LiteVoxel: Low-memory Intelligent Thresholding for Efficient Voxel Rasterization

TL;DR

LiteVoxel is a novel training pipeline for sparse-voxel rasterization that significantly reduces VRAM usage while maintaining high-quality scene reconstruction, especially in low-frequency regions.

Contribution

It introduces a self-tuning, low-memory rasterization method with adaptive pruning and loss reweighting, improving stability and efficiency over prior approaches.

Findings

Reduces peak VRAM by 40-60%.

Maintains PSNR and SSIM comparable to existing methods.

Mitigates errors in low-frequency regions and boundary instability.

Abstract

Sparse-voxel rasterization is a fast, differentiable alternative for optimization-based scene reconstruction, but it tends to underfit low-frequency content, depends on brittle pruning heuristics, and can overgrow in ways that inflate VRAM. We introduce LiteVoxel, a self-tuning training pipeline that makes SV rasterization both steadier and lighter. Our loss is made low-frequency aware via an inverse-Sobel reweighting with a mid-training gamma-ramp, shifting gradient budget to flat regions only after geometry stabilize. Adaptation replaces fixed thresholds with a depth-quantile pruning logic on maximum blending weight, stabilized by EMA-hysteresis guards and refines structure through ray-footprint-based, priority-driven subdivision under an explicit growth budget. Ablations and full-system results across Mip-NeRF 360 (6scenes) and Tanks & Temples (3scenes) datasets show mitigation of errors in low-frequency regions and boundary instability while keeping PSNR/SSIM, training time, and FPS comparable to a strong SVRaster pipeline. Crucially, LiteVoxel reduces peak VRAM by ~40%-60% and preserves low-frequency detail that prior setups miss, enabling more predictable, memory-efficient training without sacrificing perceptual quality.