Inverse Problems News
The latest Inverse Problems coverage curated for game developers.
arXiv cs.GR details UltraDiff, a differentiable ray-tracing framework that brings gradient-based optimization to ultrasound shape reconstruction. The method treats echoes as time-of-flight events, then backpropagates through a Monte Carlo forward …
arXiv cs.GR details Dirichlet Splatting, a differentiable renderer built for wave-based sensing instead of visible-light scenes. The method swaps Gaussian splats for the exact Dirichlet kernel used by finite-window DFTs, …
A new ptychography pipeline uses multi-scale non-linear spectral preconditioning to push tabletop nanoscopy past the usual detector dynamic-range bottleneck. The key shift is that strict radiometric linearity is no longer …
Researchers trained a goal-conditioned PPO policy to estimate food material parameters from fracture behavior in a single forward pass, then used CMA-ES to refine the result. On orange peeling, the …
A new inference method for diffusion models, Local MAP Sampling, aims at the point-estimation side of inverse problems rather than full posterior sampling. The authors report state-of-the-art results across image …
A new differentiable simulation framework targets one of the hardest problems in graphics and robotics: stable gradients through frictional contact. Built as a fully GPU-accelerated solver, it aims to keep …