Real-Time Light Sensing in Virtual Reality

A real-time framework for evaluating visual exposure in VR scenes.

Summary

This project explores a real-time framework for estimating how a virtual-reality scene may be experienced beyond the rendered RGB image alone. The aim is to support fast visual evaluation in VR environments, where the image shown to a viewer and the light exposure experienced by the visual system are related but not identical questions.

Motivation

Many VR workflows optimize what a scene looks like on screen, but some applications also need to reason about the light arriving at the eye. A real-time estimate is useful when designers want to compare scenes, tune lighting, or evaluate visual exposure without leaving the interactive environment.

Approach

The framework takes a rendered benchmark scene as input and produces a processed representation intended for visual evaluation. The two images below show the difference between the original scene and the derived output used by the pipeline.

Display

Benchmark VR scene used as the visual input
Benchmark VR scene: the image presented to the viewer.
Processed visual-evaluation output generated from the VR scene
Processed output used for real-time visual evaluation.

Takeaway

The project demonstrates a practical bridge between interactive VR rendering and real-time analysis. Instead of treating perception as an afterthought, the workflow keeps the evaluation loop close to the scene-design loop.