Photon Differential Splatting for Rendering Caustics

N-SfC: Robust and Fast Shape Estimation from Caustic Images

“…The simulation (cf. [9]) and subsequent gradient calculation through backpropagation mainly depends on three hyperparameters: The number of samples n l , i.e. the number of paths outgoing from the light, the number of distinct wavelengths associated with each ray nw and a smoothing parameter, which controls the size of an elliptical footprint over which the energy of each transported photon is distributed.…”

“…Thus we focus our efforts on a learned gradient descent system to circumvent those spurious local minima an achieve a plausible reconstruction for the 3D printing process at hand. We follow the simulation approach from SfC [14] for image formation and gradient generation, as they in turn build upon [9], which allows for sharper caustic edges with fewer samples.…”

“…The parameters regarding sample count and resolution were chosen as a compromise between quality and memory consumption. For more details about the influence of each rendering parameter we refer to [9].…”

Efficient rendering of caustics with streamed photon mapping

“…Such an approach enables precise estimation indirect illumination close to the edges of objects and on tively small isolated objects. risvad et al [14] proposed a photon splatting technique ich reduces noise and blur in the rendering of caustics. They d photon differentials to determine the size and shape of splats such that we achieve adaptive anisotropic flux denestimation in photon splatting.…”

“…Frisvad et al [14] proposed a photon splatting technique which reduces noise and blur in the rendering of caustics. They used photon differentials to determine the size and shape of the splats such that we achieve adaptive anisotropic flux density estimation in photon splatting.…”