Single-image human relighting aims to relight a target human under new lighting conditions by decomposing the input image into albedo, shape and lighting. Although plausible relighting results can be achieved, previous methods suffer from both the entanglement between albedo and lighting and the lack ...
First, we recover the 3D geometry by optimizing a 775 Figure 1: We introduce SIRA, a method which, from a single image, reconstructs human head avatars with high fidelity geometry and factorized lights and surface materials. This information can then be used for relighting purposes. signed ...
Generating a 3D human model from a single reference image is challenging because it requires inferring textures and geometries in invisible views while maintaining consistency with the reference image. Previous methods utilizing 3D generative models are limited by the availability of 3D training data. Op...
Re-lighting: Since light attribute is also predict in our model, it is easy to perform image relighting by setting a different Spherical Harmonic coefficient Lr, as Xn = R([Ca, Lr, Sa, Ta]). The visual results are shown in Fig. 8. Our SMR do not need additional ne...
The albedo estimated by SPAR3D has less baked-in lighting compared with SF3D and results in better relighting outcomes. Input Image Mesh Output Input Image Mesh Output Input Image Mesh Output Figure 10. Additional In-the-wild Results. We show additional results of SPAR3D on...