“Parameter-space ReSTIR for Differentiable and Inverse Rendering” by Chang, Sivaram, Nowrouzezahrai, Hachisuka, Ramamoorthi, et al. …

  • ©Wesley Chang, Venkataram Sivaram, Derek Nowrouzezahrai, Toshiya Hachisuka, Ravi Ramamoorthi, and Tzu-Mao Li

Conference:


Type:


Title:

    Parameter-space ReSTIR for Differentiable and Inverse Rendering

Session/Category Title: Inverse Rendering: Does Anybody Know How I Got Here?


Presenter(s)/Author(s):


Moderator(s):



Abstract:


    Inverse rendering algorithms that utilize differentiable rendering typically use gradient descent to optimize for 3D scene parameters. We observe that the scene changes slowly during optimization, and therefore adapt temporal reuse from the ReSTIR algorithm in real-time rendering to reuse Monte Carlo samples across gradient iterations and accelerate inverse rendering.


Overview Page: