Learning Global Motion with Compact Gaussians for Feed-Forward 4D Reconstruction
:::info Stub — Full Engineering Breakdown Coming This paper was auto-fetched from arXiv on 2026-06-01. A full breakdown with production viability rating, implementation notes, and honest limitations is being written. Subscribe to AI Letters → :::
| Authors | Mungyeom Kim et al. |
| Year | 2026 |
| Field | Computer Vision |
| arXiv | 2605.31595 |
| Download | |
| Categories | cs.CV |
Abstract
Dynamic scene reconstruction from monocular video remains a fundamental challenge in computer vision. Existing feed-forward methods predict 3D Gaussians pixel-wise for each frame, suffering from duplicated Gaussians and view-dependent biases that hinder effective learning of scene motion. We present C4G, a feed-forward 4D reconstruction framework built upon a compact set of timestamp-conditioned learnable Gaussian query tokens. Each token aggregates corresponding features across the full temporal context and decodes a 3D Gaussian whose position is modulated by the target timestamp, enabling globally coherent motion modeling without per-scene optimization. To capture fine-grained details, we further introduce a video diffusion model-based rendering enhancement module. Since our framework effectively aggregates features into Gaussians, we extend this capability to feature lifting, producing a 4D feature field that supports point tracking and dynamic scene understanding. C4G achieves strong novel-view synthesis performance using significantly fewer Gaussians and without requiring camera poses, while exhibiting stronger motion modeling and robustness to large temporal gaps.
Engineering Breakdown
The Problem
Dynamic scene reconstruction from monocular video remains a fundamental challenge in computer vision.
The Approach
We present C4G, a feed-forward 4D reconstruction framework built upon a compact set of timestamp-conditioned learnable Gaussian query tokens. Since our framework effectively aggregates features into Gaussians, we extend this capability to feature lifting, producing a 4D feature field that supports point tracking and dynamic scene understanding.
Key Results
C4G achieves strong novel-view synthesis performance using significantly fewer Gaussians and without requiring camera poses, while exhibiting stronger motion modeling and robustness to large temporal gaps.
Research Areas
This paper contributes to the following areas of AI/ML engineering:
- Image recognition
- Object detection
- Visual transformers
- Convolutional networks
- Multimodal learning
- Gaussians
:::tip Subscribe Get weekly breakdowns of papers like this in AI Letters - the newsletter for engineers building production AI systems. :::