Search papers, labs, and topics across Lattice.
BAIR脳1 convolutional neural network to align with the feature dimension dd from image featuresApr 1, 2026arXiv:2604.00509The paper introduces RT-GS, a novel Gaussian Splatting framework that incorporates microfacet material models and differentiable ray tracing to explicitly model specular reflections and transmittance. RT-GS uses separate Gaussian primitives to represent reflections and transmittance, enabling the reconstruction of distant reflections and objects behind transparent surfaces. Experiments demonstrate that RT-GS significantly improves the qualitative realism of novel view synthesis in scenes with complex specular light interactions compared to existing methods.
See-through realism gets a boost: RT-GS lets Gaussian Splatting finally handle reflections and transparency like a boss.
Gaussian Splatting is a powerful tool for reconstructing diffuse scenes, but it struggles to simultaneously model specular reflections and the appearance of objects behind semi-transparent surfaces. These specular reflections and transmittance are essential for realistic novel view synthesis, and existing methods do not properly incorporate the underlying physical processes to simulate them. To address this issue, we propose RT-GS, a unified framework that integrates a microfacet material model and ray tracing to jointly model specular reflection and transmittance in Gaussian Splatting. We accomplish this by using separate Gaussian primitives for reflections and transmittance, which allow modeling distant reflections and reconstructing objects behind transparent surfaces concurrently. We utilize a differentiable ray tracing framework to obtain the specular reflection and transmittance appearance. Our experiments demonstrate that our method successfully produces reflections and recovers objects behind transparent surfaces in complex environments, achieving significant qualitative improvements over prior methods where these specular light interactions are prominent.
