AutoLensAI Uses Physically Based Rendering to Fix Lighting After Car Image Segmentation

Cutting a vehicle cleanly from a photo does not solve the full compositing challenge, as the car still retains the lighting conditions of its original environment. AutoLensAI's automotive image pipeline addresses this by adding a physically based relighting stage after standard segmentation. The system estimates scene illumination and surface normals, then applies a low-frequency shading gain over original pixels to match the new background's lighting without altering fine details like badges or trim. Key technical components include a per-pixel roughness map, a corrected Trowbridge-Reitz GGX specular term, and a glass reflectance estimator. A built-in detail-energy check allows the pipeline to reject any frame where protected high-frequency detail is unintentionally altered.