无需训练即可实现3D重建模型的近最优量化，压缩比高达7.9倍。
3DTurboQuant: Training-Free Near-Optimal Quantization for 3D Reconstruction Models
- 利用随机旋转使参数服从已知贝塔分布，实现无需数据依赖的预计算量化。
- 在NeRF Synthetic上将3DGS压缩3.5倍，PSNR损失仅0.02dB，DUSt3R KV缓存压缩7.9倍。
- 支持无训练、无校准数据，秒级压缩，适合部署与资源受限场景。
现有压缩3D高斯泼溅（3DGS）、NeRF或基于Transformer的3D重建模型的方法均需通过逐场景微调学习数据相关的码本。本文证明此步骤非必要。这些模型中占据存储主导的参数——3DGS中的45维球谐函数和DUSt3R中的1024维键值向量——处于特定维度范围，单次随机旋转即可将任意输入映射为具有已知贝塔分布坐标的向量。这使得预先计算、数据无关的Lloyd-Max量化接近最优，距离信息论下界仅差2.7倍。我们提出3DTurboQuant，包含：(1) 基于维度的判据，可预测哪些参数可量化及量化位宽；(2) 将量化均方误差与每场景渲染峰值信噪比关联的范数分离边界；(3) 扩展旋转量化至二维哈希网格特征的条目分组策略；(4) 可组合的剪枝-量化流水线，具闭式压缩比。在NeRF Synthetic数据集上，3DTurboQuant将3DGS压缩3.5倍，PSNR损失0.02dB；将DUSt3R KV缓存压缩7.9倍，点云保真度达39.7dB。全程无需训练、无码本学习、无校准数据，压缩耗时仅数秒。代码将公开于https://github.com/JaeLee18/3DTurboQuant。
原文摘要 · Abstract (English)
Every existing method for compressing 3D Gaussian Splatting, NeRF, or transformer-based 3D reconstructors requires learning a data-dependent codebook through per-scene fine-tuning. We show this is unnecessary. The parameter vectors that dominate storage in these models, 45-dimensional spherical harmonics in 3DGS and 1024-dimensional key-value vectors in DUSt3R, fall in a dimension range where a single random rotation transforms any input into coordinates with a known Beta distribution. This makes precomputed, data-independent Lloyd-Max quantization near-optimal, within a factor of 2.7 of the information-theoretic lower bound. We develop 3D, deriving (1) a dimension-dependent criterion that predicts which parameters can be quantized and at what bit-width before running any experiment, (2) norm-separation bounds connecting quantization MSE to rendering PSNR per scene, (3) an entry-grouping strategy extending rotation-based quantization to 2-dimensional hash grid features, and (4) a composable pruning-quantization pipeline with a closed-form compression ratio. On NeRF Synthetic, 3DTurboQuant compresses 3DGS by 3.5x with 0.02dB PSNR loss and DUSt3R KV caches by 7.9x with 39.7dB pointmap fidelity. No training, no codebook learning, no calibration data. Compression takes seconds. The code will be released (https://github.com/JaeLee18/3DTurboQuant)
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。