Abstract
The multi-channel stereoscopic display technology has become an attractive frontier research area in recent years, supporting users with correct spatial perception and large vision experience. This paper studies on the easiest case of double-channel. Focusing on the problem of complex network programming in traditional synchronous display method, a design plan of double-channel stereoscopic display system on one PC based on OpenGL is proposed, which reflects some innovation of the paper. A series of performance tests have been taken and the results agree well with the need for three-dimensional effect and synchronism.
Introduction
Creating immersive, large-scale 3D environments typically requires multi-channel displays to achieve high resolution and a wide field of view. Traditionally, this is achieved through distributed computing where a "master" node synchronizes multiple "slave" nodes across a network. However, writing robust network code for perfect frame synchronization is notoriously difficult and creates a significant software bottleneck. This paper proposes a streamlined approach: a double-channel stereoscopic display system running entirely on a single workstation using OpenGL, thereby bypassing complex network programming and achieving automatic synchronization.
Methodology
The system architecture centers on a single workstation capable of quad-buffered stereo rendering. The core technical approach involves several key components:
Parallel Binocular Model
To ensure viewer comfort, the researchers utilized a parallel binocular model rather than a convergent one. By keeping virtual cameras parallel, the system avoids vertical parallax, a common cause of motion sickness and visual fatigue in 3D environments.
Scene Division and Overlap
To create a seamless image on a large screen, the visual frustum is divided into two channels. A critical aspect of this division is the inclusion of an overlap region. As illustrated in Figure 3, each channel covers a specific portion of the scene (4/7 of the total width) with a calculated overlap to facilitate edge blending and maintain visual continuity.
OpenGL Rendering Pipeline
The software leverages OpenGL's quad-buffering capabilities, managing front-left, front-right, back-left, and back-right buffers. As shown in Figure 5, this allows the GPU to handle frame timing and synchronization automatically across two rendering windows on a single OS instance, outputting the respective parts of the scene to the large screen.
Hardware Implementation
The signal travels from the PC through a VGA distributor to AP converters. These converters perform the critical tasks of edge blending and color balancing in the overlapped region before the final projection via four projectors (two per channel, utilizing polarizers and polarized glasses for the 3D effect).
Results and Discussion
The implementation successfully produced a stable double-channel 3D system with a resolution of 2048x768 at a 60Hz refresh rate. Qualitative testing confirmed a consistent and high-quality three-dimensional effect. By eliminating master-slave network communication, the design significantly improved development efficiency, allowing focus to remain on 3D content rather than the network stack.
However, limitations remain. The single-PC approach is currently constrained by physical GPU outputs, typically limiting it to two channels. Additionally, the use of all outputs for projection leaves no dedicated monitor for the operator. Future work will explore scaling to more channels and investigating auto-stereoscopic technologies to improve the user experience.
Summary updated