Real-time Rendering of Pre-integrated Neural Emitters

2026-10-05

AI Summary

This work addresses the performance bottlenecks and noise artifacts caused by runtime integration in the real-time rendering of complex emissive objects. To this end, we propose Neural Emission Fields, a method that integrates neural radiance fields with pre-integration techniques to parameterize illumination via position, normal, and viewing direction, thereby reformulating lighting computation as an offline precomputation process. Employing a dual-head architecture alongside a local-frame training strategy, the resulting model serves as a portable lighting asset transferable across scenes while handling effects such as self-occlusion at zero additional cost. A single network forward pass suffices to produce noise-free direct illumination, substantially improving both the efficiency and visual quality of real-time rendering for geometrically complex light sources.

Abstract

Direct illumination from non-trivial emitters with occluding housing, high-polygon emissive meshes, spatially varying emission, or deformable assemblies is a persistent bottleneck in real-time rendering. Current real-time solutions either rely on simplified analytical representations or fall back to runtime sampling. Analytical methods are restricted to simple emitter geometry, pure sampling-based estimators need high sampling budgets to be noise-free, and proxy representations still require runtime integration over outgoing radiance around the emitter. We present Neural Emission Fields (NEF), a representation that eliminates runtime integration by precomputing the illumination in the volume around the emitter. This neural field is parameterized by position, normal, view direction, and material parameters. Using a two-headed diffuse/glossy architecture, it yields noise-free unoccluded direct illumination from a single network evaluation per shading point. Because training is performed in the emitter's local frame, a trained NEF acts as a portable lighting asset reusable across scenes under rigid transforms. Internal interreflections, self-occlusion, spatially varying emission, and deformations are absorbed into the learned representation at zero additional runtime cost.

Problem

• real-time rendering
• direct illumination
• complex emitters
• runtime integration
• neural emission fields

Innovation

• Neural Emission Fields
• Real-time Rendering
• Pre-integration
• Direct Illumination
• Portable Lighting Asset