XPR: An Extensible Cross-Platform Point-Based Differentiable Renderer

2026-06-09

AI Summary

Existing point-based differentiable rendering methods rely on low-level implementations, hardware-specific kernels, and manual backpropagation, hindering rapid prototyping, cross-platform deployment, and reproducibility. This work proposes XPR, a framework that decouples algorithmic logic from the underlying rendering pipeline through a high-level domain-specific language (DSL). XPR employs modular, statically shaped parallel operations and leverages the XLA compiler to enable efficient execution across diverse backends, including GPUs, TPUs, and CPUs. To our knowledge, XPR is the first framework to achieve highly scalable and cross-platform compatible point-based differentiable rendering. It allows concise and faithful reimplementation of representative methods—such as 3DGS, 3DGUT, and LinPrim—in just a few hundred lines of Python code, delivering portable and high-performance execution across multiple hardware platforms.

Abstract

Point-based differentiable rendering underpins modern 3D reconstruction, novel-view synthesis, and learning-based graphics pipelines, but developing new rendering methods often requires extensive low-level implementation, hardware-specific kernels, and manually written backward passes. This limits rapid prototyping, reproducibility, exploration, and deployment, especially across diverse hardware platforms. This paper presents XPR, an extensible cross-platform framework for point-based differentiable rendering. XPR introduces a high-level programming interface that separates method-specific logic from the shared rendering pipeline, allowing users to implement new methods in a few lines of code. Its pipeline decomposes rendering into modular, statically shaped parallel operations that can be lowered by a cross-platform compiler to GPUs, TPUs, CPUs, and other ML accelerators. We demonstrate implementations of 3DGS, 3DGUT, and LinPrim, with only a few 100s lines of Python code, each of which can be compiled to a range of hardware platforms with the XLA compiler. These results show that XPR enables fast experimentation and portable execution for emerging point-based differentiable rendering systems.

Problem

• point-based differentiable rendering
• cross-platform
• rapid prototyping
• hardware portability
• reproducibility

Innovation

• differentiable rendering
• point-based rendering
• cross-platform compilation
• XLA
• modular pipeline