DriftNeRF: Cross-Ray Density Drifting for Sparse-View Neural Radiance Fields

Key points

• We attribute a specific, measurable defect to the objective itself: the rendering loss is a sum over independent rays and contains no term comparing densities that different rays assign to the same region of space, so nearby 3D points can settle on contradictory densities while every training pixel is reproduced correctly.1
• We make this quantity explicit by measuring cross-ray density disagreement directly during training, and show that it stays high in vanilla NeRF, falls only incidentally under frequency-based regularization, and moves together with held-out quality in the sparse regime.1
• We then close the gap with DriftNeRF, a drop-in regularizer that instantiates the drifting-field template on density samples: a masked exponential kernel aggregates densities from nearby points on other rays into a stop-gradient consensus target, and the resulting drifting field penalizes unsupported disagreement while leaving agreed-upon boundaries intact.1
• On NeRF-Synthetic it reduces residual disagreement by an order of magnitude, and across NeRF-Synthetic, DTU, and LLFF it improves 3-view PSNR over the strongest MLP-NeRF baseline by 0.42-0.65 dB and remains competitive with sparse-view Gaussian pipelines under a matched protocol without their depth, flow, or covisibility preprocessing.1

Sources (1)

• [1]DriftNeRF: Cross-Ray Density Drifting for Sparse-View Neural Radiance Fields
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 02:49 AM

  Neural Radiance Fields (NeRF) synthesize photorealistic novel views from dense image collections, but degrade sharply when few views are available.

  We attribute a specific, measurable defect to the objective itself: the rendering loss is a sum over independent rays and contains no term comparing densities that different rays assign to the same region of space, so nearby 3D points can settle on contradictory densities while every training pixel is reproduced correctly.

Extractive summary: sentences quoted from the sources.