Transparent Polyhedron Rendering in AR via Fresnel Weighted Mixing

The rendering process is segmented into distinct stages: obtaining reflected light colors, sampling refracted light colors from background image or ambient map, and compositing these with weighted mixing. This segmentation allows each lighting effect to be calculated independently and efficiently, then combined to achieve realistic visual results without requiring complete ray tracing of all light paths.

The patent applies partial action by selectively rendering only the most visually significant light interactions (reflection and refraction) rather than all possible light paths. The Fresnel coefficient is used to weight the contribution of reflected versus refracted light, focusing computational effort on the dominant effect at each surface point while still achieving photorealistic appearance.

The patent introduces intermediary data structures (contrast control map, internal rendering effect map, ambient map, dispersion map) that pre-process and store lighting information. These maps act as intermediaries between the complex physics of light interaction and the final rendering output, allowing sophisticated optical effects to be achieved through map sampling and composition rather than direct physical simulation at render time.

The rendering approach changes parameters dynamically based on viewing angle and material properties. The Fresnel coefficient varies with the angle of incidence, automatically adjusting the weight between reflected and refracted light components. This parameter-based approach allows accurate light interaction simulation without requiring complex geometric calculations for each light ray.

The patent performs preliminary actions by pre-processing environmental information into ambient maps and background images before rendering. The contrast control map and internal rendering effect map are prepared in advance, storing lighting characteristics that can be quickly sampled during rendering. This preliminary preparation enables fast rendering while maintaining adaptability to different AR environments, as the pre-computed maps capture environmental lighting properties.

The rendering method uses universal maps (ambient map, background image) that can represent multiple different environments. The same rendering pipeline works across diverse AR scenarios by simply changing the sampled map data, making the system both fast and highly adaptable without requiring environment-specific rendering code.