Improved Sampling for Gradient-Domain Metropolis Light Transport

generalized framework for gradient-domain Metropolis rendering includes three techniques to avoid sampling artifacts and reduce variance in sampled gradients. We (OURS) achieve visually and numerically improved results compared to previous work (GDMLT). We present a generalized framework for gradient-domain Metropo-lis rendering, and introduce three techniques to reduce sampling artifacts and variance. The first one is a heuristic weighting strat-egy that combines several sampling techniques to avoid outliers. The second one is an improved mapping to generate offset paths re-quired for computing gradients. Here we leverage the properties of manifold walks in path space to cancel out singularities. Finally, the third technique introduces generalized screen space gradient ker-nels. This approach aligns the gradient kernels with image struc-tures such as texture edges and geometric discontinuities to obtain sparser gradients than with the conventional gradient kernel. We implement our framework on top of an existing Metropolis sam-pler, and we demonstrate significant improvements in visual and numerical quality of our results compared to previous work