Yash Belhe

ybelhe@ucsd.edu

I'm a CSE PhD student at UCSD with Tzu-Mao Li and Ravi Ramamoorthi, working on all sorts of things in (differentiable) rendering, generative AI and 3D vision. I've been fortunate to work with several wonderful folks at Adobe Research, Iliyan Georgiev, Michaël Gharbi and Matt Fisher and NVIDIA Research, Thomas Müller, Merlin Nimier-David and Alex Keller.

Previously, I did my MS in ECE at Carnegie Mellon University working on computational imaging with Aswin Sankaranarayanan, and my BTech in ECE at IIT Guwahati.

CV (a differently formatted version of this page)

Papers

Highlighted papers. * = equal contribution.

• Repurposing Marigold for Zero-Shot Metric Depth Estimation via Defocus Blur Cues
  Chinmay Talegaonkar, Nikhil Gandudi Suresh, Zachary Novack, Yash Belhe, Priyanka Nagasamudra, Nicholas Antipa
  arXiv preprint, 2025.
  Turning relative depth estimators into absolute ones by using defocus blur cues.
• A Differentiable Wave Optics Model for End-to-End Computational Imaging System Optimization
  Chi-Jui Ho, Yash Belhe, Steve Rotenberg, Ravi Ramamoorthi, Tzu-Mao Li, Nicholas Antipa
  ICCV 2025.
  Efficient wave-optics-based optimization of computational imaging systems.
• Volumetrically Consistent 3D Gaussian Rasterization
  Chinmay Talegaonkar, Yash Belhe, Ravi Ramamoorthi, Nicholas Antipa
  CVPR 2025 (Highlight).
  (Mostly) mathematically correct 3D Gaussian rasterization using analytic integration.
• Spatiotemporal Bilateral Gradient Filtering for Inverse Rendering
  Wesley Chang*, Xuanda Yang*, Yash Belhe*, Ravi Ramamoorthi, Tzu-Mao Li
  SIGGRAPH Asia 2024 (conference).
  Discontinuity-preserving spatio-temporal optimizer to accelerate inverse rendering of volumes, textures, meshes.
• BSDF importance sampling using a diffusion model
  Ziyang Fu, Yash Belhe, Haolin Lu, Liwen Wu, Bing Xu, Tzu-Mao Li
  SIGGRAPH Asia 2024 (conference).
  A low-dimensional diffusion model to importance sample reflective and transmissive materials well in real-time.
• A Construct-Optimize Approach to Sparse View Synthesis without Camera Pose
  Kaiwen Jiang, Yang Fu, Mukund Varma T, Yash Belhe, Xiaolong Wang, Hao Su, Ravi Ramamoorthi
  SIGGRAPH 2024 (conference).
  Sparse view synthesis without camera poses progressively adding camera views and refining the solution.
• Importance Sampling BRDF Derivatives
  Yash Belhe, Bing Xu, Sai Praveen Bangaru, Ravi Ramamoorthi, Tzu-Mao Li
  Transactions on Graphics 2024.
  Variance reduction for BRDF derivatives by removing variance due to sign changes.
• Discontinuity-Aware 2D Neural Fields
  Yash Belhe, Michaël Gharbi, Matthew Fisher, Iliyan Georgiev, Ravi Ramamoorthi, Tzu-Mao Li
  SIGGRAPH Asia 2023 (journal).
  A new hybrid-neural-mesh image format for infinite resolution images in which edges are always sharp.
• Computational Imaging using Ultrasonically-Sculpted Virtual Lenses
  Hossein Baktash, Yash Belhe, Matteo G. Scopelliti, Yi Hua, Aswin C. Sankaranarayanan, Maysamreza Chamanzar
  International Conference on Computational Photography (ICCP), 2022.
  Reconstruct sharp images from blurry measurements from a virtual GRIN lens sculpted using ultrasound waves.