Kernel Nyström Method for Light Transport

Kernel Nyström Method for Light Transport



in this paper we propose the Colonel NIST on method for reconstructing the light transport matrix from a small number of acquired images the NIST on method exploits both linear and nonlinear data coherence in the light transport matrix for reconstructing the matrix the reconstructed light transport matrix can be used to render the scene under new lighting conditions here we show relighting results of scenes with a variety of light transport effects including specular reflection caustics light occlusion and subsurface scattering using a projector camera system we acquire columns of the light transport matrix by projecting a light beam to a diffuser that creates a point light source above the scene then with a laser system we acquire rows of the light transport matrix by shooting the laser beam at a surface point in the scene the rows of the light transport matrix obtained from the image on the diffuser after acquiring a small number of columns and rows we apply the kernel lister method to reconstruct the whole light transport matrix from the acquired columns and rows you we have evaluated the kernel mist on method using test scenes with a different light transport characteristics here is a scene with multiple objects and various types of surface properties the light transport matrix is reconstructed from hundred and fifty columns and rows all lighting effects are well preserved in the reconstructed live transport including specularity shadowing and caustics compared with the ground truth the anisotropic reflection and subtle color bleeding effect of the wax cube are accurately reproduced here is a scene with multiple glass objects which lead to complex shadows and caustics the light transport matrix was reconstructed from 250 columns and rows the dynamic caustics and shadows produced by the reconstructed light transport match the ground truth very well here is a scene with a complex shadow which scans the floor as light direction changes compared with the ground truth the complex shape of the shadow and its dynamics are well reproduced with the reconstructed light transport here is a scene with a concave object which leads to inter reflections between different parts of the object surface the color bleeding effect due to inter reflections is well captured with the reconstructed light transport notice that the subtle color bleeding from the tennis ball matches the ground truth closely here we show translucent materials captured using our method the high-resolution diffuse BSS RDFS are obtained after acquiring 50 images which take only a few minutes notice that the anisotropic subsurface scattering is well-preserved and the detailed marble veins are reproduced here we show relighting results of the captured translucent materials finally we show relighting results of the test scenes under environment lighting in conclusion we present the kernel list our method for modeling the light transport matrix from sparse sampled images of the scene we show it is efficient for modeling different kinds of light transport including specular reflections shadows caustics into reflections and subsurface scattering

Leave a Reply

Your email address will not be published. Required fields are marked *