RayPE: Ray-Space Positional Encoding for 3D-Aware Video Generation

• URL: http://arxiv.org/abs/2606.27345v2
• Date: Fri, 26 Jun 2026 06:30:48 GMT
• Title: RayPE: Ray-Space Positional Encoding for 3D-Aware Video Generation
• Abstract summary: RayPE is a positional-encoding extension that injects per-token 6D Plucker coordinates additively into the queries and keys of self-attention.<n>The resulting attention score decomposes into a content term, a geometry term, and two content and geometry cross-terms.
• Score: 43.253222907352146
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Modern video diffusion transformers position their tokens through RoPE on the (u,v,t) axes -- a description of the camera's sampling grid that says nothing about the 3D structure of the scene. We observe that the geometric relation between two camera rays is captured by the Plucker reciprocal product, which is bilinear in the two rays -- the same algebraic form as the dot product in Transformer attention. Building on this analogy, we propose RayPE, a positional-encoding extension that injects per-token 6D Plucker coordinates additively into the queries and keys of self-attention, with a query/key flip arrangement under which the symmetric identity configuration coincides exactly with the reciprocal product. The injection is additive, the resulting attention score decomposes into a content term, a geometry term, and two content and geometry cross-terms -- all of which our experiments find individually necessary. To make the encoding stable across video data with heterogeneous camera-translation scales (SfM, deep SLAM, metric), we further decouple ray direction from moment magnitude, gate the encoding by a learned function of the log-magnitude, and apply RMSNorm to align it with the QKNorm-normalized content branch. The full module adds less than 0.1% parameters to a pretrained video DiT, is zero-initialized to start from the pretrained weights, and improves camera controllability, cross-frame 3D consistency, and overall video quality on a four-dataset training mixture.

Related papers

• PE-Field 4D: Video Generation Models as Canvas [44.0217294710719]
  We show that positional encoding in video diffusion transformers provides a useful spatial bias for geometry-aware control.<n>We introduce a geometry-aware cross-attention mechanism that enables target video latent tokens to attend to structured context tokens.<n>The resulting framework provides a simple and effective approach for controllable video generation.
  arXiv  Detail & Related papers  (2026-07-17T06:22:14Z)
• CRePE: Curved Ray Expectation Positional Encoding for Unified-Camera-Controlled Video Generation [48.95338161181985]
  We propose Ray Expectation Positional Controlled (CRePE) for camera-conditioned video generation.<n>CRePE represents each image token as a depth-aware positional distribution along its source ray, capturing the projected-path geometry induced by wide-angle and fisheye cameras.<n>Design leads to more stable camera control and improves several geometry-aware and perceptual-quality metrics.
  arXiv  Detail & Related papers  (2026-05-13T03:18:26Z)
• URoPE: Universal Relative Position Embedding across Geometric Spaces [55.651792747815854]
  URoPE is a universal extension of Rotary Position Embedding (RoPE) to cross-view or cross-dimensional geometric spaces.<n>For each key/value image patch, URoPE samples 3D points along the corresponding camera ray at predefined depth anchors.<n>Standard 2D RoPE can then be applied using the projected pixel coordinates.
  arXiv  Detail & Related papers  (2026-04-20T18:52:03Z)
• SceneScribe-1M: A Large-Scale Video Dataset with Comprehensive Geometric and Semantic Annotations [88.8747004592363]
  SceneScribe-1M is a new large-scale, multi-temporal video dataset.<n>It comprises one million in-the-wild videos, each meticulously annotated with detailed textual descriptions, precise parameters, dense depth maps, and consistent 3D point tracks.<n>We demonstrate the versatility and value of SceneScribe-1M by establishing benchmarks across a wide array of downstream tasks, including monocular depth estimation, scene reconstruction dynamic point tracking, as well as generative tasks such as text-to-video synthesis, with or without camera control.
  arXiv  Detail & Related papers  (2026-04-09T08:59:33Z)
• RayRoPE: Projective Ray Positional Encoding for Multi-view Attention [46.94549341989407]
  We study positional encodings for multi-view transformers that process tokens from a set of posed input images.<n>We seek a mechanism that encodes patches uniquely, allows SE(3)-invariant attention with multi-frequency similarity, and can be adaptive to the geometry of the underlying scene.<n>RayRoPE represents patch positions based on associated rays but leverages a predicted point along the ray instead of the direction for a geometry-aware encoding.
  arXiv  Detail & Related papers  (2026-01-21T18:55:51Z)
• 3DPCNet: Pose Canonicalization for Robust Viewpoint-Invariant 3D Kinematic Analysis from Monocular RGB cameras [7.906702226082628]
  3DPCNet is a compact, estimator-agnostic module that operates directly on 3D joint coordinates.<n>Our method produces acceleration signals from video that show strong visual correspondence to ground-truth IMU sensor data.
  arXiv  Detail & Related papers  (2025-09-27T18:55:21Z)
• Cameras as Relative Positional Encoding [37.675563572777136]
  Multi-view transformers must use camera geometry to ground visual tokens in 3D space.<n>We show how relative camera conditioning improves performance in feedforward novel view synthesis.<n>We then verify that these benefits persist for different tasks, stereo depth estimation and discriminative cognition, as well as larger model sizes.
  arXiv  Detail & Related papers  (2025-07-14T17:22:45Z)
• DiffusionSfM: Predicting Structure and Motion via Ray Origin and Endpoint Diffusion [53.70278210626701]
  We propose a data-driven multi-view reasoning approach that directly infers 3D scene geometry and camera poses from multi-view images.<n>Our framework, DiffusionSfM, parameterizes scene geometry and cameras as pixel-wise ray origins and endpoints in a global frame.<n>We empirically validate DiffusionSfM on both synthetic and real datasets, demonstrating that it outperforms classical and learning-based approaches.
  arXiv  Detail & Related papers  (2025-05-08T17:59:47Z)
• Coca-Splat: Collaborative Optimization for Camera Parameters and 3D Gaussians [26.3996055215988]
  Coca-Splat is a novel approach to address the challenges of sparse view pose-free scene reconstruction and novel view synthesis (NVS)<n>Inspired by deformable DEtection TRansformer, we design separate queries for 3D Gaussians and camera parameters.<n>We update them layer by layer through deformable Transformer layers, enabling joint optimization in a single network.
  arXiv  Detail & Related papers  (2025-04-01T10:48:46Z)
• AutoDecoding Latent 3D Diffusion Models [95.7279510847827]
  We present a novel approach to the generation of static and articulated 3D assets that has a 3D autodecoder at its core. The 3D autodecoder framework embeds properties learned from the target dataset in the latent space. We then identify the appropriate intermediate volumetric latent space, and introduce robust normalization and de-normalization operations.
  arXiv  Detail & Related papers  (2023-07-07T17:59:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.