发表机构
Shanghai Jiao Tong University; Goertek Inc.; Fudan University（上海交通大学; 歌尔股份有限公司; 复旦大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对稀疏视角3DGS重建质量差的问题，提出GAPS框架，利用扩散模型生成伪视角并主动选择，结合密度自适应正则化，在LLFF和Mip-NeRF 360上显著提升PSNR、SSIM并降低LPIPS。
AI 中文摘要
从稀疏观测进行新视角合成严重欠约束。尽管三维高斯泼溅（3DGS）能够实现实时渲染，但在仅有少量视角训练时，会产生漂浮物、破碎几何形状和褪色背景。我们提出一种交替优化框架，利用预训练图像扩散模型生成几何一致的伪视角，以提供额外的3DGS监督。生成过程受深度条件ControlNet、IP-Adapter风格迁移、LoRA场景适应和img2img结构锚定约束。我们引入生成式主动伪视角选择（GAPS），在选择目标视角时平衡重建信息量和生成可靠性。其退火调度从训练早期的保守插值过渡到后期的探索性外推，逐步覆盖未观测区域。双准则准入门和不确定性加权损失拒绝不可靠的生成，而密度自适应DropGaussian减少复杂场景中的过拟合。在LLFF数据集上，使用3/6/9个视角时，我们的方法相比原始3DGS将平均PSNR提高了0.40/0.89/0.70 dB。在Mip-NeRF 360数据集上，使用12/24个视角时，增益为1.18/0.80 dB。在所有设置中，SSIM均提高，LPIPS均降低。消融实验表明，主动选择和密度自适应正则化都是必要的；只有完整方法才能在无界360度场景中将LPIPS降低到无伪视角基线以下。
英文摘要
Novel view synthesis from sparse observations is severely under-constrained. Although 3D Gaussian Splatting (3DGS) enables real-time rendering, it produces floaters, broken geometry, and washed-out backgrounds when trained with few views. We propose an alternating optimization framework that uses a pre-trained image diffusion model to generate geometrically consistent pseudo-views for additional 3DGS supervision. Generation is constrained by depth-conditioned ControlNet, IP-Adapter style transfer, LoRA scene adaptation, and img2img structural anchoring. We introduce Generative Active Pseudo-view Selection (GAPS) to balance reconstruction informativeness and generative reliability when choosing target views. Its annealing schedule shifts from conservative interpolation early in training to exploratory extrapolation later, gradually covering unobserved regions. A dual-criterion admission gate and uncertainty-weighted losses reject unreliable generations, while density-adaptive DropGaussian reduces overfitting in complex scenes. On LLFF with 3/6/9 views, our method improves average PSNR over vanilla 3DGS by 0.40/0.89/0.70 dB. On Mip-NeRF 360 with 12/24 views, the gains are 1.18/0.80 dB. SSIM improves and LPIPS decreases in every setting. Ablations show that active selection and density-adaptive regularization are both necessary; only the full method reduces LPIPS below the no-pseudo-view baseline on unbounded 360-degree scenes.
