cost while preserving visual quality. The key idea of iSR is that it first downsamples both stereo views to reduce the total number of transmitted pixels. Then, it transmits one view in full color and the other in monochrome. This removes redundant chrominance information and further reduces the required bandwidth. On the client side, iSR reconstructs full-color, high-resolution stereo frames by transferring chroma between views and enhancing spatial resolution. Extensive experiments across multiple VR games show that iSR achieves substantial bitrate reductions while maintaining high visual fidelity. These results highlight its potential for enabling high-quality VR streaming in bandwidth-limited environments.
KW - bandwidth optimization
KW - Cloud VR gaming
KW - stereo-aware colorization
KW - super-resolution
UR - https://www.scopus.com/pages/publications/105036630047
U2 - 10.1145/3793853.3795746
DO - 10.1145/3793853.3795746
M3 - Conference contribution
AN - SCOPUS:105036630047
T3 - MMSys 2026 - Proceedings of the 2026 ACM Multimedia System Conference
SP - 49
EP - 60
BT - MMSys 2026 - Proceedings of the 2026 ACM Multimedia System Conference
PB - Association for Computing Machinery, Inc
Y2 - 4 April 2026 through 8 April 2026
ER -